Re: 'Twin Triode' AM Transmiter
John Byrns
flipper <fli...@fish.net> wrote:
> First, apologies to the one who sent me the 12AU7 'twin triode'
> transmitter text file. I copied the text but not the headers and now I
> don't remember who it was.
It was probably moi if you are referring to the "ASCII schematic" I posted to
this newsgroup on Fri, 04 Mar 2011 in your thread titled "Local Osc question".
By the way your fli...@fish.net email address doesn't work!
> I've bread boarded it but had to use a much larger cathode resistor or
> else it severely distorted past, oh say, 50% mod. Ironically, less
> power meant 'more audio drive' because of the larger cathode voltage.
> That, and wanting enough sensitivity for cell phone/small MP3 players,
> meant a preamp.
I wonder if the original cathode resistor value was actually chosen for the
12AU7 as the schematic says? Other dual triodes are mentioned on the schematic
and the value shown might be more appropriate for a 12AT7 or something like that.
From your description of the problem it sounds like the voltage on the modulator
plate was too low, limiting the amount of positive modulation available? Did
you have any problem with the oscillator cutting off on serious negative
modulation? It occurred to me that shunting the modulator tube with a resistor
of appropriate value would eliminate any possibility of the oscillator dropping
out.
Another twin triode approach would be to connect the oscillator and modulator
triodes in parallel, with the B+ feed taken through an audio choke, a.k.a.
Hessing modulation. The Hessing configuration changes around some of the
modulation tradeoffs, reducing the possibility of oscillator dropout, and
providing greater carrier power at the same B+ voltage. The down side is that
it requires an expensive audio inductor and doesn't have the elegant simplicity
of the series connection, although the circuit complexity is the same other than
the addition of the audio inductor, which also serves to reduce hum from the B+
supply.
> Frequency response is better with a smaller (1nF vs 10nF) RF bypass
> and range is better tacking the (obviously untuned) wet noodle antenna
> onto the other side of the coil.
I agree 10 nF is way too large, your 1.0 nF sounds about right, I probably would
have gone with 1.5 nF on general principles, your 1 nF is probably better.
I'm not sure which end of the coil is which, however range should be best when
the "wet noodle" is connected to the long end of the coil. I wouldn't say that
the "wet noodle" is untuned since it is connected directly across the plate tank
where it serves as part of the tuning capacitance, although the impedance
matching is probably far from ideal. I suppose it is a matter of personal
preference whether or not to call this a tuned antenna.
> http://flipperhome.dyndns.org/Twin%20Triode%20AM%20Transmitter.htm
>
> I haven't put it in the box yet because I keep trying to find a way to
> do one with 2 7-pin tubes, in particular, using a pentode under a
> triode but I can't find a suitable 7 pin triode.
How about using a 6C4, which is half of a 12AU7 IIRC. Why do you want to use a
pentode modulator? The triode modulator is probably part of what provides the
greater linearity, and of course provides a lower impedance to drive that 1 nF
capacitor, extending the high frequency audio response. Is the purpose of the
pentode to eliminate the need for a preamp? If only a small amount of
additional gain is needed how about using a triode with a higher
transconductance, like the 6AB4?
> Anyway, I went ahead last night and punched out one of the sockets to
> 9 pin so it's going in there sooner or later. Now I need to decide if
> I'll use an 'expensive' 12AU7, which I've got, or a 6(7)au7 I'd have
> to buy and decide if I'm going to finish that 'tuner cap hack' so the
> dial does something.
Oops, there goes the two 7-pin tube approach! For a really 'expensive' tube how
about making it a 'Triple Triode' AM Transmitter using one of the several triple
triode compactrons, using the third section as the preamp?
--
Regards,
John Byrns
Surf my web pages at, http://fmamradios.com/
Patrick Turner
> In article <dduc47t939q95vce1fbtchalrn93s7t...@4ax.com>,
>
> flipper <flip...@fish.net> wrote:
> > First, apologies to the one who sent me the 12AU7 'twin triode'
> > transmitter text file. I copied the text but not the headers and now I
> > don't remember who it was.
>
> It was probably moi if you are referring to the "ASCII schematic" I posted to
> this newsgroup on Fri, 04 Mar 2011 in your thread titled "Local Osc question".
>
Flipper's 2 tube AM transmitter is obviously nicely simple, but for me
I'd demand more from anything I built, like low envelope THD < 1% up
to say 95% modulation at up to AF = 5kHz, and such expectations
involves having many more tubes and no silly grumbling about the cost
of a 12AU7 or the use of a 7 pin or 9 pin tube, or the number of
tubes, period.
My latest thinking is that a fine AM generator might be made using 2
cascaded 6BE6 tubes where a buffered oscillator using a 12AU7 produces
a tunable RF signal anywhere between 300kHz and 1.7MHz in 2 ranges for
testing any IF strip or MW band in an old radio. The first 6BE6 has RF
applied to one grid and AF applied to other grid to give maybe 50%
modulation. The anode output signal is applied to the g1 of the
second 6BE6 with AF applied to the g3 to increase the amount of
modulation by another 50% if needed. Such a method is nowhere in any
book, but its worth a try, because today there is no shortage of so
many tubes which one may find time to experiment with. I have
hundredss if not thousands tubes collected over the years, most have
just fallen into my hands from people who didn't want them.
Maybe Flipper should get around a bit more, because there's mountains
of free tubes available.
And quite a lot more ideas he could try.
BTW, my idea requires a 2 gang cap, one for the oscillator, one for
the RF tranny on the second 6BE6 tube. Of course plate modulation
might work better, quite easy, use a 6BQ5 in triode to modulated the
triode RF amp, and the power of the modulator adds to the RF power.
But I have a single 6BX6 pentode for my AM gene, with grid modulation,
and the AF NFB pushes the % to over 95% below 8 kHz at low THD. The RF
tranny secondary is rated at less than 100 ohms, so anything connected
don't affect the 6BX6 operation. I can over drive any AM radio I am
testing.
Patrick Turner.
John Byrns
flipper <fli...@fish.net> wrote:
> On Sat, 13 Aug 2011 16:05:03 -0500, John Byrns <byr...@sbcglobal.net>
> wrote:
>
> >In article <dduc47t939q95vce1...@4ax.com>,
> > flipper <fli...@fish.net> wrote:
> >
> >> First, apologies to the one who sent me the 12AU7 'twin triode'
> >> transmitter text file. I copied the text but not the headers and now I
> >> don't remember who it was.
> >
> >It was probably moi if you are referring to the "ASCII schematic" I posted
> >to
> >this newsgroup on Fri, 04 Mar 2011 in your thread titled "Local Osc
> >question".
>
> Yes. I thought it was but couldn't remember for sure.
>
> >By the way your fli...@fish.net email address doesn't work!
>
> That's actually the purpose of it.
>
> Long time ago I needed an email address that 'looked real' to some
> program, by now I forget what or why, and it checked for 'valid'
> extensions. Well, .net is a 'valid' extension and fish is part of the
> joke.
I thought "flipper" wasn't a fish, at least you have protested to that effect at
least twice before?
I would probably have gone with something like r...@hair.net, or similar.
Speaking of such, why do you use the "X-No-Archive: yes" header line? I think
it worked with my ISP's old usenet server, which has since gone away, but it
sure isn't respected by GigaNews, which has archived 2923 of your posts to this
group since what I assume was your first post on Tue, 21 Jun 2005 titled "Newbie
needs help"?
> Probably lots of spam bouncing off that non existent address.
That reminds me, and IIRC you have explained this before however I forgot, how
and where is flipperhome.dyndns.org hosted? And something you probably didn't
explain before, what is the dyndns.org part all about?
> >> I've bread boarded it but had to use a much larger cathode resistor or
> >> else it severely distorted past, oh say, 50% mod. Ironically, less
> >> power meant 'more audio drive' because of the larger cathode voltage.
> >> That, and wanting enough sensitivity for cell phone/small MP3 players,
> >> meant a preamp.
> >
> >I wonder if the original cathode resistor value was actually chosen for the
> >12AU7 as the schematic says? Other dual triodes are mentioned on the
> >schematic
> >and the value shown might be more appropriate for a 12AT7 or something like
> >that.
>
> I would think it was as it's even worse for all the others mentioned,
> which is a shame because I kinda wanted to use the less expensive 6DT8
> (12AT7 clone) but Po is less. Although, it looks like I underestimated
> the range due to severe interference in and around my house, plus a
> power ground problem, so a 6DT8 might be good enough once things are
> cleaned up.
That's not obvious to me without either building it, or simulating it. Don't
forget that the 12AT7 has a considerably higher mu than the 12AU7, there are
several opposing forces at work here.
> On the other hand, there's not much cost difference visa vie a 6AU7.
Can't say I ever heard of the 6AU7 before, I'll have to look it up.
> >From your description of the problem it sounds like the voltage on the
> >modulator
> >plate was too low, limiting the amount of positive modulation available?
>
> I'm not sure which pin you're referencing 'positive' to. The
> modulating triode plate swings all over the place, of course, and
> 'positive' is when trying to 'reduce' Po.
I wasn't referring to any pin, I was referring to modulation in the positive
direction. Positive modulation occurs when the voltage across the top tube
increases.
> Too low an Rk places idle too high and it grid clips when trying to
> hit peak Po.
Yes, that's essentially what I meant by "limiting the amount of positive
modulation available".
It sounds like an IC OP Amp circuit to sense the DC voltage at the plate of the
modulator tube, compare it with a "reference" provided by a potentiometer, and
use the output to control the grid bias of the modulator tube might be useful.
Then you could easily dial in the desired voltage split between the two tubes,
which wouldn't be dependent of the characteristics of the particular tube you
happened to be using.
> > Did
> >you have any problem with the oscillator cutting off on serious negative
> >modulation? It occurred to me that shunting the modulator tube with a
> >resistor
> >of appropriate value would eliminate any possibility of the oscillator
> >dropping
> >out.
>
> Yeah, it can cutoff. Right now I've got 1Meg there for a 'smidgen' of
> RF.
>
> >Another twin triode approach would be to connect the oscillator and
> >modulator
> >triodes in parallel, with the B+ feed taken through an audio choke, a.k.a.
> >Hessing modulation. The Hessing configuration changes around some of the
> >modulation tradeoffs, reducing the possibility of oscillator dropout, and
> >providing greater carrier power at the same B+ voltage. The down side is
> >that
> >it requires an expensive audio inductor and doesn't have the elegant
> >simplicity
> >of the series connection, although the circuit complexity is the same other
> >than
> >the addition of the audio inductor, which also serves to reduce hum from the
> >B+
> >supply.
>
> Yeah, the audio choke is the problem there.
I assume that the problem you are talking about is that an audio choke would add
a couple of dollars to the cost of the transmitter, or are there other problems
you see?
> I've got a double RC on B+, so it's pretty quiet, and the hum is
> coming from the lousy power ground in my upstairs workroom. Goes away
> if I pull the plug on my notebook providing the audio but it's battery
> doesn't last long.
This is a perennial problem with these transmitters. The power wiring of the
computer becomes part of the antenna/ground system and the rectifier bridge in
the computer power supply modulates the RF at 120 Hz. Suitable inductors in the
audio signal and ground leads between the computer and transmitter may help.
> Your instinct might be to suggest a ground loop but I have no chassis
> on the breadboard and no earth there, so no loop. This usually clears
> up when I put 'em in the chassis with a proper earth.
My instinct is the power supply diodes modulating the RF at the power line
frequency or 2X the power line frequency.
> I'll give you another example. Earlier I was listening to it when, pop
> HMMMMMMMMMMMMMM increased maybe 10dB. I almost went upstairs to see if
> something had 'broke' but 15 minutes later it stopped as suddenly as
> it started.
This could be some other device in the house automatically turning on and again
effectively modifying the ground system at a 120 Hz rate.
> It's nothing in my house as I've been through this before and even
> called out the power company, who's only contribution was to confirm
> "boy, there sure is a lot of noise on the lines" (with my house
> severed). They don't promise 'quiet' power, just 'power'.
Are you sure it's nothing in your house, it sounds to me like it is your
computer and possibly other devices in your house? Remember that the problem
isn't noise generatd by your computer, it's the power supply in the computer
modulating the RF being radiated by the transmitter.
> If that weren't enough I'm sort of 'in between' metropolitan areas and
> get enough residual carrier from just about every station in the state
> to heterodyne with whatever frequency you broadcast on. Just now, as I
> was listening, a station suddenly appeared on my supposedly 'clear'
> frequency, and I mean loud enough to actually listen to when I turned
> off the transmitter. I don't know, maybe a cloud somewhere moved.
Yes I have noted the same problem, especially at night. There is always a beat
note and if your transmitter frequency isn't precisely tuned to put the beat at
an infrasonic frequency there will be an annoying tone heard. Crystal control
of the transmitter will mitigate this problem.
> I say, it's a bloody nightmare trying to do anything on the AM band
> here.
>
> >> Frequency response is better with a smaller (1nF vs 10nF) RF bypass
> >> and range is better tacking the (obviously untuned) wet noodle antenna
> >> onto the other side of the coil.
> >
> >I agree 10 nF is way too large, your 1.0 nF sounds about right, I probably
> >would
> >have gone with 1.5 nF on general principles, your 1 nF is probably better.
>
> Good to have a concurring opinion.
>
> >I'm not sure which end of the coil is which, however range should be best
> >when
> >the "wet noodle" is connected to the long end of the coil.
>
> Yes, that's the end I mean. The text file showed it on the short
> (plate) end, as do the typical heptode 'wireless record player'
> schematics. That made me wonder if there was some reason they all seem
> to be on the plate but, as far as I can tell, it works best on the
> 'long end'.
Yes, which way around the coil is connected in these transmitters affects both
the audio band width and the range. I find that the audio bandwidth is greater
when the long end of the coil is connected to the grid because the circuit Q is
lowered, however the range tends to be lower because more power is absorbed by
the grid resistor. The grid resistor tends to be the major load on these
transmitters, not the antenna, which is important for linearity and proper
modulation when using grid modulation as with the simple 6BE6 type transmitters.
The high level series modulation used in this dual triode circuit should be
relatively insensitive to loading.
> > I wouldn't say that
> >the "wet noodle" is untuned since it is connected directly across the plate
> >tank
> >where it serves as part of the tuning capacitance, although the impedance
> >matching is probably far from ideal. I suppose it is a matter of personal
> >preference whether or not to call this a tuned antenna.
>
> Yes, I mean impedance matched.
>
>
> >> http://flipperhome.dyndns.org/Twin%20Triode%20AM%20Transmitter.htm
> >>
> >> I haven't put it in the box yet because I keep trying to find a way to
> >> do one with 2 7-pin tubes, in particular, using a pentode under a
> >> triode but I can't find a suitable 7 pin triode.
> >
> >How about using a 6C4, which is half of a 12AU7 IIRC.
>
> How the hell did I miss that one? I've even GOT a couple. sheesh.
>
> > Why do you want to use a
> >pentode modulator?
>
> Gain and the screen allows a bit of 'fine tuning' the cathode voltage.
>
> > The triode modulator is probably part of what provides the
> >greater linearity,
>
> If you mean vs a heptode I think just about anything would be more
> linear past 50% mod ;)
No, I meant a triode vs. a pentode used as the modulator, lower tube, in this
circuit.
> > and of course provides a lower impedance to drive that 1 nF
> >capacitor, extending the high frequency audio response.
>
> I did wonder about that but when trying to check I ran into all sorts
> of problems with this poor old notebook doing wonky things.
>
> Circuitmaker says 10kHz is down 9dB even with the 1nF but reducing the
> oscillator grid resistor to 22k brings it up to -3dB, at the cost of a
> 10dB power loss. Hmm, well, keeping the 100k and dropping C1 from 33pF
> to 6pF gets to about -3.7dB but with a doubling of Po. That's what I
> just put in and it does sound better. Hmm, the 1Meg 'no cutoff'
> resistor costs .7dB.
That sounds like the effect of the oscillator tank circuit Q that I mentioned a
few paragraphs up.
> Don't know if that would work trying to tune the entire AM band, on
> the low end, but it's probably ok for the 'reduced range' slug tuned
> coil only thing, like I did with the 'One Tube Wonder," since I run it
> on the upper end.
>
> With the same oscillator and matching idle current circuitmaker shows
> a 6AU7 pentode modulator down 5dB. That's worse but not so much as one
> might have thought from plate resistance.
>
> Well, I could try boosting the upper end with an Rk bypass.
That would presumably help, but a triode makes a better, more linear, modulator.
> > Is the purpose of the
> >pentode to eliminate the need for a preamp? If only a small amount of
> >additional gain is needed how about using a triode with a higher
> >transconductance, like the 6AB4?
>
> I don't have one to try.
IIRC the 6AB4 is half of a 12AT7.
> When I say "not much" I mean in comparison to what (audio) people
> might generally 'expect' but grid swing for 'near' 100% mod is about
> 8V, making x8 minimum for 1V in. That's a heck of a lot to 'make up'
> with just more transconductance.
Quite true, that's a lot of gain that's needed!
I like the triple triode compactron approach, or maybe a caomactron with two
triodes and a sharp cutoff pentode. The two triodes could serve as the
modulator and audio preamp, and the pentode as an "electron coupled" crystal
controlled oscillator.
> >> Anyway, I went ahead last night and punched out one of the sockets to
> >> 9 pin so it's going in there sooner or later. Now I need to decide if
> >> I'll use an 'expensive' 12AU7, which I've got, or a 6(7)au7 I'd have
> >> to buy and decide if I'm going to finish that 'tuner cap hack' so the
> >> dial does something.
> >
> >Oops, there goes the two 7-pin tube approach!
