> >Its awhile since I mucked around with a Gilner, or Turbert Cell or
> >whatever I tried to use as an am modulator with some simularity to
> >Gilbert, which does have more than 3 transistors, 6 as you say.
> >But looking at my notes and the collection of Gilbert cell schematics
> >I have, what I used was a 1/2 Gilbert cell with an LTP driving
> >balanced RF OPT, one side of LTP to 0V, other side to RF carrier F,
> >and tail is transistor collector with AF applied to base. Very simple,
> >but not as good as using tubes as I had already.
>
> Well, I think the balanced RF OPT was overkill but the 'LTP' should
> work fine, unless you were holding one side to 0V without bipolar
> supplies because there has to be 'something' across the bottom
> transistor.
>
> The basic topology you describe is the one I 'almost built':
http://flipperhome.dyndns.org/AM%20Transmitter.htm
> And, as previously mentioned, it is the solid stage version of Beamus.
> I.E., compared to the 6ME8, the bottom transistor is G1 and the top
> two are the deflection plates.
>
> > Maybe Gilbert himself
> >used 6 triodes to make his original Cell,
>
> Barrie's original cell was bipolar.
I suspected it was, nothing tubed on Google, but whatever you can do
with bjts, you can also do with tubes, well, nearly whatever anyway,
until you need a pnp tube, which can't exist, and tubes are all npn,
and I mean figuratively, not literally..
Gee, that looks really nice and simple! Makes me wanna build one, I
might have a few 6DT6 laying around.
>
> The 6BN6 is a whole different thing: a gated beam discriminator and
> limiter with the express purpose of rejecting AM.
>
> >I know you are hooked on using just one tube,
>
> Well, no. I was only 'hooked' on using one tube back when I was
> experimenting with using one tube. That was 'the point' of it: to see
> what one could do with one tube.
And now you know, and not bad considering you chose a tube which can
wear so many hats and does so many things, and which hardly anyone
knows anything about.
>
> But everything I've done since that experiment involves more than one
> tube.
>
> What would be a more accurate characterization is I like the challenge
> of getting the most'est from the least'est.
OK, but to me the 6ME6 is a "leastest with mostest" type of tube.
>
> > and someone said stuff
> >about FM being created and maybe that's true, because the audio
> >changes the effective C across a coil that determines the F so the
> >tuning F changes. I have a separate triode oscillator,
>
> So does Beamus. Well, except I used a pentode osc.
>
>
>
>
>
> > then CF buffer,
> >and audio cannot get near anything oscillating at RF. I have a pentode
> >RF amp tube, and fairly large Rk bypassed for RF, but not for AF, so
> >when AF is applied to g1 with the RF from resistance divider from CF
> >RF source, the tube current is changed fairly linearly because there
> >is local current FB at AF, so wave form wasn't so terribly bad as say
> >50% mod. But adding more tubes to get NFB applied to audio stage much
> >reduced THD in wave form, and got me closer to 100% mod. I found many
> >old radios make a real mess when they try to detect audio when
> >modulation exceeds 50%, and of course back in good old days AM was not
> >fully modulated. But when you look at radio station waves, there is
> >high % mod, and of course compression maybe. And many superhet radio
> >sets can't produce a clean 100% mod IF wave when there is a 100% mod
> >RF wave input. I'm not sure how much NFB you have but certainly it
> >does work to make AM better you have non linear tubes or feeble tubes
> >to deal with.
>
> I'm not sure what 'feeble' is supposed to mean but I estimate NFB at
> between 15 dB and 20 dB.
