Choke input PS
John Galbraith
thinking about using a choke input which would be my first try at this
architecture. There would be two supplies and three regulators, for the phase
inverter around 450V, driver tube B+ around 200V, and the second supply for
the negative supply at -300V. I already have the regulators working (that was
a trial). Turns out that all of these functions are constant current (class
A), in contrast to my power stage which has yet another capacitor input supply
that is already working fine. So I thought it might be time to try a choke
input.
I've already ordered the chokes, because I can use them in a PI filter if it
turns out the choke input won't work for me (four Lundahl 20H, 100mA chokes).
Now I am struggling with figuring out if those voltages are too high for a
choke input in practice. Basically, that 450V supply is going to need a power
tranny secondary in excess of 500V RMS, because the rule of thumb for choke
input is that VDC = .9 V RMS. With that kind of secondary voltage, the
rectifiers see 500*sqrt(2) > 700 V in a bridge configuration, so it would take
rather large diodes to hold it off. In theory then one could get away with a
filter cap rated for 500V, except any mistakes or blown circuits that shut off
the current make the output voltage go from .9*500 to sqrt(2)*500 in a choke
input. So the 500V cap would blow.
Is the choke input really worth this kind of pain? In this case, I would
basically get to pay for a 700V supply but only get 450V before the regulator.
BTW, thank you to whoever suggested using "PSU Designer II". As I reported
before, it runs in Linux under wine just fine and I am using it that way.
robert taylor
Choke supplies aren't worth it if money, size, or weight are the primary considerations.
Way more than worth it for sound quality in high end hi-fi though...
wouldn't know if this is about instrument amps, intentional distortion and such.
>
> You received this message because you are subscribed to the Google Groups "audionerdz" group.
> To post to this group, send email to audio...@googlegroups.com.
> To unsubscribe from this group, send email to audionerdz+...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/audionerdz?hl=.
>
>
Windows 7: It works the way you want. Learn more.
John Galbraith
> Choke supplies aren't worth it if
> money,
> size, or weight
no concern at all
> are the primary considerations. Way more than worth it for sound quality
> in high end hi-fi though... wouldn't know if this is about instrument
> amps, intentional distortion and such.
the choke input for a hi-fi sound upgrade. It's still painful...
John
robert taylor
One of the electricians @ work saw what I was looking @ the other day, when we were talking about dedicated lines, grounding, etc.
He said that "over the top" system that I sent the link for was just a commercial lightning rod installation system.
The salt filled 2" dia x 10' copper pipe for ground rod set in bentonite, 4 0 copper wire, should have 2-3 ohms from rod to wire.
Should follow UL96a if you want to be code, but no one checks it.
That they did LOTs of them @ the lab, on roofs all over the place, no big deal.
Hotmail: Trusted email with powerful SPAM protection. Sign up now.
robert taylor
Several ideas, methods floating around out there besides textbook if not that may be worth thinking about.
I believe Matt had good luck w/ the low DCR approach...
I had a good conversation w/ Jeffrey Jackson recently about this one.
Interesting, from one who's tried a lot of configurations, and builds the best amps I've ever heard.
John Galbraith
> Have you already designed/decided on all components/values?
> Several ideas, methods floating around out there besides textbook if
> not that may be worth thinking about. I believe Matt had good luck w/ the
> low DCR approach...
supply. So yeah, I have made some choices. I'd rather not change the
architecture or tube types, but I'll change resistor and current values any
time before I upgrade them to fancy components.
> I had a good conversation w/ Jeffrey Jackson recently about this one.
> Interesting, from one who's tried a lot of configurations, and builds the
> best amps I've ever heard.
John
robert taylor
In a nutshell, from Dennis:
Here's the goodies:
(1) The Power Transformer can benefit a little bit by being low-DCR-- but the largest
benefit will come from extra current capability, regardless of DCR. about 150 mA as a minimum.