>
> Not really. I have 2 more of the same UHF converters and one is even
> an honest to goodness Mallory. The other is an "Arvin" but without the
> 'logo' you couldn't tell them apart.
>
> > For a really 'expensive' tube how
> >about making it a 'Triple Triode' AM Transmitter using one of the several
> >triple
> >triode compactrons, using the third section as the preamp?
>
> Actually, that was a hope of mine because one of the tube discount
> dealers routinely has the 12AC10, triple '12AT7', on 'dollar days'.
> Wrong filament, though. The 6AC10 is 5 bucks but since that would
> replace 'two' tubes it's still a good deal, just not 'a buck'.
> However, power output is less (same problem as the 6DT8)
>
> Could still be done sort of 'as is' with a 6U10 containing two '12AU7'
> and a '12AX7'. It's $5.75
Could you share the name and URL of this "tube discount dealer", I'm interested
in some of the inexpensive compactrons available on "dollar days".
> Problem is it would be pushing the converter filament power (600mA vs
> 525mA) but might be okay since I don't think I'm using nearly the same
> B+ current. Could maybe exchange some heat <--- here for some heat
> ---> there.
>
> What I don't understand is why a compactron 'triple' draws 4 times the
> 'per triode' heater current, unless they did it just to be 600mA
> string compatible. What the hell, there's plenty of oil, gas, and
> coal, right?
Right, I assume that is the reason, there were also 450 mA series string
Televisions. I don't know much about series string Television sets, I would
have to hazard a guess that the 600 mA strings were used in color sets and the
450 mA strings in B&W sets that had fewer tubes?
> I also don't have a punch for the socket hole.
How about the compactron socket, non PC board variety, these seem to be hard to
come by, luckily I have two available.
John Byrns
flipper <fli...@fish.net> wrote:
> On Sun, 14 Aug 2011 17:28:25 -0500, John Byrns <byr...@sbcglobal.net>
> wrote:
>
> >In article <beie47lpvu5v90v4s...@4ax.com>,
> > flipper <fli...@fish.net> wrote:
> >
> >> On Sat, 13 Aug 2011 16:05:03 -0500, John Byrns <byr...@sbcglobal.net>
> >> wrote:
> >>
> >> >In article <dduc47t939q95vce1...@4ax.com>,
> >> > flipper <fli...@fish.net> wrote:
> >> >
>
> >> Probably lots of spam bouncing off that non existent address.
> >
> >That reminds me, and IIRC you have explained this before however I forgot,
> >how
> >and where is flipperhome.dyndns.org hosted? And something you probably
> >didn't
> >explain before, what is the dyndns.org part all about?
>
> It's a little server here at my house and the dyndns.org thing is a
> "dynamic DNS" ('free') server service. That's how it can be found 'by
> name' rather than having to post whatever the current IP is but for
> them to do it there has to be a 'domain', hence dyndns.org on the end.
> They now have other 'domain' choices but back when I first signed up
> that was the only one.
Before I commented I went to dyndns.org, I didn't dig deeply, but I didn't
notice any mention of "free"? How does this service track your ever changing IP
address, do you install a little app that connects with the dyndns.org server as
needed and keeps it updated?
> I was using a modified "webpal" running the ARM flavor of Debian Linux
> but just a month or so ago it scrambled itself and rather than screw
> with that thing again I pulled out an old P133 notebook and now run
> Windows98 with a Savant web server.
That's about what I remembered except for the Windows98 part, too bad the
"webpal" scrambled itself.
> >> >> I've bread boarded it but had to use a much larger cathode resistor or
> >> >> else it severely distorted past, oh say, 50% mod. Ironically, less
> >> >> power meant 'more audio drive' because of the larger cathode voltage.
> >> >> That, and wanting enough sensitivity for cell phone/small MP3 players,
> >> >> meant a preamp.
> >> >
> >> >I wonder if the original cathode resistor value was actually chosen for
> >> >the
> >> >12AU7 as the schematic says? Other dual triodes are mentioned on the
> >> >schematic
> >> >and the value shown might be more appropriate for a 12AT7 or something
> >> >like
> >> >that.
> >>
> >> I would think it was as it's even worse for all the others mentioned,
> >> which is a shame because I kinda wanted to use the less expensive 6DT8
> >> (12AT7 clone) but Po is less. Although, it looks like I underestimated
> >> the range due to severe interference in and around my house, plus a
> >> power ground problem, so a 6DT8 might be good enough once things are
> >> cleaned up.
> >
> >That's not obvious to me without either building it, or simulating it.
> >Don't
> >forget that the 12AT7 has a considerably higher mu than the 12AU7, there are
> >several opposing forces at work here.
>
> Yes, well, it sure wasn't obvious to me either and now that I look
> back I don't think it was a valid comparison with some inconsistent
> component values. Now I'm get about half the power with 12AT7s in the
> simulations.
If I build it with a triple triode compactron, triodes similar to 12AU7, then I
won't be able to plug in say a 12AT7 to see the truth of this, bummer.
> >> >From your description of the problem it sounds like the voltage on the
> >> >modulator
> >> >plate was too low, limiting the amount of positive modulation available?
> >>
> >> I'm not sure which pin you're referencing 'positive' to. The
> >> modulating triode plate swings all over the place, of course, and
> >> 'positive' is when trying to 'reduce' Po.
> >
> >I wasn't referring to any pin, I was referring to modulation in the positive
> >direction. Positive modulation occurs when the voltage across the top tube
> >increases.
>
> OK. Well, to me, 'positive' is with respect to something. Since B+ is
> 'fixed' more "voltage across" could be cathode 'more negative', if you
> see what I mean.
>
> Just explaining why I wasn't sure.
Positive modulation means increasing the "carrier" level, nagative modulation
means reducing the "carrier" level towards zero.
> >> Too low an Rk places idle too high and it grid clips when trying to
> >> hit peak Po.
> >
> >Yes, that's essentially what I meant by "limiting the amount of positive
> >modulation available".
> >
> >It sounds like an IC OP Amp circuit to sense the DC voltage at the plate of
> >the
> >modulator tube, compare it with a "reference" provided by a potentiometer,
> >and
> >use the output to control the grid bias of the modulator tube might be
> >useful.
> >Then you could easily dial in the desired voltage split between the two
> >tubes,
> >which wouldn't be dependent of the characteristics of the particular tube
> >you
> >happened to be using.
>
> I don't think that would help the 'too high idle' issue because you
> still have to pick the idle point.
??? Well yes, it seems obvious that you have to pick the idle point, am I
missing something?
I have an alternate idea for stabilizing the idle point that doesn't require an
evil OP Amp IC. My idea is to eliminate the modulator tube cathode resistor and
directly ground the modulator cathode, then remove the modulator tube grid
resistor and replace it with two resistors, one connected to the midpoint of the
two tubes, and the second one connected to the negative supply. The ratio of
the two resistors would be chosen to set the idle point to the desired value.
> I think it's more than a couple of bucks and then there's a potential
> space problem.
??? Well yes, it seems obvious that you have to pick the idle point, am I
missing something?
I have an alternate idea for stabilizing the idle point that doesn't require an
evil OP Amp IC. My idea is to eliminate the modulator tube cathode resistor and
directly ground the modulator cathode, then remove the modulator tube grid
resistor and replace it with two resistors, one connected to the midpoint of the
two tubes, and the second one connected to the negative supply. The ratio of
the two resistors would be chosen to set the idle point to the desired value.
> >> I've got a double RC on B+, so it's pretty quiet, and the hum is
> >> coming from the lousy power ground in my upstairs workroom. Goes away
> >> if I pull the plug on my notebook providing the audio but it's battery
> >> doesn't last long.
> >
> >This is a perennial problem with these transmitters. The power wiring of
> >the
> >computer becomes part of the antenna/ground system and the rectifier bridge
> >in
> >the computer power supply modulates the RF at 120 Hz.
>
> I wouldn't have thought 120Hz rectifier noise could make it through
> the switcher.
It isn't "rectifier noise", the 60 Hz line frequency switches the rectifiers on
and off 120 times a second, connecting and disconnecting the electrical wiring
of your house from the transmitter RF wise, 120 times a second. RF capacitors
across the diodes in the power supply can help minimize this by maintaining a
continuous connection across the rectifier diodes at RF frequencies.
> > Suitable inductors in the
> >audio signal and ground leads between the computer and transmitter may help.
>
> I'll keep that in mind but, as mentioned, the issue has usually gone
> away once I get things in a properly grounded chassis.
>
> >> Your instinct might be to suggest a ground loop but I have no chassis
> >> on the breadboard and no earth there, so no loop. This usually clears
> >> up when I put 'em in the chassis with a proper earth.
> >
> >My instinct is the power supply diodes modulating the RF at the power line
> >frequency or 2X the power line frequency.
> >
> >> I'll give you another example. Earlier I was listening to it when, pop
> >> HMMMMMMMMMMMMMM increased maybe 10dB. I almost went upstairs to see if
> >> something had 'broke' but 15 minutes later it stopped as suddenly as
> >> it started.
> >
> >This could be some other device in the house automatically turning on and
> >again
> >effectively modifying the ground system at a 120 Hz rate.
>
> That's the first thing I check but, besides knowing there are things
> outside my property producing huge amounts of hash, I don't have
> anything that comes on for 15 minutes 'only once' in 6 hours.
>
> >> It's nothing in my house as I've been through this before and even
> >> called out the power company, who's only contribution was to confirm
> >> "boy, there sure is a lot of noise on the lines" (with my house
> >> severed). They don't promise 'quiet' power, just 'power'.
> >
> >Are you sure it's nothing in your house,
>
> I can guarantee it was nothing in my house when the power company came
> out because they severed it from the grid while checking the
> distribution transformer. And while they were doing it you could
> barely get *any* AM station for all the deafening noise interference.
>
> > it sounds to me like it is your
> >computer and possibly other devices in your house? Remember that the
> >problem
> >isn't noise generatd by your computer, it's the power supply in the computer
> >modulating the RF being radiated by the transmitter.
>
> I understand what you're saying but there's more than one problem
> here.
Yes, it sounds like you are fighting a noise problem, in addition to the
"modulation hum" problem I was discussing.
> >> If that weren't enough I'm sort of 'in between' metropolitan areas and
> >> get enough residual carrier from just about every station in the state
> >> to heterodyne with whatever frequency you broadcast on. Just now, as I
> >> was listening, a station suddenly appeared on my supposedly 'clear'
> >> frequency, and I mean loud enough to actually listen to when I turned
> >> off the transmitter. I don't know, maybe a cloud somewhere moved.
> >
> >Yes I have noted the same problem, especially at night. There is always a
> >beat
> >note and if your transmitter frequency isn't precisely tuned to put the beat
> >at
> >an infrasonic frequency there will be an annoying tone heard. Crystal
> >control
> >of the transmitter will mitigate this problem.
>
> Not so easy to tune a crystal, though ;)
True, however that's the beauty of it!
> Sounds a bit circular to me.
>
> I'm seeing about a 3dB difference in circuitmaker
>
> >> > and of course provides a lower impedance to drive that 1 nF
> >> >capacitor, extending the high frequency audio response.
> >>
> >> I did wonder about that but when trying to check I ran into all sorts
> >> of problems with this poor old notebook doing wonky things.
> >>
> >> Circuitmaker says 10kHz is down 9dB even with the 1nF but reducing the
> >> oscillator grid resistor to 22k brings it up to -3dB, at the cost of a
> >> 10dB power loss. Hmm, well, keeping the 100k and dropping C1 from 33pF
> >> to 6pF gets to about -3.7dB but with a doubling of Po. That's what I
> >> just put in and it does sound better. Hmm, the 1Meg 'no cutoff'
> >> resistor costs .7dB.
> >
> >That sounds like the effect of the oscillator tank circuit Q that I
> >mentioned a
> >few paragraphs up.
>
> Probably, but why would Q go down with lower C?
What C are you talking about?
> >> Don't know if that would work trying to tune the entire AM band, on
> >> the low end, but it's probably ok for the 'reduced range' slug tuned
> >> coil only thing, like I did with the 'One Tube Wonder," since I run it
> >> on the upper end.
> >>
> >> With the same oscillator and matching idle current circuitmaker shows
> >> a 6AU7 pentode modulator down 5dB. That's worse but not so much as one
> >> might have thought from plate resistance.
> >>
> >> Well, I could try boosting the upper end with an Rk bypass.
> >
> >That would presumably help, but a triode makes a better, more linear,
> >modulator.
>
> I meant the triode. Well, both but I was looking at the triode when I
> made the comment.
>
> >> > Is the purpose of the
> >> >pentode to eliminate the need for a preamp? If only a small amount of
> >> >additional gain is needed how about using a triode with a higher
> >> >transconductance, like the 6AB4?
> >>
> >> I don't have one to try.
> >
> >IIRC the 6AB4 is half of a 12AT7.
>
> In that case I've simmed it, except with a 12AT7 in both positions.
>
> Well, putting it in just the bottom 'helps' but only gets about half
> way there.
>
> >> When I say "not much" I mean in comparison to what (audio) people
> >> might generally 'expect' but grid swing for 'near' 100% mod is about
> >> 8V, making x8 minimum for 1V in. That's a heck of a lot to 'make up'
> >> with just more transconductance.
> >
> >Quite true, that's a lot of gain that's needed!
> >
> >I like the triple triode compactron approach, or maybe a caomactron with two
> >triodes and a sharp cutoff pentode. The two triodes could serve as the
> >modulator and audio preamp, and the pentode as an "electron coupled" crystal
> >controlled oscillator.
>
> Got one in mind? And why the pentode osc?
I was thinking a 6AV11 triple triode. The pentode was for the parallel
modulation version where a pentode operating as an electron coupled oscillator
might provide for deeper negative modulation then a simple triode oscillator
could do. A separate floating screen supply would be required to make the
pentode oscillator work with the series modulation scheme though. I was
thinking of something like a 6BH11, although the pentode section is a little
more muscular than I would want.
> >Could you share the name and URL of this "tube discount dealer", I'm
> >interested
> >in some of the inexpensive compactrons available on "dollar days".
>
> Sure.
>
> http://www.abcvacuumtubes.com/
>
> http://www.thetubecenter.com/
Thanks, both of those sites appear to be the same business, I guess they get
more "shelf space" that way.
John Byrns
flipper <fli...@fish.net> wrote:
> On Sun, 14 Aug 2011 17:28:25 -0500, John Byrns <byr...@sbcglobal.net>
> wrote:
>
> >In article <beie47lpvu5v90v4s...@4ax.com>,
> > flipper <fli...@fish.net> wrote:
> >
> <snip>
>
> >> >> I've bread boarded it but had to use a much larger cathode resistor or
> >> >> else it severely distorted past, oh say, 50% mod. Ironically, less
> >> >> power meant 'more audio drive' because of the larger cathode voltage.
> >> >> That, and wanting enough sensitivity for cell phone/small MP3 players,
> >> >> meant a preamp.
> >> >
> >> >I wonder if the original cathode resistor value was actually chosen for
> >> >the
> >> >12AU7 as the schematic says? Other dual triodes are mentioned on the
> >> >schematic
> >> >and the value shown might be more appropriate for a 12AT7 or something
> >> >like
> >> >that.
> >>
> >> I would think it was as it's even worse for all the others mentioned,
> >> which is a shame because I kinda wanted to use the less expensive 6DT8
> >> (12AT7 clone) but Po is less. Although, it looks like I underestimated
> >> the range due to severe interference in and around my house, plus a
> >> power ground problem, so a 6DT8 might be good enough once things are
> >> cleaned up.
> >
> >That's not obvious to me without either building it, or simulating it.
> >Don't
> >forget that the 12AT7 has a considerably higher mu than the 12AU7, there are
> >several opposing forces at work here.
> >
>
> Following up on the 12AT7 (and clones) 'power' and Rk, just look at
> the plate curves. In particular, since we aren't using positive grid
> drive, look at 0 grid plate current. It depends, of course, on plate
> voltage but the ratio is similar regardless so just pick 100V to
> compare and, rough eyeballing it, a 12AU7 will do about 12mA but the
> 12AT7 will only do 8mA. Po will be current squared so the ratio comes
> out around .44 for the 12AT7.
However what we really care about is the 0 grid plate saturation voltage at a
more reasonable peak modulator current on the order of about 3.33 mA where the
difference between the 12AU7 and 12AT7 is small.
> My sims don't match that simple analysis 'exactly', partly because I
> pick 'standard' resistor values and the oscillator is then a 12AT7
> too, but it's 'in the ball park'.
>
> I don't know if it's 'right' but I calculate plate RF power by the
> generated Vrms times average cathode current and, using that formula,
> get about 45mW with the 12AT7s, which should be plenty enough for a
> room transmitter. I calculate about 80mW with the 12AU7s.
I'd have to give it some thought before forming an opinion of whether your
method of calculating power might be right. It does make me nervous because it
doesn't make any attempt to take efficiency into account. Where are you
measuring "Vrms"? What values of "average cathode current" and "Vrms" did you
obtain for the 12AU7 and 12AT7?
> I also got to thinking about where the distortion is and, unless one
> does something silly with the preamp, it's overwhelmingly dominated by
> the modulator triode because it's being operated in what I call 'power
> mode'. I.E. at max mod current is swinging from max to cutoff, the
> worse case for distortion, and it's not all that 'terrific' at reduced
> levels either because it simply isn't biased, nor can it be, for 'best
> (low) distortion'.
I assume the problem is at the cutoff end, max current isn't that great. The
parallel or Heising modulation approach eliminates the need to operate the tube
near cutoff, eliminating that source of distortion. The parallel approach does
have the down side of using the modulator tube less efficiently, although that
shouldn't be an issue in the type of transmitter we are talking about here.