Feeble means low Ia and low gm, and thast usually means low "figure of
merit". For example, EF86 is a gutless wonder of a tube which ws used
in thousands of amps in 1955 where one wanted gain between 80 and 150
and with only one nine pin socket and 0.7mA of anode current. It
replaced the use of a 12AU7 with 2 halves cascaded to get similar
gain, but with more total Ia and more R&C parts. I'd rather have the
12AU7. Look at Quad-II with EF86. Bloody awful! To get the gain needed
Quad used 180k RLa resistors and following bias R for KT66 were 680k
which often began to develop a high positive biasing voltage when
tubes age a bit. One shouldn't have Rg more than 120k max, ever, but
that weighs down the gain of feeble EF86. Later, in 1995, Chinese who
bought Quad used Andy Grove's revised Quad-II-Forty circuit with 6SH7
in pentode, almost identical in operation to early Quad-II. 6SH7 is an
octal tube with potentially far higher gm than EF86, or the 6SJ and
6SJ7 family of pentodes. Why Grove went with the original Quad-II
schematic is a mystery when it would have been much better to use 6SH7
in triode followed by 6SN7 as an LTP with CCS tail is beyond me. I
have repaired and re-wired a few Quad-II-Forty, because once you take
a careful look beyond the nice looking Chinese paint, its a POS amp,
and they have 6SH7 operating with Ia way below where its figure of
merit would be high, and with 470k biasing for the KT88 instead of the
KT66. Instead of gutless wonder pentodes with Ia at 1mA, I like more
linear triodes with 4 or 5 mA at least.
Enough about feeble tubes. In RF work, I like to see healthy Ia
because capacitances anywhere tend to have low Z paths at RF.
15-20dB FB is good, enough to do something worth doing, assuming the
gear makes maybe 7% Dn at 90% mod without NFB.
But I found that with NFB applied, the modulation bandwidth with
undistorted wave at say 95% mod was reduced, and to be expected of
course. In other words, modulation can be 95% OK with AF from say 15Hz
to 5kHz, with envelope wave having THD < 1%, and enough to test most
AM radios, but above 5kHz, the envelope THD increases. OK though at
50% mod, good from 15Hz to 20kHz, less than maybe 0.5%. Its hard to
measure because one needs a linear detector, But then with dual trace
CRO one can compare input AF with recovered AF at detector output of
receiver, and if that detector is my double triode type with two CF
triodes then any visible distortion is likely to be in receiver mixer
and IF stage, not in sig gene, not in detector.
The AM transmissions here from radio stations are pretty good, and
probably with lower THD + IMD than in 1955, so recievers are the weak
link. My own AM set has cascode vari-mu twin triode input, 6AN7 mixer,
then 6BX6 IF amp with unbypassed Rk for some local current FB and no
AGC voltage is applied to it. 6BX6 or EF80 is a sharp cut off pentode
with strong gm normally used for 4.5Mhz TV IF stages. Its operating
with fixed bias. AGC voltage is generated by the detector CF stage,
and it works on mixer just a bit, and on RF input tube. I have
variable spacing between IFT1 like Hellicrafter used on their
communications radios with 3 IFTs to vary the Q for the IF band. But I
vary between flat top pass band and twin peak, so with two IFTs, and
IFT2 with normal 9kHz pass band, I get 18kHz pass band and 9kHz of AF
bandwidth. Some HF emphasis in R&C network boosts the AF BW to 10kHz.
I hate to seem immodest, but my AM radio outperforms ALL other AM
radios I have ever placed beside it to compare them, including Quad's
AM tuner which I think is next best. My point is that OK, one makes a
nice trick thing like a decent AM "mini-transmitter" and you can put
on a CD in the garage and fool ppl into believing there's this nice
new radio station without adds and some BS artist, but what about the
radio receiver perfomance?
Most AM radios are shit boxes, fair dunkum. The worst are the 3
transistor types added to an AM/FM receiver-tuner, 1.5kHz of audio BW
if you are lucky and very high THD IMD. Tubed types vary between
passable and "it needs total re-engineering". Passable means 4kHz BW,
typical THD/IMD 4%. Good means 7kHz BW, THD/IMD < 2%, and Excellent
means 10kHz and Dn < 1%.
There's no use going for more BW because AM transmissions are limited
to less than 10kHz, usually. Only 4.5kHz, where stations are close
together.
Patrick Turner.