(2) The rectifier doesn't matter at all-- as to DCR. It is super-active, and can address current/voltage changes rather instantaneously. BUT-- just as in the transformer-- extra current capability here is a winner, regardless of DCR.
(3) The input choke should be low-DCR. This is because we want a TIGHT COUPLE to the rectifier output. Once the power leaves the rectifier, it is no longer adjustable-- actively. Now, we have passive components, unless we install voltage/current regulation-- but we don't want to listen to those things struggling to "keep-up" with music.... so they're best left out.
(4) Capacitors should be low-capacitance. This is because we want any "tuned" circuits to stay above human hearing--- by several orders. You can say here that we want SPEED. That simplifies it.
(5) Rules-of-thumb in designing the total amplifier:
(A) High-Current power transformer.
(B) High-Transparency rectifier.
(C) CHOKE-INPUT.
(D) low-DCR chokes.
(E) SMALL capacitors.
(F) Low-DCR wiring-- everywhere.
(G) Anything that is in-series with power or signal-- low-DCR, and physically short.
(H) Anything that LOADS power--or signal--(plate & cathode resistors, input impedance, etc.),--- VERY high DCR.
----D----
Ground wire needs to be a minimum of 10 or 12 gauge ROMEX, with one stand of Kimber TCSS litz in parallel to it in
You want chokes that are 10 ohms DCR, or less, and with a small core, and 1/2 HY or lower in inductance.
An example of this, a two stage amp, built w/ separate supplies, a 76 into a 245:
(L1/C1/L2/C2/L3/C3)
· Signal 500 VA PT and a 5U4GB hybred Graetz full-wave bridge rectifier,
· L1 is Triad C-56U, 35 mHY @ .75 ohms DCR @ 2A.,
· C1 is a 10 uF round oil,
· L2 was a Triad C-40X, .32 HY @ 10 ohms DCR @ 600 mA.,
· C2 was a 40 uF ASC oil from Michael Percy,
· L3 is another Triad C-40X, and
· C3 is another Percy 40 uF ASC.
· Oh, one other BIG BIG BIG improvement came when we significantly dropped the OP-POINT on the 245s so that they more closely reflected a Golden Ratio percentage ( 62% ) of the Manufacturer's suggestions.
- - - - - - - - - - - - - - - - - - -
This is not what Jeffrey uses.
He says there is too much ripple (and audible hum with sensitive speakers, like the 109dB horns he uses).
He uses L1 calculated to critical inductance.
L2 is about half that. L3 about half L2, and a high quality, nickle piece.
Caps are smaller than what Dennis and Jeff Medwin use, and the last one bypassed w/ a white Solen tin foil.
--Usually uses MV rectifiers too.
These are much better sounding amps IMO, but costlier.
http://www.jeffreywjackson.com/products.htm click picts for links for a few details.
>
> You received this message because you are subscribed to the Google Groups "audionerdz" group.
> To post to this group, send email to audio...@googlegroups.com.
> To unsubscribe from this group, send email to audionerdz+...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/audionerdz?hl=.
>
>
John Galbraith
> The "low DCR" isn't Jeffrey's main "thing," it's part of Dennis Fraker's L
> SES idea. I believe this is what Matt used, and posted about, maybe a
> year ago?
> In a nutshell, from Dennis:
>
> Here's the goodies:
> (1) The Power Transformer can benefit a little bit by being low-DCR-- but
> the largest benefit will come from extra current capability, regardless of
> DCR. about 150 mA as a minimum.
the transformer? It's true you get low resistance at DC (that's what "DCR"
is, right?). Maybe high frequency performance it just doesn't matter in the
power transformer? Maybe you want it poor, to reject line noise?
> (2) The rectifier doesn't matter at all--
> as to DCR. It is super-active, and can address current/voltage changes
> rather instantaneously. BUT-- just as in the transformer-- extra current
> capability here is a winner, regardless of DCR.
> (3) The input choke should
> be low-DCR. This is because we want a TIGHT COUPLE to the rectifier
> output. Once the power leaves the rectifier, it is no longer adjustable--
> actively.