> I mean, if we were doing a preamp we'd use a high value load
> impedance, with idle point picked for the (most) 'linear' operation,
> so the signal component was 'small' in comparison but that directly
> conflicts with the need to produce power.
>
> Btw, my idea of using a 6ME8 (or other sheet beam tube) to variably
> (modulation) subtract carrier should solve that problem since the
> oscillator remains constant while one 'side' is simply subtracted from
> the other.
Since you hadn't mentioned the 6ME8 in this thread before, I went looking for
where you mentioned it and found that it was in your "Local Osc question" thread
from back in February. In that thread you also said "However, after looking at
the heptode and dual control pentode curves I'm now thinking there's a sneakier
way to do it, which is to use a PP transformer driven by the plate and
screen(s)." Have you seen this design?
http://www.radiomuseum.org/forum/hi_fi_am_transmitter.html
> Just to get my feet wet I've made it oscillate but am now working on
> an idea to get a sine, instead of semi-square, out of it since I'd
> like to avoid tacking on tuned circuits to clean up the RF.
John Byrns
flipper <fli...@fish.net> wrote:
> On Fri, 19 Aug 2011 23:29:25 -0500, John Byrns <byr...@sbcglobal.net>
> wrote:
>
> >In article <cu9t475fc6m8avao5...@4ax.com>,
> > flipper <fli...@fish.net> wrote:
> >
> >> On Sun, 14 Aug 2011 17:28:25 -0500, John Byrns <byr...@sbcglobal.net>
> >> wrote:
> >>
> >> >In article <beie47lpvu5v90v4s...@4ax.com>,
> >> > flipper <fli...@fish.net> wrote:
<Snip>
> >> Following up on the 12AT7 (and clones) 'power' and Rk, just look at
> >> the plate curves. In particular, since we aren't using positive grid
> >> drive, look at 0 grid plate current. It depends, of course, on plate
> >> voltage but the ratio is similar regardless so just pick 100V to
> >> compare and, rough eyeballing it, a 12AU7 will do about 12mA but the
> >> 12AT7 will only do 8mA. Po will be current squared so the ratio comes
> >> out around .44 for the 12AT7.
> >
> >However what we really care about is the 0 grid plate saturation voltage at
> >a
> >more reasonable peak modulator current on the order of about 3.33 mA where
> >the
> >difference between the 12AU7 and 12AT7 is small.
>
> I don't follow that at all.
>
> Modulator plate voltage will be 'whatever' to pull the 'semi follower'
> cathode current across the oscillator (That's a bit of an
> approximation that would be even closer to true with a pentode
> modulator that's close to a variable CCS). With the 12AU7 that's
> around 130V. It'll be different with the 12AT7 in both positions,
> which is why the simple comparison is only 'ball park'.
I get the feeling that we are interpreting the operation of this circuit in
different ways. Many of your comments seem to indicate that you believe that it
uses cathode modulation, as was popular in many Television transmitters of the
vacuum-tube era. On the other hand I believe this circuit is a standard "high
level plate modulated" design. The operation of the two circuits is completely
different and I think we are going to have trouble following what each oter are
saying unless we can reconcile the actual operation of the circuit.
<Snip>
> >> I don't know if it's 'right' but I calculate plate RF power by the
> >> generated Vrms times average cathode current and, using that formula,
> >> get about 45mW with the 12AT7s, which should be plenty enough for a
> >> room transmitter. I calculate about 80mW with the 12AU7s.
> >
> >I'd have to give it some thought before forming an opinion of whether your
> >method of calculating power might be right. It does make me nervous because
> >it
> >doesn't make any attempt to take efficiency into account.
>
> I said plate power, not radiated. From what I understand that's where
> the FCC part 15 100mW (max) is measured.
Yes, I realized that you weren't talking about radiated power, but felt that you
were taking a very roundabout way of calculating "plate power"! "Plate power"
is easily calculated by multiplying the voltage across the RF tube by the
current through it, without modulation. For example, assume a 160 volt power
supply, further assume that the cathode of the oscillator tube under carrier
only conditions rests at 100 volts, and finally that the current through the
tube is 1.67 mA. This gives us 160 volts - 100 volts = 60 volts across the
oscillator tube, multiply 60 volts times 1.67 mA and we have a "plat power" of
100 mW as per the FCC part 15 regulations.
> > Where are you
> >measuring "Vrms"?
>
> The osc plate.
>
> > What values of "average cathode current" and "Vrms" did you
> >obtain for the 12AU7 and 12AT7?
>
> The 6pF/100k 12AU7 version is, at 0 mod, approx 20Vrms on the osc
> plate with 3.9mA through Rk. The 12AT7 variant is about 14.8Vrms and
> 2.6mA.
>
> Modulated there's 'more' on the peaks and 'less' on the troughs but it
> averages the same. Which is a good thing because it should ;)
I meant to ask, but forgot, what is the oscillator cathode voltage for the 12AU7
and 12AT7 under the same conditions as you measured Vrms and the current through
Rk?
Are these numbers taken from your breadboard or your "sims"?
<Snip>
> >> I also got to thinking about where the distortion is and, unless one
> >> does something silly with the preamp, it's overwhelmingly dominated by
> >> the modulator triode because it's being operated in what I call 'power
> >> mode'. I.E. at max mod current is swinging from max to cutoff, the
> >> worse case for distortion, and it's not all that 'terrific' at reduced
> >> levels either because it simply isn't biased, nor can it be, for 'best
> >> (low) distortion'.
> >
> >I assume the problem is at the cutoff end, max current isn't that great.
>
> I don't believe it's a matter of the current magnitude. It's that your
> swinging from 0 grid to cutoff. 'The whole ball of wax', so to speak,
> whatever it is and they just aren't 'linear' over full scale.
The point I was trying to make was that if you look at the curves, the greatest
nonlinearity is when you try to drive the tube towards cutoff.
> > The
> >parallel or Heising modulation approach eliminates the need to operate the
> >tube
> >near cutoff, eliminating that source of distortion. The parallel approach
> >does
> >have the down side of using the modulator tube less efficiently, although
> >that
> >shouldn't be an issue in the type of transmitter we are talking about here.
>
> I was thinking one might be able to accomplish the same thing with a
> CCS in parallel across the oscillator, to raise the idle point so the
> actual modulating current is a 'fraction' of the total, but that would
> take a beefier tube and simply burn off the extra current as heat. Not
> very efficient.
We are only talking about a 100 mW transmitter, the 12AU7 modulator should be
able to easily handle the added current, the main problem is that the saturation
voltage, at 0 grid, will be higher with the added current, necessitating
increasing the power supply voltage, or other change, to compensate, further
increasing the load on the 12AU7 modulator, although it should still be up to
the job in a 100 mW transmitter.
The way the big boys dealt with this problem in real broadcast transmitters was
to operate the lower tube in class D rather than class A. This has two
advantages, first it eliminates the effects of tube nonlinearity, and second it
greatly increases the efficiency since the modulator tube is either ON or OFF,
none of this analog stuff. A low pass audio filter is required between the
plate of the modulator and the cathode of the RF tube to remove the switching
frequency however.
<Snip>
> >Since you hadn't mentioned the 6ME8 in this thread before, I went looking
> >for
> >where you mentioned it and found that it was in your "Local Osc question"
> >thread
> >from back in February. In that thread you also said "However, after looking
> >at
> >the heptode and dual control pentode curves I'm now thinking there's a
> >sneakier
> >way to do it, which is to use a PP transformer driven by the plate and
> >screen(s)." Have you seen this design?
> >
> >http://www.radiomuseum.org/forum/hi_fi_am_transmitter.html
>
> Yeah, someone posted that one back then too.
That would have been me again, I didn't remember your mentioning this idea
before, or my posting the link, I guess a few relevant brain cells bit the dust.
> I think the 6ME8 might be able to do more power and ought to be more
> linear, though.
Why would more power be needed, at least if you are following the FCC rules?
> Amusingly he mentions the 6ME8 too but feels that isn't any 'fun'
> since "that's what it was made for."
I suspect that it might be more "fun" than he feels it would be, interesting
problems always crop up in projects of that sort.
> >> Just to get my feet wet I've made it oscillate but am now working on
> >> an idea to get a sine, instead of semi-square, out of it since I'd
> >> like to avoid tacking on tuned circuits to clean up the RF.
>
> I've been trying to work on the class A oscillator thing but spice is
> producing insanity. I get things like it claiming 2kV swings and
> negative screen voltages so there's clearly something wrong with the
> models when used that way.
What is the "class A oscillator thing", I don't remember your mentioning this, I
suppose a few more brain cells than I thought bit the dust?
John Byrns
flipper <fli...@fish.net> wrote:
> On Sun, 21 Aug 2011 11:57:11 -0500, John Byrns <byr...@sbcglobal.net>
> wrote:
>
> >In article <0sgu47pmpm0489eg2...@4ax.com>,
> > flipper <fli...@fish.net> wrote:
> >
> >> On Fri, 19 Aug 2011 23:29:25 -0500, John Byrns <byr...@sbcglobal.net>
> >> wrote:
> >>
> >> >In article <cu9t475fc6m8avao5...@4ax.com>,
> >> > flipper <fli...@fish.net> wrote:
> >> >
> >> >> On Sun, 14 Aug 2011 17:28:25 -0500, John Byrns <byr...@sbcglobal.net>
> >> >> wrote:
> >> >>
> >> >> >In article <beie47lpvu5v90v4s...@4ax.com>,
> >> >> > flipper <fli...@fish.net> wrote:
<Snip>
> >I get the feeling that we are interpreting the operation of this circuit in
> >different ways. Many of your comments seem to indicate that you believe
> >that it
> >uses cathode modulation, as was popular in many Television transmitters of
> >the
> >vacuum-tube era. On the other hand I believe this circuit is a standard
> >"high
> >level plate modulated" design. The operation of the two circuits is
> >completely
> >different and I think we are going to have trouble following what each oter
> >are
> >saying unless we can reconcile the actual operation of the circuit.
>
> Well, I don't know what 'names' to give it and, similarly, providing a
> name doesn't tell me the difference. I'm not familiar with them.
What I call "high level plate modulation" feeds the modulated oscillator or RF
amplifier tube plate circuit with a series combination the B+ voltage and the
audio modulating voltage.
What I call "cathode modulation" uses what is effectively a variable current
source, current controlled by the modulating audio, in the cathode circuit of
the oscillator or RF amp tube. The grid is referenced to the end of the current
source that isn't connected to the cathode, this causes the grid to also be
modulated in such a way that the voltage across the current source does not have
to be large.
> The way I see it the modulator tube is setting current which, of
> course, means voltage too if we presume the osc is a (relatively)
> resistive load.
>
> I'm guessing you say we're setting plate voltage and current follows.
Yes, if we are talking triode modulators I would say "we're setting plate
voltage and current follows", if we were talking pentode modulators I would say
that we are setting the current, however either way around works.
<Snip>
> >Yes, I realized that you weren't talking about radiated power, but felt that
> >you
> >were taking a very roundabout way of calculating "plate power"! "Plate
> >power"
> >is easily calculated by multiplying the voltage across the RF tube by the
> >current through it, without modulation. For example, assume a 160 volt
> >power
> >supply, further assume that the cathode of the oscillator tube under carrier
> >only conditions rests at 100 volts, and finally that the current through the
> >tube is 1.67 mA.
>
> Well, there's nothing 'at rest'. hehe. B+ is on the other side of a
> coil and plate is oscillating. And the osc cathode has 2Vrms on it as
> well.
>
> > This gives us 160 volts - 100 volts = 60 volts across the
> >oscillator tube, multiply 60 volts times 1.67 mA and we have a "plat power"
> >of
> >100 mW as per the FCC part 15 regulations.
>
> What happened to 'RMS'?
Nothing, it's still there, just doesn't have anything to do with plate input
power.
> Well, ok, if I take 'averages' then the 12AU7 'plate power' is 64mW.
I took the term "plate power" to mean DC input power to the final stage, I
believe that's the way the FCC specifies it.
> >> > Where are you
> >> >measuring "Vrms"?
> >>
> >> The osc plate.
> >>
> >> > What values of "average cathode current" and "Vrms" did you
> >> >obtain for the 12AU7 and 12AT7?
> >>
> >> The 6pF/100k 12AU7 version is, at 0 mod, approx 20Vrms on the osc
> >> plate with 3.9mA through Rk. The 12AT7 variant is about 14.8Vrms and
> >> 2.6mA.
> >>
> >> Modulated there's 'more' on the peaks and 'less' on the troughs but it
> >> averages the same. Which is a good thing because it should ;)
> >
> >I meant to ask, but forgot, what is the oscillator cathode voltage for the
> >12AU7
> >and 12AT7 under the same conditions as you measured Vrms and the current
> >through
> >Rk?
>
> Well, depends on how one defines "same conditions."
By "same conditions" I meant that Vrms, Icathode, and Vcathode for a given tube
were all measured under the same conditions.
> I'm setting them
> for max mod before clipping so Rk and current is different. Second,
> since they're 'pairs', the presented oscillator load is not
> necessarily the same.
>
> The present 12AU7 osc cathode is at 143Vav with 3.75mA and the 12AT7
> is at 140Vav with 2.16mA
Those voltages are somewhat higher than I expected, and the currents are
different than you stated above, so something has changed.
> >Are these numbers taken from your breadboard or your "sims"?
>
> Sims
You said in your original post to this thread that you had bread boarded the
circuit, have you compared Vrms, Icathode, and Vcathode on the bread board with
the same on the sims?
One problem with this transmitter circuit is that there is no way to adjust the
oscillator current, independently of the modulator plate voltage, like the
loading control does in a "real" transmitter does.
<snip>
> Btw, I discovered why I was getting confusing distortion numbers but
> I'll go into that in another message.
I will patiently await your explanation of the distortion.
<snip>
> >The way the big boys dealt with this problem in real broadcast transmitters
> >was
> >to operate the lower tube in class D rather than class A. This has two
> >advantages, first it eliminates the effects of tube nonlinearity, and second
> >it
> >greatly increases the efficiency since the modulator tube is either ON or
> >OFF,
> >none of this analog stuff. A low pass audio filter is required between the
> >plate of the modulator and the cathode of the RF tube to remove the
> >switching
> >frequency however.
>
> Yeah. A bit beyond a 'uhf converter box' special, though ')
A dedicated constructor could probably pack it all into the "uhf converter box".
I figure it could be done with 6 dual triodes, or 4 compactrons. Fitting in a
suitable power supply might be the major problem.
<snip>
> >> I've been trying to work on the class A oscillator thing but spice is
> >> producing insanity. I get things like it claiming 2kV swings and
> >> negative screen voltages so there's clearly something wrong with the
> >> models when used that way.
> >
> >What is the "class A oscillator thing", I don't remember your mentioning
> >this, I
> >suppose a few more brain cells than I thought bit the dust?
>
> Well, the typical cathode-grid oscillator operates semi class C, grid
> drive setting the bias and all that. The RF is dirty unless filtered,
> like with a second tuned tank. That's one reason I went with the plate
> oscillator to begin with.
Now you got me on the name thing, what is a "cathode-grid oscillator" and a
"plate oscillator"?
John Byrns
flipper <fli...@fish.net> wrote:
> On Tue, 16 Aug 2011 19:16:27 -0500, flipper <fli...@fish.net> wrote:
>
> <snip>
>
> >
> >On another note, since I have a preamp, and a triode mode pentode to
> >boot, meaning I could get much more gain, I ran some sims using
> >modulator plate to preamp cathode NFB. But it doesn't seems to help
> >much.
>
> I since tried a number of FB approaches and none seemed to 'work' but
> I think I've found out why. There's RF everywhere, and it isn't
> 'clean'.
>
> In particular, the modulator plate has what looks more like a
> sawtooth, partly because of the osc positive grid drive, and that
> splatters harmonics all over the spectrum. And, being a triode, it
> appears on the modulator cathode as well. And the grid, the preamp
> plate, and even it's grid. That covers everything, including potential
> FB sources.
>
> With no input at all, just carrier on, spice shows at least -40dB of
> 'signal' on every harmonic and 1kHz too (which spice thinks is the
> fundamental because I just 'delayed' the signal generator start till
> after the sim completes).
>
> The first question that arises is whether that appears in the AM
> envelope or is just a 'local artifact' of the circuit. Murphy's Law,
> of course, dictates it's in the envelope.
I would expect that it doesn't affect the "envelope", but does produce
additional spectra in the output, at harmonics of the carrier frequency.
> It can be improved by increasing the RF bypass but that screws the
> 10kHz response.
>
> There's a dramatic improvement by inserting a 47k grid stopper on the
> osc to limit grid current but then I worry about osc startup.
Did this "grid stopper" have any effect on "Vrms"?
> The best improvement is adding a 100mH (preferably more, 1H) choke to
> the modulator plate but where am I going to find a 100mH, much less
> 1H, choke good to 2 MHz? A quick check at Mouser only produced ones
> with self resonance in the 50kHz range.
>
> At any rate, with a 100mH choke (ideal model), a modest (100pF) Rk
> filter,
Where in the circuit is this "Rk" filter, which cathode?
> and NFB from modulator Rk to the input 'stereo mixer' I can
> get the sims to almost .1%, measured at the modulator Rk, limited by
> the preamp triode (which isn't terrific at 160V).
Did you try feedback from the modulator plate to the preamp cathode with a
dominant pole in the feedback path to knock down the "RF" grunge?
At what level, modulator plate swing, and modulation percentage, was the "almost
.1%" measured?
Did you make any measurements of distortion without the feedback connected?
Wouldn't it be better to measure the actual envelope distortion of the entire
transmitter system, rather than the audio at the modulator cathode, using an
envelope detector circuit, perhaps Patrick's cathode follower plus Ge diode?
John Byrns
flipper <fli...@fish.net> wrote:
> If it's not in the envelope then we wouldn't care, as far as 'audio
> quality' went.
>
> Unfortunately, even if it weren't 'normally' in the envelope having it
> get into the FB loop probably puts it there.
I would think only to the extent that nonlinearities in the feedback loop
generate distortion and grunge in the audio band, from the RF.
> >> It can be improved by increasing the RF bypass but that screws the
> >> 10kHz response.
> >>
> >> There's a dramatic improvement by inserting a 47k grid stopper on the
> >> osc to limit grid current but then I worry about osc startup.
> >
> >Did this "grid stopper" have any effect on "Vrms"?
>
> Oh, of course. Everything does. And Murphy's Law is in full force,
> meaning anything that improves 10kHz response, or distortion, reduces
> power output.
>
> >> The best improvement is adding a 100mH (preferably more, 1H) choke to
> >> the modulator plate but where am I going to find a 100mH, much less
> >> 1H, choke good to 2 MHz? A quick check at Mouser only produced ones
> >> with self resonance in the 50kHz range.
> >>
> >> At any rate, with a 100mH choke (ideal model), a modest (100pF) Rk
> >> filter,
> >
> >Where in the circuit is this "Rk" filter, which cathode?
>
> The modulator Rk.
>
> That's the only place there's a hope of getting something without RF
> riding all over it so the choke is to RF isolate the modulator plate
> from the osc cathode and then apply a little more filtering to the
> modulator cathode.
This statement that "the choke is to RF isolate the modulator plate from the osc
cathode" appears to contradict the previous statement that the filter is in "The
modulator Rk"?
> >> and NFB from modulator Rk to the input 'stereo mixer' I can
> >> get the sims to almost .1%, measured at the modulator Rk, limited by
> >> the preamp triode (which isn't terrific at 160V).
> >
> >Did you try feedback from the modulator plate to the preamp cathode with a
> >dominant pole in the feedback path to knock down the "RF" grunge?
>
> That was the first thing I tried (sim) but there's so much RF there
> it's almost impossible to filter.
You only need to knock it down to a level where it doesn't cause trouble. Also
the audio signal at the modulator plate is presumably much larger than at the
cathode so the flat loss of the feedback network, before any filtering, is
proportionately greater, greatly attenuating the "sawtooth".
> I don't think it matters, though, because both approaches presume the
> osc is a resistive load. I.E. if we 'control' the voltage then it
> better be resistive or current isn't linear. But if current visa vie
> voltage is linear then we can accomplish the same thing by controlling
> current.
I'm not sure what you are trying to say here? Are you simply saying that it
doesn't matter where the feedback is taken from, under the assumption that the
oscillator looks like a resistor to the modulator?
That raises two questions, one how much like a resistor does the oscillator
actually look, and two, how linear is "Vrms" vs. either voltage or current?
> >At what level, modulator plate swing, and modulation percentage, was the
> >"almost
> >.1%" measured?
> >
> >Did you make any measurements of distortion without the feedback connected?
>
> Sure, in spice. Feedback gave about a 12dB improvement, if you get the
> noise off it.
>
> I 'calibrated' the preamp gain for max mod at 1V so for 'nominal'
> distortion measurement I use 156mV, -16dB.
>
> As with everything else 'it depends' on everything else but typical,
> open loop, might be around .6% nominal up to between 4% and 5% at max
> mod.
>
> Frankly, it "sounds good," for an AM radio that is, but "since I had a
> preamp..." You know how that goes.
>
> >Wouldn't it be better to measure the actual envelope distortion of the
> >entire
> >transmitter system, rather than the audio at the modulator cathode, using an
> >envelope detector circuit, perhaps Patrick's cathode follower plus Ge diode?
>
> Maybe but that opens more than one can of worms. Linearity of the
> demod, filtering, slew distortion, phase shift... etc.
>
> That strikes me as way more work than is due a 'simple' transmitter.
It strikes me as practically no work at all with "sims" where it is easy to
build a perfectly linear envelope demodulator, as I have done when running
"sims".
John Byrns
flipper <fli...@fish.net> wrote:
> On Sun, 21 Aug 2011 18:07:53 -0500, John Byrns <byr...@sbcglobal.net>
> In this circuit that sounds a little like a distinction without much
> of a difference.
>
> I mean, with a triode it seems a bit esoteric to debate you're
> 'controlling' one vs the other.
That's based on the assumption that the oscillator looks like a resistor and
further that the oscillator output, "Vrms" is proportional to voltage and current
> >> The way I see it the modulator tube is setting current which, of
> >> course, means voltage too if we presume the osc is a (relatively)
> >> resistive load.
> >>
> >> I'm guessing you say we're setting plate voltage and current follows.
> >
> >Yes, if we are talking triode modulators I would say "we're setting plate
> >voltage and current follows", if we were talking pentode modulators I would
> >say
> >that we are setting the current, however either way around works.
>
> Sounds like you mean if the current source isn't 'good enough' of a
> current source then it becomes setting plate voltage but the plate
> voltage changes with the mod in either case.
>
> I think we understand the operation but are just using different words
> for the same thing. I mean, except for a triode not being a 'well
> regulated' current source (which I noted in my comment about it being
> 'closer to true' if using a pentode) the topology is what you
> described for "cathode modulation."
No, I guess I didn't clearly make my point, there is a significant difference
between "cathode modulation" and the series modulation" scheme used in the "Twin
Triode" AM transmitter. In both cases the modulator is in the cathode circuit,
however the reference point for the RF drive and bias to the oscillator grid is
very different for the two modulation schemes. With cathode modulation the RF
drive and bias for the oscillator grid is referenced to ground, while with
"series modulation" the RF drive and bias for the oscillator grid is referenced
to the oscillator cathode, it makes quite a big difference.
> But a triode isn't a 'well regulated' voltage source either.
>
> Got a 'name' for using a voltage controlled resistor?
Not offhand.
> Well, that's the thing. The term 'plate power' is used all over the
> place but seems to be one of those things everyone is just supposed to
> automatically 'know' because I never saw it defined before and
> presumed it was RF power generated.
There are at least three powers of interest;
DC plate power, used in the FCC Part 15 rules
RF Power Generated, broadcast transmitters are rated this way.
Power dissipated by the plate of the final tube, important in choosing the
correct tube.
> >> >> > Where are you
> >> >> >measuring "Vrms"?
> >> >>
> >> >> The osc plate.
> >> >>
> >> >> > What values of "average cathode current" and "Vrms" did you
> >> >> >obtain for the 12AU7 and 12AT7?
> >> >>
> >> >> The 6pF/100k 12AU7 version is, at 0 mod, approx 20Vrms on the osc
> >> >> plate with 3.9mA through Rk. The 12AT7 variant is about 14.8Vrms and
> >> >> 2.6mA.
> >> >>
> >> >> Modulated there's 'more' on the peaks and 'less' on the troughs but it
> >> >> averages the same. Which is a good thing because it should ;)
> >> >
> >> >I meant to ask, but forgot, what is the oscillator cathode voltage for
> >> >the
> >> >12AU7
> >> >and 12AT7 under the same conditions as you measured Vrms and the current
> >> >through
> >> >Rk?
> >>
> >> Well, depends on how one defines "same conditions."
> >
> >By "same conditions" I meant that Vrms, Icathode, and Vcathode for a given
> >tube
> >were all measured under the same conditions.
>
> Hehe. That would be a good example for "circular definition." "Same
> conditions" means under the "same conditions."
>
> Well, B+ is the same.
>
> >> I'm setting them
> >> for max mod before clipping so Rk and current is different. Second,
> >> since they're 'pairs', the presented oscillator load is not
> >> necessarily the same.
> >>
> >> The present 12AU7 osc cathode is at 143Vav with 3.75mA and the 12AT7
> >> is at 140Vav with 2.16mA
> >
> >Those voltages are somewhat higher than I expected, and the currents are
> >different than you stated above, so something has changed.
>
> What stated where above?
>
> 160-143=17Vdc times 3.75mA is the 64mW given.
>
> Things do change, though, as I tweak for frequency response, or
> whatever. For example, increase the osc grid resistor back to 100k and
> the numbers become 130V cathode and 3.85mA for 116mW.
>
> Then increase the coupling cap back to 100pF, as in the original, and
> it becomes 136V cathode with 4.05mA for 97mW.
>
> Put the 470 Ohm RK and 10nF RF bypass back in, like the original, and
> it's 118V cathode with 7.06mA for 297mW. Unfortunately, it begins
> clipping at around 43% mod and the frequency response is dismal with a
> 20dB drop from 150Hz (peak) to 10kHz.
>
> Everything changes everything else and I'm constantly tweaking so ask
> me tomorrow and, odds are, something will be a little different.
Right, that's precisely why I asked for the modulator plate voltage under the
same conditions as your first reported measurements of "Vrms" and average Ik.
> >> >Are these numbers taken from your breadboard or your "sims"?
> >>
> >> Sims
> >
> >You said in your original post to this thread that you had bread boarded the
> >circuit, have you compared Vrms, Icathode, and Vcathode on the bread board
> >with
> >the same on the sims?
>
> It's 'close' but, since power changes, B+ varies as I tweak so
> everything else is a bit different too.
>
> >One problem with this transmitter circuit is that there is no way to adjust
> >the
> >oscillator current, independently of the modulator plate voltage, like the
> >loading control does in a "real" transmitter does.
>
> Sounds like the parallel CCS I mentioned.
They don't do the same thing, more degrees of freedom are needed to set the DC
plate power at 100 mW and optimize the modulation at the same time.
> ><snip>
> >
> >> Btw, I discovered why I was getting confusing distortion numbers but
> >> I'll go into that in another message.
> >
> >I will patiently await your explanation of the distortion.
>
> Posted.
Noted.
> ><snip>
> >
> >> >The way the big boys dealt with this problem in real broadcast
> >> >transmitters
> >> >was
> >> >to operate the lower tube in class D rather than class A. This has two
> >> >advantages, first it eliminates the effects of tube nonlinearity, and
> >> >second
> >> >it
> >> >greatly increases the efficiency since the modulator tube is either ON or
> >> >OFF,
> >> >none of this analog stuff. A low pass audio filter is required between
> >> >the
> >> >plate of the modulator and the cathode of the RF tube to remove the
> >> >switching
> >> >frequency however.
> >>
> >> Yeah. A bit beyond a 'uhf converter box' special, though ')
> >
> >A dedicated constructor could probably pack it all into the "uhf converter
> >box".
> >I figure it could be done with 6 dual triodes, or 4 compactrons. Fitting in
> >a
> >suitable power supply might be the major problem.
>
> Like a "funny car" with a shell that looks like a Ford Taurus but
> everything inside is custom. Actually, so is the shell.
>
> You knew what I meant ;) The point is to use the UHF converter close
> to 'as is', except the circuit of course, in order to scavenge the
> parts.
>
> I also thought about scavenging the tubes but out of 4 converters not
> a one was good so that's not something to count on.
>
> ><snip>
> >
> >> >> I've been trying to work on the class A oscillator thing but spice is
> >> >> producing insanity. I get things like it claiming 2kV swings and
> >> >> negative screen voltages so there's clearly something wrong with the
> >> >> models when used that way.
> >> >
> >> >What is the "class A oscillator thing", I don't remember your mentioning
> >> >this, I
> >> >suppose a few more brain cells than I thought bit the dust?
> >>
> >> Well, the typical cathode-grid oscillator operates semi class C, grid
> >> drive setting the bias and all that. The RF is dirty unless filtered,
> >> like with a second tuned tank. That's one reason I went with the plate
> >> oscillator to begin with.
> >
> >Now you got me on the name thing, what is a "cathode-grid oscillator"
>
> Typical AA5 LO
>
> > and a
> >"plate oscillator"?
>
> Typical 'wireless phonograph" osc (and the AES kit copy, and the twin
> triode, and the 'one tube wonder').
The "typical AA5 LO", and the "typical wireless phonograph osc", and IIRC the
"one tube wonder", and the "Twin Triode AM transmitter" all use the same
oscillator circuit, namely a Hartely oscillator. This can be easily seen by
drawing the AC equivalent circuits for each, removing the power supplies, and
deleting the extra grids.
John Byrns
flipper <fli...@fish.net> wrote:
> On Mon, 22 Aug 2011 18:14:55 -0500, John Byrns <byr...@sbcglobal.net>
> The RF grunge is already in the audio band. That's why spice shows
> 'signal' and all harmonics with no input signal applied.
That's curious, do you know how the RF grunge, which starts as an RF sawtooth,
get translated down into the audio band?
> I said two things, a "choke to the modulator plate"..... and a "modest
> (100pF) Rk filter."
>
> The choke is between the modulator plate and the RF bypass cap and the
> modest 100pF filter in across the modulator Rk.
OK, that's what I suspected, at least as far as the choke goes, and I was
beginning to suspect that the "(100pF) Rk filter" was on the modulator cathode.
Since you aren't building an analog modulator, not a class D modulator, did you
try adding a capacitor from the modulator plate to ground? By applying a little
filter design theory you can probably get flatter response to 10 kHz, and
greater attenuation at the carrier frequency, by using two capacitors and the
inductor, appropriately chosen, between the modulator plate and oscillator
cathode.
> I was trying to get a clean signal there.
>
> >> >> and NFB from modulator Rk to the input 'stereo mixer' I can
> >> >> get the sims to almost .1%, measured at the modulator Rk, limited by
> >> >> the preamp triode (which isn't terrific at 160V).
> >> >
> >> >Did you try feedback from the modulator plate to the preamp cathode with
> >> >a
> >> >dominant pole in the feedback path to knock down the "RF" grunge?
> >>
> >> That was the first thing I tried (sim) but there's so much RF there
> >> it's almost impossible to filter.
> >
> >You only need to knock it down to a level where it doesn't cause trouble.
>
> A heck of a lot easier said than done.
>
> > Also
> >the audio signal at the modulator plate is presumably much larger than at
> >the
> >cathode so the flat loss of the feedback network, before any filtering, is
> >proportionately greater, greatly attenuating the "sawtooth".
>
> You lost me. What is "flat loss?"
"Flat loss" is essentially the loss of the network without the equalizing or
filtering components in place.
> The difference isn't as much as you might think, like 2.5, or so, to
> one, and the RF hash is the same proportion.
Exactly, that is what I expected, so what do you gain by taking the feedback
from the modulator cathode?
> >> I don't think it matters, though, because both approaches presume the
> >> osc is a resistive load. I.E. if we 'control' the voltage then it
> >> better be resistive or current isn't linear. But if current visa vie
> >> voltage is linear then we can accomplish the same thing by controlling
> >> current.
> >
> >I'm not sure what you are trying to say here? Are you simply saying that it
> >doesn't matter where the feedback is taken from, under the assumption that
> >the
> >oscillator looks like a resistor to the modulator?
> >
> >That raises two questions, one how much like a resistor does the oscillator
> >actually look, and two, how linear is "Vrms" vs. either voltage or current?
>
> Yes, that's the question alright.
>
> However, I may have been mistaken. The RF negative tips follow the
> cathode. Specifically the positive 'peaks' of the sawtooth.
You lost me, I must have missed your mistake, whatever it was?
> Or, to visualize, if you look at the lovely AM modulated RF waveform
> and then put the cathode on a second trace you see cathode voltage
> snuggled right up against the bottom of the AM envelope following it.
> Actually vice versa but, as the saying goes, you get the picture.
Yes, that is more or less as I visualized it, although we may not be visualizing
quite the same thing. When you talk about the "AM envelope" what are you really
talking about, where is the "scope" connected, and what do you mean by "snuggled
right up", aren't the two traces separated by the saturation voltage of the tube
at the grid drive level applied?
> This leads one to think, whether current is linear not, we can
> certainly control NFB by the cathode voltage, since the AM envelope
> faithfully follows it.
It's not obvious to me that that follows, you may be making a few assumptions
that may or may not be justified.
> However, in the immortal words of Capt Kirk, we have another problem.
> The AM envelope is following the 'peaks' of the sawtooth but the
> sawtooth amplitude changes with modulation depth, because of the
> power. I.E. it's 'thickness' (RMS magnitude) is more on the modulation
> peaks than at the troughs.
You are beginning to sound like Patrick with this talk of sawtooths!
There are few details here that need justification, first I need to understand
what you mean by the "AM envelope" and how you are measuring it vs. the
oscillator cathode voltage? At what point in the circuit are you looking at the
"AM envelope"?
> But a *filter* averages it. So the filtered signal is closer to the
> actual AM envelope in the power troughs and further from the envelope
> at the power peaks.
What is the "it" that the filter averages? Are you sure there is really a
problem here?
> In short, 2'nd harmonic distortion.
I don't see that, the sawtooth peaks and valleys should average out to the value
we expect. The modulator tube itself will introduce significant 2'nd harmonic
though.
> Actually need a 'peak detector' for the sawtooth.
I believe we want the average of the sawtooth, although I may be wrong, have you
looked carefully at the oscillator cathode current waveform over its conduction
interval?
> Of course, we can put a 'big RF bypass cap' to absorb the oscillator
> current spikes creating that sawtooth as long as we don't mind
> bandwidth being only a kHz or two.
>
> This behavior is apparently because the oscillator is being heavily
> overdriven, saturating the triode, because there's a significant
> separation between plate and cathode when I limit positive grid drive.
> Not only that but a phase shift appears with the cathode leading the
> anode.