("super-active"?) Sounds like he wants a low output impedance, probably across
a wide bandwidth, which is a more standard way of expressing the design spec.
I would agree, if that's what he meant.
> Now, we have passive components, unless we install
> voltage/current regulation-- but we don't want to listen to those things
> struggling to "keep-up" with music.... so they're best left out.
slow it down (which in a real paper I would express as something like "reduce
the bandwidth and degrade the transient response") and could be heard easily,
and one would probably rather have the noisy B+ than those degradations. That
is because the current sourced by the power supply is literally following the
music, and the regulator has to track it.
However, on constant current stages like a differential pair, cathode
follower, or common cathode, a regulated supply seems like it might offer some
advantages. Both the supply current and voltage should always be fixed, and
the regulator helps with the fixed voltage part. The fixed voltage helps keep
the current fixed depending on the current source bias circuit, because the
current sources need a reference somewhere and a stable supply improves the
performance of those references.
Granted, the regulator probably still needs to maintain high bandwidth, but at
least to first order the amplifier stage is not modulating the current or
asking for much performance from the regulator. Another way of thinking
about this is that the regulator does not have to *slew* quickly (a large
signal parameter, different than but related to bandwidth). Second order
effects might include the finite output resistance of the current sources,
which could modulate the current with the music ever so slightly even in one
of these stage types. But these small second order effects might be more
bearable than ripple and a bunch of other crap on the unregulated power
supply.
My regulator that I have right now uses LM317, and MJE340 devices. The NPN
has f_t of 10Mhz, which is probably slower than the LM317 because it is a high
voltage device (300V). If the regulator has (I'm guessing) 1MHz bandwidth,
that seems pretty good to me. I would like it even higher to complete
squashing any RF transients from my rectifier or other crap that comes through
my wall outlet, but at some point it is splitting hairs. And in return, I get
totally constant DC supply rails and current source references. My scope
cannot even measure any residual ripple or RF after the regulator (well, it
can see RF junk depending on which outlet I use).
> (4)
> Capacitors should be low-capacitance. This is because we want any "tuned"
> circuits to stay above human hearing--- by several orders. You can say
> here that we want SPEED. That simplifies it.
wide bandwidth. He has discovered that small capacitors do this better than
large capacitors, because the large capacitor comes with more inductance that
makes it act more like an inductor at some frequency. I agree, but it might
be possible to build a high quality, large capacitor, with low inductance.
Then you get the "speed" but you get to keep the large size. I'm hoping that
these Mundorf brand "tube-caps" are like that; I've bought four of them.
One can show this parameter trivially on a network analyzer.
> Ground wire needs to be a minimum of 10 or 12 gauge ROMEX, with one stand
> of Kimber TCSS litz in parallel to it in You want chokes that are 10 ohms
> DCR, or less, and with a small core, and 1/2 HY or lower in inductance.
bus. I can't wait to try it. Using the litz for better high frequency
performance sounds like a great idea; I'm going to try that too. The bulk of
the low frequency currents will flow through the bar, and the litz will take
over when the bar's inductance becomes a factor.
> An example of this, a two stage amp, built w/ separate supplies, a 76 into
> a 245: (L1/C1/L2/C2/L3/C3)
The 3dB frequency will be high, but the filter's slope above that frequency
will be super steep (6*6 components = 36 dB per octave? Is that right?).
Still, those are pretty small L's...
Wow, that ended up being a long post...
John
robert taylor
> This part makes me cringe, a little
I keep thinking the guy stumbled onto something he liked the sound of, then tried to come up with a "folk explination" of how it worked.
And he claims to be an engineer.
A lot of mud-slinging on that forum from two camps, the engineers explaining why it can't work, and a few who just say "just build it and listen; then tell us what you hear."
Not exactly what I would do, (Yes, I have heard Dennis' amps, very nice, but not like Jeffrey's).