We aren't trying to build a linear amplifier here, we want a good amount of
drive voltage on the grid to achieve high efficiency, or at least we would in a
real transmitter.
Adding a third harmonic trap in the oscillator cathode circuit might help
minimize the sawtooth current effect.
> What seemed like a relatively simple thing, just pop in a little
> feedback, is anything but.
I don't know, the feedback seems like a separate issue, from the sawtooth, to
me. I still would like to understand how the RF sawtooth is being translated
down into audio frequency noise?
> >> >Wouldn't it be better to measure the actual envelope distortion of the
> >> >entire
> >> >transmitter system, rather than the audio at the modulator cathode, using
> >> >an
> >> >envelope detector circuit, perhaps Patrick's cathode follower plus Ge
> >> >diode?
> >>
> >> Maybe but that opens more than one can of worms. Linearity of the
> >> demod, filtering, slew distortion, phase shift... etc.
> >>
> >> That strikes me as way more work than is due a 'simple' transmitter.
> >
> >It strikes me as practically no work at all with "sims" where it is easy to
> >build a perfectly linear envelope demodulator, as I have done when running
> >"sims".
>
> Did you try using it for feedback?
No, I used it just for measuring, in the "sims" the demodulator itself was
implemented with ideal components, followed by a filter that did what I wanted,
flat audio passband while attenuating the RF by the amount I wanted.
In the tube era the big boys did use feedback rectifiers with transmitters
employing low level modulation, or screen grid modulation.
Patrick Turner
> I stuck a picture of the trace on the web page
>
> http://flipperhome.dyndns.org/Twin%20Triode%20AM%20Transmitter.htm
>
> > The modulator tube itself will introduce significant 2'nd harmonic
> >though.
>
> >> Actually need a 'peak detector' for the sawtooth.
>
> >I believe we want the average of the sawtooth,
>
>
> > although I may be wrong, have you
> >looked carefully at the oscillator cathode current waveform over its conduction
> >interval?
>
The discussion about the twin triode AM generator continues between JB
and Flipper.
I had a look at the oscilloscope wave you have shown at
http://flipperhome.dyndns.org/Twin%20Triode%20AM%20Transmitter.htm
I don't see much 2H.
on my PC, the envelope has a majority green portion which if
considered alone looks like an AM RF wave with a sine wave audio tone,
with about 75% modulation %.
There's an added yellow band tacked on the bottom of the green
envelope. Is the yellow band the detected AF with RF ripple included?
If so, there isn't a huge amount of 2H.
The positive peaks of the yellow band look similar to negative peaks
of yellow band. What I think you need to do is filter the ripple away
and then compare recovered AF with green envelope shape.
Ideally, you should be able to set the AM mod to 25%, then set the
recovered audio level on a dual trace CRO to line up with envelope
shape, and when you increase the mod % the recovered AF should exactly
follow the envelope shape to 100% mod.
The CRO picture looks very much like what a simulator program might
generate. I wonder perhaps your wave form is not really real? I don't
see a picture of an oscilloscope anywhere, you know, one with a screen
and graticules etc.
Perhaps you ought to have a look at how other people test their AM
waves. Try having a look at
http://www.cleanrf.com/applications.html
There's a section there about trapezoid CRO shapes with what looks
like a green triangle on a CRO screen. That tells you a lot.
Anyway, who am I to tell anyone to read more about it. I must learn to
STFU.
Patrick Turner.
John Byrns
flipper <fli...@fish.net> wrote:
> On Mon, 22 Aug 2011 23:06:18 -0500, John Byrns <byr...@sbcglobal.net>
> wrote:
>
> >In article <0ou557dsta8h2bpiu...@4ax.com>,
> > flipper <fli...@fish.net> wrote:
<Snip>
> >> The RF grunge is already in the audio band. That's why spice shows
> >> 'signal' and all harmonics with no input signal applied.
> >
> >That's curious, do you know how the RF grunge, which starts as an RF
> >sawtooth,
> >get translated down into the audio band?
>
> I don't think any 'translation' is needed. A sawtooth is full of
> frequencies.
A sawtooth is full of frequencies that are all harmonics of the fundamental
frequency, it doesn't contain any frequencies that are lower than the
fundamental frequency, so the sawtooth which in this case is the RF carrier
frequency should not contain any audio frequency components. Of course the
sawtooth rides on the audio output of the modulator tube when there is a signal
input.
> Unless spice is lying to me.
That may be worth considering, if you are seeing audio frequency components with
no audio input signal applied!
<Snip>
> >> But a *filter* averages it. So the filtered signal is closer to the
> >> actual AM envelope in the power troughs and further from the envelope
> >> at the power peaks.
> >
> >What is the "it" that the filter averages?
>
> The sawtooth
>
> > Are you sure there is really a
> >problem here?
>
> Yes.
>
> >> In short, 2'nd harmonic distortion.
> >
> >I don't see that, the sawtooth peaks and valleys should average out to the
> >value
> >we expect.
>
> No, because it's thickness changes but the AM envelope is mated to the
> peaks, not the average.
>
> I stuck a picture of the trace on the web page
>
> http://flipperhome.dyndns.org/Twin%20Triode%20AM%20Transmitter.htm
This picture is interesting from several perspectives
First it does not appear to show what you say it is supposed to show, I can see
neither the "sawtooth waves" that you describe, nor the snuggling or mating that
you talk of. All I can see is some squared off quantizing that I assume is an
artifact of your simulator. I think you need to zoom in on only a few cycles of
RF for a better view, at both the modulation peaks and the valleys. It would
also be useful to show the oscillator cathode current waveform.
I also notice two notations near the oscillator tube on the schematic, "48 mW
RF" and "80 mW PP". I assume that the "80 mW PP" represents the DC Plate Input
Power calculated as I described. I further assume that "48 mW RF" is supposed
to represent the gross RF power generated by the tube, but I can't for the life
of me figure how you came up with this number?
Further you don't specify the value of the capacitor and two resistors that make
up the "Wire Ant." which may be useful in calculating the actual net RF power
output. Details on the tapped inductor in the plate circuit, the inductance of
each side and the mutual inductance between them, would also be useful in
calculating the gross RF power.
Related to that, I think you said that the tap is asymmetrically placed on the
coil, it would be interesting to see a similar "Sim" with the plate and grid
ends of the coil interchanged.
Secondly this picture drives home the point, which you have already described
for me in numbers, that this circuit is not operating as I have always assumed
it should. Specifically the oscillator cathode is only operating something on
the order of 20 volts below the B+ voltage under carrier conditions, I would
think that the oscillator cathode should be roughly triple that voltage below
the B+ voltage under carrier conditions, or something on the order of 100 volts
on the cathode rather than 140 volts. I believe the transmitter would work
better at this operating point, however the "resistance" of the oscillator would
need to be increased, which is that additional degree of freedom that I was
speaking of. I will have to give some thought to the best way of achieving this
additional degree of freedom before I purchase a 6AV11 and warm up my soldering
iron to build the "Triple Triode" AM Transmitter.
It would also be interesting to see the modulator grid to cathode voltage on the
picture you posted, the one that shows several complete cycles of modulation, so
one could know how close the modulator is getting to the 0 grid line.
<Snip>
Patrick Turner
> On Tue, 23 Aug 2011 07:26:23 -0700 (PDT), Patrick Turner
>
> See the green square up in the shown fraction of the schematic sitting
> on the oscillator plate? That's the green trace in the plot area. See
> the yellow square up there sitting on the oscillator cathode? That's
> the yellow trace in the plot.
>
> To the left of the plot is the circuit label in addition to color. "A"
> (green) is V2B_1. I.E. tube V2B pin 1. "B" (yellow) is V2B_3. I.E.
> tube V2B (same tube) pin 3. V2B is the oscillator and pin 1 is the
> plate, pin 3 the cathode.
>
> Byrns asked me what I 'meant' by the cathode signal being snuggled up
> against the AM envelope so I said I posted a picture of it and, well,
> there ya go. That's what I meant.
>
> No, the yellow band is not recovered AF with RF ripple included
> although it makes sense it would look like that as it's the audio
> being applied to the oscillator cathode thereby modulating the RF.
> AI.E Rather than this 'AF' being the 'result' of demodulating the AM
> envelope the AM envelope is the result of this AF modulating it.
>
> The yellow trace 'ripple' is the sawtooth Byrns and I have been
> discussing and you see it is not a constant magnitude. The AM envelope
> follows the 'peaks' of the ripple so it seems to me an 'average'
> ripple value has an added 2'd harmonic relative to the AM envelope.
I'm not so sure is has added much 2H.
>
> >Ideally, you should be able to set the AM mod to 25%, then set the
> >recovered audio level on a dual trace CRO to line up with envelope
> >shape, and when you increase the mod % the recovered AF should exactly
> >follow the envelope shape to 100% mod.
>
> >The CRO picture looks very much like what a simulator program might
> >generate. I wonder perhaps your wave form is not really real? I don't
> >see a picture of an oscilloscope anywhere, you know, one with a screen
> >and graticules etc.
>
> It looks like a simulator program because it is and, as a general
> rule, it's probably safer to presume a thing is what it looks like
> than to imagine it's something else.
>
> I never said it was a CRO picture and all the discussion has been
> "spice tells me <insert description>"
>
> I just popped that picture on the web page because it was a quick and
> easy way to show Byrns.
OK, so we are discussing the workings of a circuit by means of
illustrated imagination, rather seeing real observation of an actual
working circuit.
I'm using grid mod in my RF gene. There's a buffered oscillator
feeding a 6BX6 RF amp with tunable anode tank OPT. The RF enters via
about 100pF from osc buffer, then the AF enters via a grid bias R, and
so RF and AF are added, applied to grid, The cathode has 3k3 Rk,
bypassed only at RF only, so Ia is fairly linearly changed by AF
because of the current FB, and of course the gm is also altered, and
RF gets modulated. I tried a lot of pentodes and all gave a lot of 2H
above 80% mod, so then added a detector to produce AF from the output
AM. then applied this to a diff amp with AF input so the AF sent to
the pentode grid contains an error signal which helps reduce envelope
THD about 5 times,
and I get about 95% mod that looks a lot better.
Ordinary triangular wave AF is a good test and the straightness of the
wave mod should remain straight up to high mod%.
I don't recall looking at the signal at my pentode cathode, but it
would have a large AF content with varying RF ripple, like your
simulator shows.
> >Perhaps you ought to have a look at how other people test their AM
> >waves. Try having a look at
> >http://www.cleanrf.com/applications.html
>
> >There's a section there about trapezoid CRO shapes with what looks
> >like a green triangle on a CRO screen. That tells you a lot.
>
> Thank you but what I'm looking at would be difficult to see on a 'real
> scope' because it rides on 160Vdc.
Why? if you have a floating resistor or coil or cap someplace, one may
use a dual trace CRO and go to each side of the item and you can see
the difference between the two signals which are each between the itrm
ends and 0V.
> So, if DC coupling, you have a
> comparatively small signal shrunk down so that 160V 'fits' on the 8
> vertical division screen and it would difficult to align both traces
> to discern just 'how close' or nor they were. AC couple, so you could
> enlarge just the AC component, and you lose what the DC relationship
> is and, again, you couldn't tell if, as I described, the AM envelope
> 'rode the peaks' of the sawtooth ripple.
Without looking over your shoulder and seeing exactly what you're
measuring, I really should STFU.
>
> With a 'real CRO' you'd deduce that is "probably the case" because "it
> makes sense" and "near as I can tell" the trace looks like it.
>
> >Anyway, who am I to tell anyone to read more about it. I must learn to
> >STFU.
>
> No one minds good natured assistance, Patrick. It's the wise cracks
> people take offense to.
Well, I didn't really want to spend too much time on this project of
yours when I know I'd probably go about the whole process in a
different and much more complicated manner. I'm not game to say its
more sophisticated because I'm still using tubes when anyone else who
just went for performance would use chips. I did have a Topward
function generator fulla chips with every known possible analog wave.
Extremely complex inside, and I repaired it upteen times. Finally an
accident with 240V mains getting in via its output fried multiple
things internally and into the bin it went.
I dunno exactly how many tubes I have in my RF gene home brew tester,
maybe 5 or 6, and all simulation is in my head, and if I can't imagine
how something might work myself then its no use me trying it. Usually,
implemented imaginings have frightful defects as things one can't
forsee spoil the outcome, but then that forces one to think more, do
more, learn more, and I end up with an acceptable result that looks
well on a real CRO.
I also have 59 other projects my customers want me to complete by
last Friday.
Good luck.
Patrick Turner.
John Byrns
flipper <fli...@fish.net> wrote:
> On Tue, 23 Aug 2011 18:59:39 -0500, John Byrns <byr...@sbcglobal.net>
> wrote:
>
> >In article <q4c6579ipm34nfst6...@4ax.com>,
> > flipper <fli...@fish.net> wrote:
> >
> <snip>
>
> >>
> >> I stuck a picture of the trace on the web page
> >>
> >> http://flipperhome.dyndns.org/Twin%20Triode%20AM%20Transmitter.htm
> >
> >This picture is interesting from several perspectives
> >
> >First it does not appear to show what you say it is supposed to show, I can
> >see
> >neither the "sawtooth waves" that you describe, nor the snuggling or mating
> >that
> >you talk of.
>
> Oh for Christ's sake. Tell me again you don't see the cathode waveform
> snuggled up against the AM envelope.
I don't see the cathode waveform snuggled up against the AM envelope! You said
something to the effect that cathode waveform was snuggled up against the tips
of the sawtooth riding on the audio at the modulator plate. Not only couldn't
see any sawtooth, but there was a fair amount of yellow distance between the osc
cathode and the mod plate!
> Doesn't mater. Circuitmaker is not correct.
>
> > All I can see is some squared off quantizing that I assume is an
> >artifact of your simulator.
>
> What did you do? Download it and then try to blow it up 10x? That's
> the only way I can imagine you finding any 'squared off quantizing'.
Certainly, when I didn't see any sawtooth, and saw that there was considerable
distance between the two waveforms I naturally enlarged it to see if I could see
the sawtooth you said was there, as I said all I could see was the quantizing
effect.
> I put up another one zoomed in on the RF.
You must have taken it down before 7:30 AM CST this morning as it wasn't there
when I went looking for it.
> Both are now worthless (see below) but I'll leave them up till I redo
> things.
They aren't there.
> > I think you need to zoom in on only a few cycles of
> >RF for a better view, at both the modulation peaks and the valleys. It
> >would
> >also be useful to show the oscillator cathode current waveform.
>
> I can't get at osc cathode current unless I make some circuit mods to
> provide a measurement point. It can't probe current in tube pins.
OK, that sounds reasonable.
> However, there's something wrong with the model, see below.
>
> >I also notice two notations near the oscillator tube on the schematic, "48
> >mW
> >RF" and "80 mW PP". I assume that the "80 mW PP" represents the DC Plate
> >Input
> >Power calculated as I described.
>
> Yes.
>
> > I further assume that "48 mW RF" is supposed
> >to represent the gross RF power generated by the tube, but I can't for the
> >life
> >of me figure how you came up with this number?
>
> Exactly the way I told you it was calculated. RF Vrms times the
> average current through the modulator.
I originally suspected this calculation wasn't quite correct, and was going to
try working out the exact formula, but I abandoned that when I misread your
answer to one of my questions so that it appeared there was not just a minor
error in your RF power number, but what was a really gross error. I just
realized that I had misread one of your answers causing me to believe there was
a gross error. Now that I realize my reading error, I may have to go back to
trying to verify your formula mathematically. I believe the correct formula
must incorporate the oscillator conduction angle among other factors.
> >Further you don't specify the value of the capacitor and two resistors that
> >make
> >up the "Wire Ant." which may be useful in calculating the actual net RF
> >power
> >output.
>
> For radiated power, yes, but I don't consider it realistic enough to
> count on. It's just there as a relative comparison for me.
Radiated power would require more than simply the impedance parameters of the
antenna.
> It's just 30pF into .177 ohm. The shunt resistor is a dummy 100Meg so
> spice won't bitch about 'singular matrix' due to a cap feeding a cap.
Do you really mean 0.177 ohms?! Was the antenna model connected for your "sims"?
> > Details on the tapped inductor in the plate circuit, the inductance of
> >each side
>
> 5uH and 110uH
Where did these numbers come from, especially the 5 uH number, are they measured
values?
> > and the mutual inductance between them,
>
> Who the heck knows? It's just a 'typical' local osc type similar to
> the Miller OSC-70.
"Who the heck knows" may be a valid answer for the actual physical coil, but I
assume you had to put in some number for the "sims", or are the two coils not
coupled at all in your "sims"?
> This may be part of the modeling problem, though. See below.
>
> > would also be useful in
> >calculating the gross RF power.
> >
> >Related to that, I think you said that the tap is asymmetrically placed on
> >the
> >coil, it would be interesting to see a similar "Sim" with the plate and grid
> >ends of the coil interchanged.
>
> It doesn't oscillate in circuitmaker.
OK, that's a useful answer. I wonder why? The "Zenith" circuit oscillates with
the OSC-70 coil connected either way around.
> >Secondly this picture drives home the point, which you have already
> >described
> >for me in numbers, that this circuit is not operating as I have always
> >assumed
> >it should. Specifically the oscillator cathode is only operating something
> >on
> >the order of 20 volts below the B+ voltage under carrier conditions, I would
> >think that the oscillator cathode should be roughly triple that voltage
> >below
> >the B+ voltage under carrier conditions, or something on the order of 100
> >volts
> >on the cathode rather than 140 volts.
>
> That's not too far off from what the actual circuit is doing and this
> is the beginning of the "see below" stuff.
>
> The 'real life' numbers are
>
> B+ is about 168V and the osc cath is sitting at 111V.
>
> The modulator Rk has 3.96V on it, meaning idle current is 2.8mA vs
> 4.3mA in the sim.
>
> Osc plate RF is 1.3Vrms and 'antenna' volts (big end of coil) is
> 8.8Vrms. Unless my 10x probe was seriously loading it down both of
> those are way less than circuitmaker is saying.
That's interesting/reassuring.
> > I believe the transmitter would work
> >better at this operating point, however the "resistance" of the oscillator
> >would
> >need to be increased, which is that additional degree of freedom that I was
> >speaking of. I will have to give some thought to the best way of achieving
> >this
> >additional degree of freedom before I purchase a 6AV11 and warm up my
> >soldering
> >iron to build the "Triple Triode" AM Transmitter.
>
> Well, you can plug it back in because the sim is not accurate.
I'm not sure what it is that you say I can plug back in?
I imagine that we will hear from Patrick now that you have said that "the sim is
not accurate"!
> I'm not sure what's causing it but this circuit is so sensitive to
> just about everything that it may be due to many circuitmaker
> components being 'ideal'. Like, capacitors have no resistive losses
> and I have always had trouble with magnetic devices, especially
> transformers.
>
> Tube models can also be problematic as, for example, some authors
> don't bother modeling grid current. That's not the case here, the
> 12AU7 model does do grid current, but who knows what other
> approximations were made. In fact, I have two 12AU7 models and,
> surprise, they don't give the same results. The other one is even
> worse.
>
> I also didn't include the few ohms of coil resistance. Putting that in
> knocked down the RF amplitude but didn't affect osc cathode voltage,
> so it wasn't 'that problem'.
>
> While none of those 'compromises' may be 'serious' all by themselves
> they apparently add up to some significant miscalculations in this
> circuit. It's good enough to show it 'works' but not good enough for
> the kind of detail we're looking for.
My guess would be that it is mainly the coil and pobably the tube model. I
suspect that the tube model doesn't have to be very accurate, but it must not
include dumb mistakes that may get by for audiophile tube amps, but will be a
disaster for large signal RF "sims".
John Byrns
flipper <fli...@fish.net> wrote:
> On Wed, 24 Aug 2011 15:40:26 -0500, John Byrns <byr...@sbcglobal.net>
> wrote:
>
> >In article <0cj857ltbq31bngi9...@4ax.com>,
> > flipper <fli...@fish.net> wrote:
> >
> >> On Tue, 23 Aug 2011 18:59:39 -0500, John Byrns <byr...@sbcglobal.net>
> >> wrote:
> >>
> >> >In article <q4c6579ipm34nfst6...@4ax.com>,
> >> > flipper <fli...@fish.net> wrote:
> >> >
> >> <snip>
> >>
> >> >>
> >> >> I stuck a picture of the trace on the web page
> >> >>
> >> >> http://flipperhome.dyndns.org/Twin%20Triode%20AM%20Transmitter.htm
> >> >
> >> >This picture is interesting from several perspectives
> >> >
> >> >First it does not appear to show what you say it is supposed to show, I
> >> >can
> >> >see
> >> >neither the "sawtooth waves" that you describe, nor the snuggling or
> >> >mating
> >> >that
> >> >you talk of.
> >>
> >> Oh for Christ's sake. Tell me again you don't see the cathode waveform
> >> snuggled up against the AM envelope.
> >
> >I don't see the cathode waveform snuggled up against the AM envelope!
>
> Then you need better glasses, or something, because there is not so
> much as 1 pixel of anything or any color between the green (plate) and
> yellow (cathode) waveforms.
I'm getting new glasses tomorrow, hopefully I'll be able to see better then,
although I am beginning to suspect that I am misinterpreting the picture.
Clearly the green area represents the modulated carrier on the oscillator plate,
the individual cycles of which can't be resolved. I had assumed that the
modulator plate signal was represented by the boundary between the black and the
yellow areas, but I am beginning to realize that the yellow area itself is the
modulator plate, a.k.a. oscillator cathode, signal with the sawtoofth waves to
fine to be resolved. It is certainly true that the in the picture the green and
yellow areas are "snuggled" right up to each other, but what do they really
represent? It would help to show an expanded "sim" covering maybe 10 or so
cycles of RF. I have my doubts if this picture is really showing what it seems
to, both because the simulator picture software may be messing things up, and
more importantly this picture makes me seriously wonder about what is going on
in the 12AU7 model and what it is really doing?! Do you have the "code" for the
12AU7 model?
> That is as blooming "snuggled up against" as on can get.
Yes assuming that you are talking about the yellow area being snuggled up
against the green area, the question is what that means and what the simulator
is actually doing.
<Snip>
> >Radiated power would require more than simply the impedance parameters of
> >the
> >antenna.
>
> Not since I can put a 'watts' probe on it.
With such a short antenna most of the "radiated power" is probably actually lost
in the ground system.
The "Watts probe" sounds interesting, does it measure both real and imaginary
power, and show which direction the power is flowing, resolving forward and
reflected power?
<Snip>
> >> > Details on the tapped inductor in the plate circuit, the inductance of
> >> >each side
> >>
> >> 5uH and 110uH
> >
> >Where did these numbers come from, especially the 5 uH number, are they
> >measured
> >values?
>
> Measured with the slug wherever it was. Somewhere around 1.1 MHz in
> circuit.
>
> >> > and the mutual inductance between them,
> >>
> >> Who the heck knows? It's just a 'typical' local osc type similar to
> >> the Miller OSC-70.
> >
> >"Who the heck knows" may be a valid answer for the actual physical coil, but
> >I
> >assume you had to put in some number for the "sims", or are the two coils
> >not
> >coupled at all in your "sims"?
>
> I used a 'transformer' model and, if 'there', it's whatever the
> circuitmaker model has.
I would assume that a transformer model has a coupling coefficient close to
unity? I suspect that the real coil is nowhere near as tightly coupled as a
transformer model.
<Snip>
> >> >Related to that, I think you said that the tap is asymmetrically placed
> >> >on
> >> >the
> >> >coil, it would be interesting to see a similar "Sim" with the plate and
> >> >grid
> >> >ends of the coil interchanged.
> >>
> >> It doesn't oscillate in circuitmaker.
> >
> >OK, that's a useful answer. I wonder why? The "Zenith" circuit oscillates
> >with
> >the OSC-70 coil connected either way around.
>
> Well, in my experience yes and no. Which doesn't make sense and I know
> it doesn't make sense but last time I tried I couldn't get the 'one
> tube wonder' to oscillate with it the 'right' way, short end on plate,
> even though I swear it did when I first breadboarded the thing.
>
> Must be 'something' not the same but it shows they don't oscillate 'no
> matter what' and we now know the models in circuitmaker are not
> 'perfect' in this kind of configuration.
>
> Actually, I want to revisit the 'one tube wonder' with it back like
> the zenith because I suspect the large plate swing contributed to some
> of the G3 alignment issues I ran into but that breadboard is gone now
> and I haven't had time to make another one, not to mention I only have
> 3 coils and all of them are 'in use' at the moment.
I would like to try the Twin Triode AM transmitter circuit, but am not thrilled
with collecting all the needed parts to start from scratch. I am toying with
deconstructing my "Zenith" phono oscillator and rebuilding it as a Twin Triode
circuit, replacing the 12SA7GT with a 12SN7GT dual triode. The power
transformer isn't rated to handle a preamp tube in addition to the 12SN7GT, but
I could use the preamp tube socket hole for a plug in balanced 600 ohm to grid
transformer and drive it with a CBS labs audimax and volumax processor pair.
John L Stewart
The "Watts probe" sounds interesting, does it measure both real and
imaginary
power, and show which direction the power is flowing, resolving forward
and
reflected power?
>>>>>>>For that John, you probably need to call on Agilent/HP, something
you may have tried once during your broadcast daze. Or perhaps Bird.
I've used both.
...................................................
I would like to try the Twin Triode AM transmitter circuit, but am not
thrilled
with collecting all the needed parts to start from scratch. I am toying
with
deconstructing my "Zenith" phono oscillator and rebuilding it as a Twin
Triode
circuit, replacing the 12SA7GT with a 12SN7GT dual triode. The power
transformer isn't rated to handle a preamp tube in addition to the
12SN7GT, but
I could use the preamp tube socket hole for a plug in balanced 600 ohm
to grid
transformer and drive it with a CBS labs audimax and volumax processor
pair
>>>>>>>>>>>>>> Works only if you can find your seldom used soldering
iron. Is it worth a bet you will never try that cct?
John L Stewart, who can still recognize the handle end of a
screwdriver!!
.........................................................
--
Regards,
John Byrns
Surf my web pages at, http://fmamradios.com/
+-------------------------------------------------------------------+
|Filename: 6550A Tungsol 4h.jpg |
|Download: http://www.audiobanter.com/attachment.php?attachmentid=241|
+-------------------------------------------------------------------+
--
John L Stewart
John Byrns
John L Stewart <John.L.Stew...@audiobanter.com> wrote:
> > The "Watts probe" sounds interesting, does it measure both real and
> > imaginary
> > power, and show which direction the power is flowing, resolving forward
> > and
> > reflected power?
>
> >>>>>>>For that John, you probably need to call on Agilent/HP, somethingyou
> >>>>>>>may have tried once during your broadcast daze. Or perhaps Bird.I've
> >>>>>>>used both.
During my ³broadcast daze² Bird was the word, during my post broadcast days we
used mostly HP stuff, Agilent didn't come along until very late in the game IIRC.
> ...................................................
>
> > I would like to try the Twin Triode AM transmitter circuit, but am not
> > thrilled with collecting all the needed parts to start from scratch.
> > I am toying with deconstructing my "Zenith" phono oscillator and rebuilding
> > it as a Twin Triode circuit, replacing the 12SA7GT with a 12SN7GT dual triode.
> > The power transformer isn't rated to handle a preamp tube in addition to the
> > 12SN7GT, but I could use the preamp tube socket hole for a plug in balanced
> > 600 ohm to grid transformer and drive it with a CBS labs audimax and volumax
> > processor pair
>
> >>>>>>>>>>>>>> Works only if you can find your seldom used solderingiron. Is
> >>>>>>>>>>>>>> it worth a bet you will never try that cct?
>
> John L Stewart, who can still recognize the handle end of ascrewdriver!!
Well it depends on whether or not I can bear to let go of the AES circuit and
reuse the chassis. I guess I will have to put which way to go out for an
advisory vote by the group. See this web page for information on my
implementation of the AES phono oscillator kit:
http://www.fmamradios.com/AES_osc.html
So everyone, is this thing worth keeping or should I repurpose it?
If I decide to repurpose the chassis, which seems likely at this point, the
chances are all most 100% that I will implement the series modulation circuit.
If I can't bear to repurpose the chassis the chances that I build another one
are probably only around 50%, and the project would take longer.
> John L Stewart, who can still recognize the handle end of a
> screwdriver!!
Was there any doubt that you could recognize the handle end of a screwdriver?
John Byrns
flipper <fli...@fish.net> wrote:
You will be happy to hear that I had my eyes examined for the new glasses
yesterday, it will be about a week before I actually have them on my face
though, so if you post anymore waveform pictures in the next week, they will
probably still confuse me.
I was just looking at the latest "twin triode" schematic on your web page here:
http://flipperhome.dyndns.org/Twin%20Triode%20AM%20Transmitter.htm
I notice a few changes, but I don't see the inductor that you put between the
modulator plate and the oscillator cathode, I also don't see the capacitor from
the modulator cathode to ground? I thought these components were necessary for
the negative feedback to work properly, and to eliminate RF grunge?
The narrative on your web page says that C1 affects the 10 kHz frequency
response. I know you mentioned this in one of your previous posts, so I will go
back and see exactly what you said about it, however I am curious weather you
have made any investigation of the effect of this capacitor, beyond the
frequency response issue?
I have long wondered if this capacitor might contribute to high frequency
distortion on heavy negative modulation, so I am curious if you have seen any
distortion problem related to this capacitor? I have an idea for a fix using
the AES (P-C70-OSC) to reduce the extent of this problem, I will comment on this
possible fix later.
> What would be the reason to scavenge it? Cost?
Cost is not an issue as I already have 98% of the parts required for the twin
triode version, 97%, if I should decide to use a triple triode compactron.
> It can't be availability because, while they may not be 'everywhere',
> you can still get everything you'd need. For example, you can get the
> osc coil at AES (P-C70-OSC) and, if you want a 'conventional' one,
> Allied has a power transformer (6K27VF).
>
> I obviously think a UHF converter box is a good deal since, if your
> timing is good, you can get one off Ebay for the same, or less,
> (including shipping) than just the Allied 6K27VF so the chassis,
> sockets (if you can use them), power switch (w/knob), and pretty
> enclosure are 'free'. It takes a bit more elbow grease, though,
> because you have to first strip it.
>
> The worst thing is paying to ship onesies, like the coil from AES and
> a tube or two from ABCvacTubes (which is why I usually buy some
> 'extras' or wait till I need more than just one) but that's offset by
> not needing to buy a chassis (UHF converter)
>
> Personally, I hate dismantling prior projects but that's just me.
>
I don't particularly care for the �Zenith� circuit and would dismantle it in a
minute if it weren't for the fact that I documented the project so perhaps I
should keep it intact as part of the documentation. It is probably history,
however I haven't made the final decision yet.
John Byrns
flipper <fli...@fish.net> wrote:
> On Fri, 26 Aug 2011 22:15:18 -0500, John Byrns <byr...@sbcglobal.net>
> wrote:
>
> >In article <aqkd5790nafq9evkd...@4ax.com>,
> > flipper <fli...@fish.net> wrote:
> >
> >You will be happy to hear that I had my eyes examined for the new glasses
> >yesterday, it will be about a week before I actually have them on my face
> >though, so if you post anymore waveform pictures in the next week, they will
> >probably still confuse me.
> >
> >I was just looking at the latest "twin triode" schematic on your web page
> >here:
>
> Ah. I was about to post something so good timing.
>
> >http://flipperhome.dyndns.org/Twin%20Triode%20AM%20Transmitter.htm
> >
> >I notice a few changes, but I don't see the inductor that you put between
> >the
> >modulator plate and the oscillator cathode, I also don't see the capacitor
> >from
> >the modulator cathode to ground? I thought these components were necessary
> >for
> >the negative feedback to work properly, and to eliminate RF grunge?
>
> That was based on circuimaker but it's not modeling the osc right so I
> decided to just go ahead and try some things. But, for starters, I
> don't have a 100mH choke so, what the heck, try it without one.
> Besides, the grunge didn't seem as large as circuimaker was
> predicting.
I wonder if that is an artifact of problems with the 12AU7 model?
> My first shot did, however, have a modulator Rk bypass but I'm not
> convinced it's needed or that it has a desirable effect. Jury is still
> out on that, though, because the dern thing just magically changed
> 'tonal quality' on me a couple of times (noticeably 'bright' as if too
> much treble, or lack of bass) and I can't figure out why. Maybe
> something is 'loose' on the breadboard but I'll have to get it to
> settle down before coming to a conclusion.
>
> I had three choices: 1. demod and feedback, but I decided that was too
> complex for this go around.
>
> The other two were either modulator plate or cathode feedback.
>
> The problem with plate feedback is the FB resistor loads the osc, like
> the 1 Meg we tacked on to prevent osc cutoff but much more,
> significantly limiting negative mod.
I didn't stop to think about what value the feedback resistor would end up
being, I think you are telling me that the feedback resistor would be
considerably less than 1 Meg, which sounds logical, so that it can't simply be
used to replace the existing 1 Meg resistor.
> That left modulator cathode feedback.
>
> I'm also back to thinking it's 6 of one or a half dozen of the other
> since both depend on the osc behaving like a resistive load, which is
> almost certainly not true. That limits just how effective the NFB can
> be but, as previously mentioned, I saw a visually obvious improvement
> in negative mod.
>
> >The narrative on your web page says that C1 affects the 10 kHz frequency
> >response. I know you mentioned this in one of your previous posts, so I
> >will go
> >back and see exactly what you said about it, however I am curious weather
> >you
> >have made any investigation of the effect of this capacitor, beyond the
> >frequency response issue?
> >
> >I have long wondered if this capacitor might contribute to high frequency
> >distortion on heavy negative modulation, so I am curious if you have seen
> >any
> >distortion problem related to this capacitor?
>
> Bingo.
>
> I ran into that while trying to assess 10kHz response and it seems
> almost as if what circuitmaker reports as frequency response is how
> much it will slew distort on deep negative mod because I get no
> amplitude effect when changing those values but, rather, changes in
> the waveform distortion.
>
> For example, since I was getting no amplitude improvement with 47k vs
> 100k I went back to 100k for more radiated power. That naturally led
> me to see if increasing C1 would also have no effect but after
> increasing both I saw an increase in the negative mod distortion. I
> did so I tried the RF bypass, C3, and got the same results.
>
> Btw, that distortion is there no matter what the values are, it's just
> worse with higher C1, R3, and C3 values.
>
> At any rate, the feedback circuit I used did significantly improve
> that but I have no means to measure distortion. All I can say is I
> think it 'sounds better' but that's dern tough to asses just listening
> to an AM radio and after a few dozen 'listening tests' my head just
> spins.
>
> > I have an idea for a fix using
> >the AES (P-C70-OSC) to reduce the extent of this problem, I will comment on
> >this
> >possible fix later.
>
> I'm interested to hear your idea because I'm convinced this is what I
> was seeing in the 'one tube wonder', but to a larger degree.