And Jeffrey's explanations' make sense, are backed by measurements, actual tests, and his amps sound better than any others I've ever heard.
Caps-- I haven't heard anything from anyone personally using Mundorf tube caps in a PSU. Might work well; I like their using 2 windings to cancel out inductances, and their low ESR. (I really like their silver/oil coupling caps, used them in both the amps we brought to your house that time).
The "standard" (for all the good builders I know) for caps is still motor runs, and to by-pass the last one. Solen fast caps if you're tight on space. No electrolytics.
Windows 7: I wanted simpler, now it's simpler. I'm a rock star.
robert taylor
L1 is below critical inductance, therefore this is still a cap input filter.
As I see it, L1 is seen as a small resistor...
> (L1/C1/L2/C2/L3/C3)
> Ah, so he is compensating for small L and C values by using lots of poles.
Especially quieter ones, as needed for sensitive speakers.
Matt Wiebe
--
robert taylor
Have you used one of these w/ sensitive speakers, something ~108-110dB? (Do they hum?)
r
From: mwi...@unm.edu
To: audio...@googlegroups.com
Subject: Low DCR
Date: Sun, 22 Nov 2009 07:03:02 -0700
Hotmail: Trusted email with Microsoft's powerful SPAM protection. Sign up now.
Matt Wiebe
robert taylor
I don't know how to "see" how quiet B+ is...
Yes, in PSUD you can see more ripple, (than the other supply described) but I don't know how to correlate how much more audible/objectionable this would be.
I dont EVEN want to go into the mud-slinging "debate" on the asylum;
I'm not trying to say just because it shows more ripple it can't work (like so many have), I don't know.
Just looking for another report, if anyone else tried this with higher sensitivity drivers.
(Jeffrey had told me he tried it, that it worked w/ lower efficiency (<100dB) systems, but hummed too much at listening distance for more sensitive drivers).
Date: Tue, 24 Nov 2009 07:17:18 -0700
Matt Wiebe
robert taylor
Low DCR PSU has a noticable amount more ripple.
I can't hear the difference.
I see the difference as more ripple in a sim.
I really do need to break down and build another PSU.
I bought the parts months ago...
so many projects, so little time...
I have five amp projects now in pieces, waiting to be put together...
And two speaker projects... :(
Matt Wiebe
jsn
supply would be a good thing. I don't have any idea what the DCR of
my existing supplies is.
Is there a target DCR figure for this? Does the power transformer DCR
factor in significantly? Are these typically choke-input, or does
choke or cap input not matter?
jsn
robert taylor
I was definetialy going to try this on a little 6N1P > parellel 1626;
have parts to build PSUs both ways, and swap out--e.perimental toy.
Was thinking about a more high end amp, with an Aa interstaged coupled to a 4-65.
Have to build the little one & see how it goes first I guess.
robert taylor
He (Dennis & Jeff) call it choke input only, but it's below critical inductance, so the engineer types say it's not...
They are pretty good amps, but I really thinkJeffrey's are noticably better...
And also Low DCR, and even lower cap values, but L1 is @ critical inductance.
Just not built to Dennis L SES rules
from an earlier post:
Here's the goodies:
(1) The Power Transformer can benefit a little bit by being low-DCR-- but the largest
benefit will come from extra current capability, regardless of DCR. about 150 mA as a minimum.
(2) The rectifier doesn't matter at all-- as to DCR. It is super-active, and can address current/voltage changes rather instantaneously. BUT-- just as in the transformer-- extra current capability here is a winner, regardless of DCR.
(3) The input choke should be low-DCR. This is because we want a TIGHT COUPLE to the rectifier output. Once the power leaves the rectifier, it is no longer adjustable-- actively. Now, we have passive components, unless we install voltage/current regulation-- but we don't want to listen to those things struggling to "keep-up" with music.... so they're best left out.
(4) Capacitors should be low-capacitance. This is because we want any "tuned" circuits to stay above human hearing--- by several orders. You can say here that we want SPEED. That simplifies it.