Unfortunately my idea probably wouldn't help the one tube wonder. In the "twin
triode" circuit I see two unwanted things influencing the oscillator grid bias,
it would be interesting to explore these with a good simulation model. The
first effect, which would be common to both circuits, is that the oscillator
grid bias circuit is sort of like the AM envelope detector that was recently
discussed in another recent thread. The average grid bias is effectively the
output of an envelope detector, so if the "detector" goes into slew rate
limiting while the modulation is approaching the negative modulation peak, then
the bias on the oscillator grid will be greater than it would otherwise be. It
is not immediately obvious what the precise effect of this would be, it may
cause some self compensating feedback, a good "sim" would be useful here. My
improvement idea would have no effect, as far as I can visualize, on this
"problem".
The second problem, which applies only to the "twin triode" circuit, is that the
audio signal from the modulator is effectively in series with the B+, as in all
plate modulated transmitters, and in this design that voltage, varying with
modulation also appears across C1. So the C1*R1 time constant must charge fast
enough not to disturb the oscillator bias. This bias shift at high audio
frequencies is what my idea for a fix is designed to remedy.
The idea is this, the big tapped winding of the oscillator coil is connected
only in the tuned grid circuit, part of the winding is not used for the drive
from the plate. Instead the second winding on the AES (P-C70-OSC) coil is
connected in the plate circuit to provide the necessary feedback. I'm not sure
if the second winding has a suitable number of turns, but I vaguely remember
that it is similar to the short end of the main winding and so should work in
this way. Connecting the oscillator coil in this manner removes the modulated
B+ from across C!, eliminating any possible bias shift problem from this source.
Unfortunately, I don't believe this fix is relevant to the "one tube wonder",
nor would it appear to be a fix for the equivalent part of the problem in the
"twin triode". The question relative to the "twin triode" is which problem is
dominant, which would determine whether the "fix" would do any good.
I found your mention of distortion caused by C3 interesting, as it hadn't
occurred to me, beyond its effect on frequency response and possible distortion
in the modulator caused by excess capacitive loading of the modulator. I assume
that traditional plate modulated broadcast transmitters don't suffer from this
charging time constant problem because the modulator used a push pull circuit
topology with a low source impedance which could drive C3 to where it needed to
be. The interesting question is whether the later PDM transmitters suffer from
the problem you describe, to any extent?
> >> What would be the reason to scavenge it? Cost?
> >
> >Cost is not an issue as I already have 98% of the parts required for the
> >twin
> >triode version, 97%, if I should decide to use a triple triode compactron.
>
> ok
>
> >> It can't be availability because, while they may not be 'everywhere',
> >> you can still get everything you'd need. For example, you can get the
> >> osc coil at AES (P-C70-OSC) and, if you want a 'conventional' one,
> >> Allied has a power transformer (6K27VF).
> >>
> >> I obviously think a UHF converter box is a good deal since, if your
> >> timing is good, you can get one off Ebay for the same, or less,
> >> (including shipping) than just the Allied 6K27VF so the chassis,
> >> sockets (if you can use them), power switch (w/knob), and pretty
> >> enclosure are 'free'. It takes a bit more elbow grease, though,
> >> because you have to first strip it.
> >>
> >> The worst thing is paying to ship onesies, like the coil from AES and
> >> a tube or two from ABCvacTubes (which is why I usually buy some
> >> 'extras' or wait till I need more than just one) but that's offset by
> >> not needing to buy a chassis (UHF converter)
> >>
> >> Personally, I hate dismantling prior projects but that's just me.
> >>
> >
> >I don't particularly care for the łZenith˛ circuit and would dismantle it in
> >a
> >minute if it weren't for the fact that I documented the project so perhaps I
> >should keep it intact as part of the documentation. It is probably history,
> >however I haven't made the final decision yet.
>
> Well, if you don't care for it then dismantle away ;) Maybe take a
> good picture set for the historical record.
Good point, I'm glad you suggested it; I should get some modern digital photos
of it before ripping it apart.
One of the major advantages of dismantling the "Zenith" circuit is that it
eliminates a large number of decisions, layout, parts choice, etc. that I would
otherwise have to agonize over, which drags out with me. This way I just unplug
the 12AS7GT, unsolder a few R's and C's, solder in some new and different R's
and C's, and finally plug in a 12SN7 from my stash, and I am on the air with the
new circuit. The agonizing decisions can be delayed until after the basic
circuit is built and proved out. There are two or three deferred decisions that
would ultimately have to be made. First, which one of three input transformer
options to choose will probably be simply a matter of my mood at the time the
choice is made. The second issue is how to disguise the old holes on the front
panel. There are two possible directions this could take, first use a trim
plate to cover the existing holes allowing complete freedom for the new layout,
or try to cover the existing holes with the new connectors. The new front panel
will include a 600 Ohm input to the transformer, and the existing high impedance
stereo mixer inputs, the modulation control will be history. The "stereo"
inputs will probably be a three circuit phone jack with switching contacts to
disconnect the transformer from the input. I'm not sure if the 600 Ohm input
will be a female three pin XLR connector, a three terminal barrier strip, or
perhaps a five terminal barrier strip.
I would like to incorporate some sort of carrier level stabilization circuit to
stabilize the carrier level when the tube is changed, although I may find this
is an unnecessary gilding of the lily. I have two different ideas for how to do
this if it appears desirable. First copy what the big boys did in the PDM
transmitters and use an RF rectifier to derive a voltage proportional to the
carrier level and use that voltage to control the bias on the modulator tube.
The second approach is to use a simple servo to control the DC voltage on the
modulator plate and indirectly the carrier level. The second approach is
probably more appropriate for this transmitter, as we don't really care what the
carrier level is, but do care about the modulator plate voltage because it
affects the modulation capability of the transmitter. In the later case as you
point out it would not be a good idea for the input to the servo circuit to
present a load on the modulator plate of less than 1 Meg. With luck none of
this will prove necessary and the circuit will be stable enough across tube
changes on its own. A third approach, the poor man's approach, would be to
provide a pot in the modulator cathode/grid circuit so the operating point could
be manually adjusted.
John Byrns
flipper <fli...@fish.net> wrote:
> On Sat, 27 Aug 2011 16:49:43 -0500, John Byrns <byr...@sbcglobal.net>
> I would think it probable. That and perhaps the crude coil model
> contributes.
IIRC you said the inductance values for the two ends of the coil were measured
values, which implies that you have some sort of inductance meter? If you do I
think it is a simple mater to measure the coil parameters necessary to build an
accurate model of the coil. I'm not going to attempt to describe how to do this
because over the past several days I have come to realize that my brain cells
that held this information have died off from disuse over the past three or four
years and I don't want to embarrass myself.
My gut feeling is that not modeling the coil correctly probably makes the "sim"
results somewhat inaccurate but isn't the major cause of the error with the
sawtooth, I suspect that that comes from the 12AU7 model. I will have to
printout the model you posted and try to understand it, I didn't look closely at
it but noticed that it may use some functions I am not familiar with and would
have to look up. It could be the grid model as you noted, or it could be
related to the model's saturation behavior, or who knows what.
> I did some surfing and discussions of 'tube models' invariably involve
> amplifiers with the models themselves derived from either static plate
> curves or theoretical derivations of them so it wouldn't surprise me
> if things are 'a bit different' at RF. For example, while simply
> tacking on some discrete inter-electrode capacitances may be fine on
> the audio end I wonder if that's valid at RF since, in reality,
> they're not 'separate' from the internal fields.
I'm not sure that I believe that 1.0 MHz is a high enough frequency to cause the
types of problems you are alluding to, my guess is that the problems relate to
the large signal behavior of the model, where the "twin triode" oscillator
pushes the tube into different portions of its charisteric curves than even a
class B audio amplifier does.
It might be possible to check this out by simply scaling ALL the L's and C's in
the "sim" to put the oscillator frequency at say 1 kHz, and see if the same
effects are occurring, or if the behavior fundamentally changes at the lower
frequency.
<Snip>
> I had noted that the osc grid resembled an AM envelope detector but it
> wasn't clear to me just 'how close' because the cathode is doing the
> modulation.
>
> What I see does 'look like' what you theorize, though. If you
> visualize a scope trace of it, negative modulation looks as if it
> can't get to the negative peak 'fast enough' and then, after the
> turnaround, rises more quickly than it fell. It makes negative
> modulation, if you figure the negative peak should be half way between
> the positive peaks on either side, look 'skewed' to the right (very
> much like using photoshop's 'free transform' skew function to drag the
> negative modulation peak over to the right.)
>
> >The second problem, which applies only to the "twin triode" circuit, is that
> >the
> >audio signal from the modulator is effectively in series with the B+, as in
> >all
> >plate modulated transmitters,
>
> Interesting you say that because now I have modulator Rk feedback so
> it's closer to a modulated current source. How does that affect our
> understanding, since you've been pointing out the two types operate
> 'quite differently'?
You lost me here, what two types have I been pointing out operate quite
differently? If you are talking about a current source modulator vs. a voltage
source, I'm not sure that makes much difference relative to the problem I am
talking about here involving C1 in the "twin triode" transmitter. The type of
modulator does make a difference in a real transmitter, even if the modulated
stage is perfectly linear in all senses, because the impedance the modulator
sees is actually a transformed version of the antenna/ground system impedance,
which may be a perfect 50 ohms at the carrier frequency, but may differ
considerably at the sideband frequencies in both magnitude and phase. It may be
somewhat a matter of perspective, but I think a stiff modulator copes better
with these real world conditions, although proper design of the
transmitter/antenna coupling networks and transmission line length in degrees
can go a long way towards minimizing this problem. Many real broadcast
transmitters used tetrode modulators, however they invariably used feedback from
the modulator plates to the audio input stage grids to minimize distortion in
the class B modulator and to provide a low source impedance to drive the load.
In the case of these mini transmitters, without sophisticated antenna matching
networks, it is my belief that the load on the oscillator is dominated by the
grid resistor, which is relatively constant, and not the antenna. That means
that the value of R3, the grid resistor, must be chosen to provide the optimum
load for the oscillator, and C! will affect the load seen by the oscillator.
The correct load is especially important in achieving linear modulation without
clipping in grid modulated circuits like the Zenith and "one tube wonder", and
is less important in the high level modulated "twin triode" circuit which is
more tolerant of loading.
> > and in this design that voltage, varying with
> >modulation also appears across C1. So the C1*R1 time constant must charge
> >fast
> >enough not to disturb the oscillator bias. This bias shift at high audio
> >frequencies is what my idea for a fix is designed to remedy.
>
> That's 6pF into 2.2K on mine. 12MHz pole an 'audio' problem?
Oops sorry, typo, that should have read "the C1*R3 time constant"! Also keep in
mind when thinking about this that the audio potential across the oscillator
tube is maybe an order of magnitude greater than the bias level on the grid, so
that even the tail end of the time constant charge/discharge curve is going to
have a greater effect than expected.
> >The idea is this, the big tapped winding of the oscillator coil is connected
> >only in the tuned grid circuit, part of the winding is not used for the
> >drive
> >from the plate. Instead the second winding on the AES (P-C70-OSC) coil
>
> I'm already out of luck as my coil does not have a second winding like
> the P-C70-OSC.
Your schematic shows a second winding, and while the coil on your breadboard
obviously isn't a P-C70-OSC, I thought you said at one point that you had a
P-C70-OSC, I guess I misunderstood.
> > is
> >connected in the plate circuit to provide the necessary feedback. I'm not
> >sure
> >if the second winding has a suitable number of turns, but I vaguely remember
> >that it is similar to the short end of the main winding and so should work
> >in
> >this way. Connecting the oscillator coil in this manner removes the
> >modulated
> >B+ from across C!, eliminating any possible bias shift problem from this
> >source.
>
> Well, I think I get the idea but don't see how it'll work because
> there's audio on Rk, so bias would have to track that to remain
> 'constant'.
>
> It might explain what seemed to me 'odd' bypass cap values on the
> screen and Rk. Maybe that's intended to 'compensate' for slew limiting
> above 3kHz, or so.
The last two paragraphs don't make any sense to me, presumably because my
erroneous mention of "R1" when I meant "R3" confused you.
> >Unfortunately, I don't believe this fix is relevant to the "one tube
> >wonder",
>
> I don't see why. It's the same circuit except for using a 6CS6 instead
> of the 12SA7.
I'm not sure how the 12AS7 comes into this? What I am trying to say is that my
potential fix is only applicable to that part of the problem unique to the high
level plate modulated oscillator used in the "twin triode" transmitter, it
doesn't, at least I don't think it does, have any impact on the "detector slew
rate" part of the problem which affects the grid modulated oscillators like the
"Zenith" and "one tube wonder" circuits, as well as the high level plate
modulated oscillator used in the "twin triode" transmitter.
> >nor would it appear to be a fix for the equivalent part of the problem in
> >the
> >"twin triode". The question relative to the "twin triode" is which problem
> >is
> >dominant, which would determine whether the "fix" would do any good.
> >
> >I found your mention of distortion caused by C3 interesting, as it hadn't
> >occurred to me, beyond its effect on frequency response and possible
> >distortion
> >in the modulator caused by excess capacitive loading of the modulator.
>
> Well, drawing on your theory I'd imagine it affects loading of the
> 'osc grid AM detector'. Affects grid input impedance and those
> positive grid currents are going straight into it.
Wouldn't the grid current be a second order effect compared with the plate
current?
> > I assume
> >that traditional plate modulated broadcast transmitters don't suffer from
> >this
> >charging time constant problem because the modulator used a push pull
> >circuit
> >topology with a low source impedance which could drive C3 to where it needed
> >to
> >be. The interesting question is whether the later PDM transmitters suffer
> >from
> >the problem you describe, to any extent?
>
> <snip>
> Well, that's why I breadboard. You can try things and defer all those
> agonizing decisions on how the 'final build' should go till you work
> out how you want it to work.
>
> >I would like to incorporate some sort of carrier level stabilization circuit
> >to
> >stabilize the carrier level when the tube is changed, although I may find
> >this
> >is an unnecessary gilding of the lily.
>
> A good example. You can breadboard the 'simple' way and worry about
> gilding the Lilly later.
However the same applies to reusing a chassis, where you can also leave worrying
about gilding the lily until later.
> Btw, things like this are one reason why I tend to resist 'fancy'
> ideas, at least to begin with, because I am more likely to breadboard
> a simple circuit I'm not sure of than a complex one I'm even more
> unsure of.
Yes, the need for an operating point stabilization circuit won't be obvious
until I have the basic circuit running. A good "sim", with the parameters of
the tube model varied over the likely range to be encountered in real life.
> > I have two different ideas for how to do
> >this if it appears desirable. First copy what the big boys did in the PDM
> >transmitters and use an RF rectifier to derive a voltage proportional to the
> >carrier level and use that voltage to control the bias on the modulator
> >tube.
> >The second approach is to use a simple servo to control the DC voltage on
> >the
> >modulator plate and indirectly the carrier level. The second approach is
> >probably more appropriate for this transmitter, as we don't really care what
> >the
> >carrier level is, but do care about the modulator plate voltage because it
> >affects the modulation capability of the transmitter. In the later case as
> >you
> >point out it would not be a good idea for the input to the servo circuit to
> >present a load on the modulator plate of less than 1 Meg. With luck none of
> >this will prove necessary and the circuit will be stable enough across tube
> >changes on its own.
>
> It's obvious that's where I would start since I did ;)
I'm not as sanguine about this issue as you are.
> It's self bias so all you really need do is make Rk large enough to
> cover worst case with the tradeoff being you put out a little less
> power when a 'hot' tube is present. But that's going to be pretty much
> the case no matter what 'stabilization' scheme you use because the
> (worst case) tube can't do more than it can do.
>
> Tighter bias feedback than Rk provides might get a few percents more
> power but it doesn't seem to me worth the added complexity in a
> 'simple' transmitter.
>
> > A third approach, the poor man's approach, would be to
> >provide a pot in the modulator cathode/grid circuit so the operating point
> >could
> >be manually adjusted.
I will try to post a schematic of my proposed 12SN7 transmitter circuit later
tomorrow.
It also occurs to me that the "plate" winding of the oscillator transformer can
probably be translated to the cathode circuit without changing the AC equivalent
circuit. This would allow the use of a single winding coil with my partial
"fix" idea. I will try to post a schematic of this too.
John Byrns
John Byrns <byr...@sbcglobal.net> wrote:
>
> I will try to post a schematic of my proposed 12SN7 transmitter circuit later
> tomorrow.
>
> It also occurs to me that the "plate" winding of the oscillator transformer
> can
> probably be translated to the cathode circuit without changing the AC
> equivalent
> circuit. This would allow the use of a single winding coil with my partial
> "fix" idea. I will try to post a schematic of this too.
I have posted a schematic of my "Twin Triode" AM Transmitter proposal here:
http://fmamradios.com/TTAMXmtr.gif
This version incorporates my "fix" for the grid bias modulation problem and also
illustrates the use of a coil with a single tapped winding, the two winding AES
(P-C70-OSC) coil is not required for this version.
Any and all comments are welcome.
Patrick Turner
> In article <byrnsj-24DAED.13464728082...@news.giganews.com>,
Your AM transmitter looks simple, and the bottom tube cathode
modulates the top tubes RF efforts. Alter the Ia and you alter the
amplitude.
But what's the envelope like?
I recall someone putting a similar 2 triode circuit like yours into a
magazine about 16 years ago when ppl here were turning back to tubes
and fixin' up their old AM radios. Just used one 12AU7.
Patrick Turner.
John Byrns
flipper <fli...@fish.net> wrote:
> On Sun, 28 Aug 2011 20:09:47 -0500, John Byrns <byr...@sbcglobal.net>
> wrote:
>
> >In article <byrnsj-24DAED....@news.giganews.com>,
> > John Byrns <byr...@sbcglobal.net> wrote:
> >
> >>
> >> I will try to post a schematic of my proposed 12SN7 transmitter circuit
> >> later
> >> tomorrow.