(5) Rules-of-thumb in designing the total amplifier:
(A) High-Current power transformer.
(B) High-Transparency rectifier.
(C) CHOKE-INPUT.
(D) low-DCR chokes.
(E) SMALL capacitors.
(F) Low-DCR wiring-- everywhere.
(G) Anything that is in-series with power or signal-- low-DCR, and physically short.
(H) Anything that LOADS power--or signal--(plate & cathode resistors, input impedance, etc.),--- VERY high DCR.
----D----
Ground wire needs to be a minimum of 10 or 12 gauge ROMEX, with one stand of Kimber TCSS litz in parallel to it in
You want chokes that are 10 ohms DCR, or less, and with a small core, and 1/2 HY or lower in inductance.
An example of this, a two stage amp, built w/ separate supplies, a 76 into a 245:
(L1/C1/L2/C2/L3/C3)
· Signal 500 VA PT and a 5U4GB hybred Graetz full-wave bridge rectifier,
· L1 is Triad C-56U, 35 mHY @ .75 ohms DCR @ 2A.,
· C1 is a 10 uF round oil,
· L2 was a Triad C-40X, .32 HY @ 10 ohms DCR @ 600 mA.,
· C2 was a 40 uF ASC oil from Michael Percy,
· L3 is another Triad C-40X, and
· C3 is another Percy 40 uF ASC.
· Oh, one other BIG BIG BIG improvement came when we significantly dropped the OP-POINT on the 245s so that they more closely reflected a Golden Ratio percentage ( 62% ) of the Manufacturer's suggestions.
- - - - - - - - - - - - - - - - - - -
This is not what Jeffrey uses.
He says there is too much ripple (and audible hum with sensitive speakers, like the 109dB horns he uses).
He uses L1 calculated to critical inductance.
L2 is about half that. L3 about half L2, and a high quality, nickle piece.
Caps are smaller than what Dennis and Jeff Medwin use, and the last one bypassed w/ a white Solen tin foil.
--Usually uses MV rectifiers too.
These are much better sounding amps IMO, but costlier.
http://www.jeffreywjackson.com/products.htm click picts for links for a few details.
If you're really interested, I've followed the thread for quite a while, and have saved details. Yeah, Jeff does get pretty specific, and he's even selling transformers & chokes now.
I'll try to attach picts, see if it works.
guess I ought to write him & see what he wants for them, try some out on that little 12B4...
Or maybe build a PP 6C45, I finally scored some input trans...
> Subject: Re: Low DCR
> From: j...@boozhoundlabs.com
> To: audio...@googlegroups.com
jsn
the concepts seem solid. Not sure about that whole "Golden Ratio of
the Manufacturer's Specs" though! :)
jsn
robert taylor
He is...
You should meet the guy, real fun to talk to.
or "seriously," (his business is called "serious stereo," he sells little DC monobloc 2A3 amps for $18k) you should call him up, & talk to him some time. Just leave a message, & he'll call back--& talk your leg off.
Quite entertaining, can go from sound multi-discipline engineering, to "what planet are you from" stuff and back in the same paragraph.
Always end up scratching my head after talking to Dennis...
Yeah, I think it's more mojo, & lucky stuff he stumbled on, but that's cool.
Anyway, he's kinda into the golden ratio, incorporates it in all kinds of stuff...
Even has caps wound in that length to dia proportion, says they sound better.
Was telling me about all these test he ran showing how much energy was given off the voice coil as sound vs heat, then insisted I come up to Livingston Montana to listen to everything to verify...
Uh, yeah...
I can afford that...
> Date: Tue, 24 Nov 2009 11:31:14 -0700
> the concepts seem solid. Not sure about that whole "Golden Ratio of
> the Manufacturer's Specs" though! :)
>
> jsn
> >
Matt Wiebe
There is more then enough to read about on audioasylum. In brief it
ends up being less then 15ohms from B+ to ground through the power
trans secondary. Choke input.
Matt
> .
>
>