> >>
> >> It also occurs to me that the "plate" winding of the oscillator
> >> transformer
> >> can
> >> probably be translated to the cathode circuit without changing the AC
> >> equivalent
> >> circuit. This would allow the use of a single winding coil with my
> >> partial
> >> "fix" idea. I will try to post a schematic of this too.
> >
> >I have posted a schematic of my "Twin Triode" AM Transmitter proposal here:
> >
> >http://fmamradios.com/TTAMXmtr.gif
> >
> >This version incorporates my "fix" for the grid bias modulation problem and
> >also
> >illustrates the use of a coil with a single tapped winding, the two winding
> >AES
> >(P-C70-OSC) coil is not required for this version.
>
> Oh, ok. Well, that's what I called the typical AA5 'cathode-grid'
> Hartley. Except, of course, there's not a modulator under it in the
> AA5.
>
> Well, we know circuitmaker isn't modeling the other one right so for
> what it is or not worth circuitmaker says the AM envelope is not
> symmetrical. The 'upper' envelope is a smaller amplitude than the
> lower.
>
> However, I think that's again due to the strange osc cathode voltage
> it comes up with and, so, 'not right'.
That is a disappointing result, I wonder if the problem is somehow inherent in
the morphed circuit, or if I morphed it incorrectly as I didn't draw out the
steps as I had planned to and instead morphed it in my head, or is it a problem
with the simulator? Did you use the same component values & 12AU7 as in your
original "sim", to eliminate double diddling?
That circuit wasn't the one I originally planed on posting first, I had intended
on posting my preferred "two winding" design first, as it has fewer changes from
your circuit. It would be interesting to see what your simulator does with my
preferred proposal. There are two steps to my preferred circuit. The first
step is to move the lead of C2 that is connected to L1 pin 3 to L1 pin 4
instead, and then scale the inductance of L1, adjust the core in the physical
implementation, to maintain the same carrier frequency so that the reactance of
C2 at the carrier frequency will remain the unchanged, before running a "sim".
The second step, which yields my target design, is to split the tapped coil into
two separate windings by disconnecting the grid part of the coil from the center
tap at pin 4, along with the end of C2 that had been moved to pin 4 in the first
step, and then reconnecting them both to the oscillator cathode, yielding my
proposed two winding circuit.
John Byrns
Patrick Turner <in...@turneraudio.com.au> wrote:
> On Aug 29, 11:09 am, John Byrns <byr...@sbcglobal.net> wrote:
> > In article <byrnsj-24DAED.13464728082...@news.giganews.com>,
> > John Byrns <byr...@sbcglobal.net> wrote:
> >
> >
> >
> > > I will try to post a schematic of my proposed 12SN7 transmitter circuit
> > > later
> > > tomorrow.
> >
> > > It also occurs to me that the "plate" winding of the oscillator
> > > transformer
> > > can
> > > probably be translated to the cathode circuit without changing the AC
> > > equivalent
> > > circuit. This would allow the use of a single winding coil with my
> > > partial
> > > "fix" idea. I will try to post a schematic of this too.
> >
> > I have posted a schematic of my "Twin Triode" AM Transmitter proposal here:
> >
> > http://fmamradios.com/TTAMXmtr.gif
> >
> > This version incorporates my "fix" for the grid bias modulation problem and
> > also
> > illustrates the use of a coil with a single tapped winding, the two winding
> > AES
> > (P-C70-OSC) coil is not required for this version.
> >
> > Any and all comments are welcome.
> > John Byrns
> >
> > Surf my web pages at, http://fmamradios.com/
>
> Your AM transmitter looks simple, and the bottom tube cathode
> modulates the top tubes RF efforts. Alter the Ia and you alter the
> amplitude.
>
> But what's the envelope like?
The envelope should be identical to that of the design you refer to below,
assuming the same component values are used, as the two circuits use the same
identical RF and audio topology with the power supply/modulator simply
translated in the circuit, while maintaining the same RF/AC topology, to
minimize one undesirable RF/Audio interaction in the original circuit.
> I recall someone putting a similar 2 triode circuit like yours into a
> magazine about 16 years ago when ppl here were turning back to tubes
> and fixin' up their old AM radios. Just used one 12AU7.
It sounds like the circuit you are talking about is the same identical circuit
that I first encountered 17 years ago here on usenet in the form of an ASCII art
schematic, which I posted for flipper's edification earlier this year, and on
which he based his "Twin Triode AM Transmitter" which is the subject of this
thread.
Patrick Turner
> In article <b29c9d9c-a71d-4de0-8f1d-569b7ee60...@u7g2000pro.googlegroups.com>,
>
> - Show quoted text -
I'm sure dozens of ppl have tried to make an AM transmitter with just
one 12AU7, since say 1947, and probably hundreds who have made a
transmitter with just two tubes, ie, maybe pair of 6J5, or 6C4, or 6J7
etc, etc, etc.
Such a thing makes an interesting back-to-your-schoolboy-years hobby
thinking & delights.
But as soon as you make one with a buffered oscillator, AM detector
and diff pair amp for the modulator amp and perhaps Push Pull for low
THD then there is a much greater challenge - and a better looking
envelope.
The first transmitter I made at age 14 was a converted AM radio where
a second 6V6 was added where the IF tube was, and then the existing AF
amp used the OPT to cathode modulate the added 6V6 RF output.
Oscillator was made using the mixer socket, with a very simple
oscillator, not buffered.
Worked fine, no NFB, but fairly low % of mod, OK to talk to my friend
down the street.
Patrick Turner.
John Byrns
flipper <fli...@fish.net> wrote:
> On Mon, 29 Aug 2011 20:38:28 -0500, John Byrns <byr...@sbcglobal.net>
> Took me a few minutes to decipher those instructions but I figured it
> out.
>
> Good news is it 'works' and the plate signal looks right. Bad news is
> the plate signal is only 6.8Vrms so there's little RF power from
> there.
>
> More bad news, the 'big endian' coil we get 'mo power' from looks
> asymmetrical like the other one. Worse actually. It's not only
> asymmetrical but on deep mod the 'top side' is heavily distorted from
> what 'looks like' severe slew limiting. Which seems odd since that's
> what we're supposedly trying to fix.
No, the slew rate problem isn't the one I am trying to fix, your scheme of
reducing the time constant is the only way anyone has suggested to fix the slew
rate problem. The problem I am trying to fix is a slightly different but
related problem, which is also "fixed" by reducing the time constant as you have
done.
> The same 'tweaks' work with this one so, by using a 47k series grid
> limiter, I could get osc cathode to ride down around 122V, which
> eliminates the 'cathode crunch' up against the plate signal, so I have
> a sub fraction more semi confidence it might mean something.
>
> I left all values the same as mine except for increasing C2 to 180pF
> to get a similar frequency.
OK, that would reduce the capacitive reactance at the carrier frequency, that
may or may not be a big deal, it will probably increase the operating Q of the
coil considerably, resulting in greater high frequency rolloff.
> On mine I went back to 47k for R3 and it seems to have brightened
> things up a smidgen and it dawned on me,
Yes, that is part of the effect I noticed working with the "Zenith" circuit.
> when I was measuring
> frequency response and getting 'lower amplitude' at 10kHz I was using
> full mod. So maybe the small signal response does improve and it's
> slew limiting I was seeing limit the amplitude. Maybe just wishful
> thinking and I should take it back up to the scope but, for now, I'm
> listening to it.
Could be, I will have to take a few minutes to think about what you are saying
here.
John Byrns
Patrick Turner <in...@turneraudio.com.au> wrote:
> The first transmitter I made at age 14 was a converted AM radio where
> a second 6V6 was added where the IF tube was, and then the existing AF
> amp used the OPT to cathode modulate the added 6V6 RF output.
Did you use the existing OPT secondary winding, that normally drives the
speaker, to cathode modulate the added 6V6 RF output, or was the cathode
modulation achieved by other means?
This type of cathode modulation sounds like a 14 year old scheme, I hate to
think of the things I tried at age 14.
> Oscillator was made using the mixer socket, with a very simple
> oscillator, not buffered.
The separate oscillator and modulated amplifier are a big step beyond what we
are talking about here, buffering the oscillator would be gilding the lily.
> Worked fine, no NFB, but fairly low % of mod, OK to talk to my friend
> down the street.
Assuming the original OPT secondary was used to provide the cathode modulation,
I would expect a considerably greater % of mod and considerably lower distortion
if you had used Heising modulation, which you could have done with the same
parts.
John Byrns
flipper <fli...@fish.net> wrote:
> On Tue, 30 Aug 2011 21:30:00 -0500, John Byrns <byr...@sbcglobal.net>
> wrote:
>
> >In article <l0no5798fblja6of9...@4ax.com>,
> > flipper <fli...@fish.net> wrote:
> <snip>
>
> >
> >> when I was measuring
> >> frequency response and getting 'lower amplitude' at 10kHz I was using
> >> full mod. So maybe the small signal response does improve and it's
> >> slew limiting I was seeing limit the amplitude. Maybe just wishful
> >> thinking and I should take it back up to the scope but, for now, I'm
> >> listening to it.
> >
> >Could be, I will have to take a few minutes to think about what you are
> >saying
> >here.
I took 6 days off from thinking about exactly how I will rebuild my AES phono
oscillator with a high level series modulation circuit. In addition to my
wife's project to reduce what I call domestic entropy, which will probably
extend well into 2012, when you are "94", Patrick's number not mine, health
issues crop up which demand attention. The most time consuming though was
purchasing a new car for my wife, which involved scouring the country for one
that was built before, or shortly after, the tsunami in Japan ended the
availability of the optional $395.00 paint job she wanted. The tsunami seems to
have ended the production of a particular type of metallic flakes that were used
in the desired paint.
> Well, I did some more testing.
>
> I started with the intent to put my air variable in and see how it
> works over the whole AM band and, as suspected, I had to increase C1
> to 20pF for it to oscillate on the lower end.
>
> My cap is a typical AA5 dual gang with grounded frame so that meant I
> had to put it on the coil big end only, which meant I had to screw the
> slug full in.
>
> Still, I was able to get around 540 to 1700 and figured the top end
> would drop when I tacked the antenna back on but, to my surprise, not
> only did the top end come down but the lower end of the range came up
> to, say, 650 minimum.
That's interesting, do you have any tentative explanation for the low end
increasing from 540 kHz to 560 kHz when you connected the antenna, at first
glance it doesn't seem to compute?
> The BIG surprise though was the 10kHz amplitude response I talked
> about previously. With everything else remaining the same it changes
> with carrier frequency from rather lousy on the low end to looking
> pretty decent at the very top, and deep mod slew distortion follows
> hand in hand.
That shouldn't have been a big surprise, when the tuning is done with a variable
capacitor the circuit Q increases when tuned to lower frequencies, remember that
the grid resistance forms the main load on the oscillator, so the band width
becomes narrower at the low end, plus there is a double whammy because not only
does the Q increase, but the carrier frequency is lower further narrowing the
bandwidth.
> I don't understand why but it settles the tuning range question. Pick
> C2 so the range is pushed up to the top end.
The "top end" seems to be a better choice anyway because frequency availability
is better up there, and if you are going for a fancy matched antenna to maximize
range, the antenna is likely to be more efficient at the "top end".
Don Pearce
>It's posted on the web page so see what you think.
>
>http://flipperhome.dyndns.org//Twin%20Triode%20AM%20Transmitter.htm
Only the photo of the breadboard is showing. And the text - pale blue
upon white - is near illegible.
d
Don Pearce
OK, a second try at loading the page sorted it.
d
Don Pearce
>Ok, good.
>
>I've also noticed some odd Internet behavior but don't know if it's my
>end or the ISP (it's usually not my end).
I notice the site is directed through dyndns. Is it on your own home
PC or does the ISP host it?
d
Don Pearce
>It's here at home on a old P133 notebook running Windows98SE. The web
>server is Savant. That's all it does.
>
>Local LAN access is zippity do da fine.
I think the problem is the dynamic dns server. It probably needs
reminding who you are from time to time. The web site would work much
better if you hosted it your ISP.
d
John Byrns
sp...@spam.com (Don Pearce) wrote:
However it would be lacking in cool factor if hosted at the "ISP".
John Byrns
flipper <fli...@fish.net> wrote:
> On Sun, 04 Sep 2011 20:25:25 -0500, flipper <fli...@fish.net> wrote:
>
> <snip>
>
> >
> >Some sort of modulation meter would be more useful but neither easy
> >nor particularly inexpensive, compared to the rest of it.
> >
>
> Well, I kludged up a simple 'modulation indicator' in spice, but I
> haven't actually built it.
> It's posted on the web page so see what you think.
>
> http://flipperhome.dyndns.org//Twin%20Triode%20AM%20Transmitter.htm
>
> something like the LM3915 but the modulator drive has an interesting
> problem in that trying to sense 'max' positive mod may not be so
> simple due to positive grid drive potentially clipping the signal. It
> shouldn't if everything is 'right', but it might not be, and then
> there's the matter of 'calibrating' the thing even if all else is
> okay.
>
> Positive mod isn't the 'real' (or biggest) problem anyway, it's
> carrier cutoff on negative mod, and 0 modulator current has the
> advantage of being 0 no matter what 'idle' or peak positive mod is. So
> I look for modulator cathode approaching 0, rather than worry about
> 'calibrating' to positive mod, then peak hold/decay it for visual.
> (Note, the LM3915 doesn't peak hold so you could easily miss seeing
> carrier cutoff events even if you got it calibrated right)
>
> The yellow 'mod' LEDs are for visual effect. The R15/C19 filter senses
> 'idle current' and U1B, then, generates (rectified half sine) pulses
> proportional to positive mod to drive the LEDs. We don't care if it's
> terribly accurate as it's sort of like those 'musical lights' things
> where light intensity dances around to the music.
>
> The idea is a sort of 'glow indicator'. Like, maybe, aiming the LEDs
> into a diffuser, or not. I haven't worked out the details. Probably
> need to buy some 'wide angle' flat LEDs and see just how well they
> disperse and how it looks.
>
> In theory it would be nice to know our peaks are -3dB, or whatever,
> but for something this simple I think people would slam it as high as
> they can anyway, so a red 'too much' indicator is probably sufficient,
> and calibration, as mentioned, would be non trivial so it's unlikely
> -3dB would have much meaning to the actual problem. At least that was
> my thinking: a modulation 'idiot light', to use a car analogy.
>
> A simple tripler off the heater supply should be enough to power it.
It's an interesting coincidence that you are working on this idea, as I have
been thinking along similar lines the past couple of days.
As you allude to above, I think the important thing would be to monitor the
onset of oscillator dropout and flash warning indicator at that point.
The idea that struck me was to use a neon bulb like some 1950s era tape
recorders used as recording level indicators. To a first approximation a neon
bulb and two resistors would do the job. One connection of the neon bulb would
be go to the modulator plate, with the second connection to the junction of two
resistors forming a voltage divider between B+ and ground. The potential at the
junction of the two resistors would set the firing voltage for the neon bulb,
and the fact that the firing voltage is higher than the conduction voltage would
provide a sort of faux peak hold effect. Calibration of this circuit should be
easy as we only have to identify the point where the oscillator drops out, not
the precise point of 100% modulation.
There is at least one obvious problem with this elegantly simple scheme, which
is that when the neon bulb fires it would drag the modulator plate down to a
lower voltage, creating serious distortion. This problem could be overcome at
the cost of an additional triode operating as a cathode follower to drive the
neon bulb, buffering the modulator plate. Unfortunately besides requiring the
extra triode, an additional higher voltage power supply would be required for
the cathode follower plate, to prevent it from saturating before the point where
the oscillator drops out.
Most old vacuum tube era professional AM broadcast modulation monitors required
calibration vs. an external source such as Patrick's trapezoidal CRO display.
There was one old modulation monitor though which was self calibrating, that was
the Gates M5693 Modulation Monitor. The ideas from this monitor should be easy
to adapt to the modern solid state opamp era. A year ago I worked out a scheme
to implement this self calibrating idea in a single quad opamp to drive an
analog peak modulation meter. A little additional circuitry, as in the actual
MS5693, could provide for driving a red peak flasher LED, and the analog meter
could be replaced with an orange LED "VU" meter.
An AM demodulator would be required to drive this circuit, the oscillator plate,
which is a lower impedance point than the grid circuit where the antenna is
connected, might provide enough RF drive voltage for a demodulator diode without
unduly loading the oscillator circuit.
If you want to puzzle out the self-calibrating principle used in the M5693, the
operating instruction manual and schematic can found here.
http://louise.hallikainen.org/~harold/BroadcastHistory/uploads/GatesM5693.pdf
Don Pearce
>That wouldn't explain the recent 'in general' odd Internet behavior
>I'm seeing since none of that is dyndns related.
>
>> It probably needs
>>reminding who you are from time to time.
>
>I have a semi static IP. I haven't ever seen it change 'on it's own',
>it's just not 'guaranteed'.
>
It will have an assigned lifetime and will expire if you don't log in
for some time. That time is probably upwards of a week.
>> The web site would work much
>>better if you hosted it your ISP.
>>
>>d
>
>I prefer the control and it being ISP 'independent'.
Hardly ISP-independent. Your IP address still comes from the block
assigned to your current ISP.
d
Don
fli...@fish.net says...
Where one *finds* 6DR4s, though, I don't know
>as ABCvacuumtubes doesn't list it and they seem to be rather rare.
http://www.tubedepot.com/tcp8.html $4.95