I met both Jeff and Dennis at RMAF. Both seemed to be very pleasant gentlemen. In fact, I also got to hear Dennis' 2A3 amps. Probably the best I've come across so far.
Dennis is a little on the whacky side when he's trying to explain circuits. Let's just say he takes a little different approach than most of us (you should ask him about the relationship between cows and amps). Hey, you can do math all day, but in the end it's the results that count.
Anyway, what might come across as gruff, intolerant, or arrogant in one of these forums has a entirely different appeal is person.
jh
PS - I don't do the low dcr thing. I design for critical stability, which requires an optimum and finite value for dcr.
Dennis is a little on the whacky side when he's trying to explain circuits. Let's just say he takes a little different approach than most of us (you should ask him about the relationship between cows and amps). Hey, you can do math all day, but in the end it's the results that count.
Anyway, what might come across as gruff, intolerant, or arrogant in one of these forums has a entirely different appeal is person.
jh
PS - I don't do the low dcr thing. I design for critical stability, which requires an optimum and finite value for dcr.
IMHO, while it is may be possible to make a low DCR, low inductance and low capacitance PSU work, that doesn't mean it's a good idea. I suspect it needs a lot of tuning and would probably be quite critical. If any of the amplifier design conditions were to change, like the voltage or current draw, then the PSU would probably have to be re-tuned. Sounds like a lot of hassle and grief for no obvious tangible benefit.
When did that ever stop a true believer? Part of why I think these supplies are apparently working ok is that they are going into class A amps that are seldom overloaded, few supply transients.
A smaller cap may drain its (smaller) charge faster, but its the caps impedance we care about, is it not? If a bigger cap happens to manage having both lower ESR and ESL, should it not be 'better'?
Check back in at AA. Henry Pasternak, the approach's most vocal and technically capable critic, has returned under various pseudonyms, now in agreement with its validity and trying to work out the underlying theory. It appears to be a viable and valid PS topology. Don't shoot the baby with the messenger!
I daresay it is but I also get the impression, from various people's contributions to the discussion, that it needs to be optimised. As I said before, I fail to see any obvious benefit of the approach to justify either the hassle of making it work or the thousands of words that were spent arguing about it on AA.
When Jeff Medwin only preached 'low DCR', most people agreed with him, myself included. It was patently obvious that excessive DCR in iron components was detrimental to regulation. He was preaching to the converted and that should have been the end of the argument but it didn't seem to satisfy Jeff. He had to have his say anyway and he rubbed a few people up the wrong way in the process with his overbearing attitude, including Henry Pasternak.
Later, when Jeff got onto the 'low inductance, low capacitance' kick, he had some difficulty selling it, perhaps because it wasn't his original idea. It seems to have come from Dennis Fraker and Jeff found he couldn't defend it convincingly. Dennis's contributions didn't help, because he was patronising and talked in riddles.
Those guys were pretty goofy, but I personally wouldn't go as far as saying their system was wonderful, although it is very difficult to tell under show circumstances. The best sounds I heard at the show were of conventional design but highly developed, usually driving speakers that were way out of my price range.
The problem with Medwin on AA is he writes as though these amps are a second coming and anyone who won't agree has bad hearing. There were a lot of rooms at that show that if anyone thought they were any good, I'd question their hearing.
I personally use the lowest DCR power transformers and chokes I can reasonably get and agree with his philosophy in that regard. But that rant about the G2 supply and that big high current choke being able to provide good regulation was bizarre.
John
That could be, I never got to hear their room. My exposure to the amps was after-hours in a shootout (in the Galibier room). Even up against my own 2A3 design (prototype). A new Quicksilver design also fared well. This was also my chance to finally hear a top-end digital amp. Oh my, it was like AM radio in comparison. Really bad.
jh
To be fair, the findings of Henry Paternack are not directly related to low DCR. Though the components used do have an inherently low DCR I'm not sure if that is or is not important.
Henry has been looking deeply into first LC sections composed of a low value input choke followed by a low value capacitor. We're talking about in the neighborhood of a couple hundred millihenry and a couple of microfarad. What Henry found is that it is possible to achieve very good regulation with a filter section like this; MUCH better than a simple C-input filter and perhaps as good as or better than an L-input filter that meets critical inductance.
In a nutshell, what he found is that the ripple voltage across the cap is not limited to the peak output voltage of the power transformer as is the case with an ordinary C-input filter. Instead, the small input L allow the average ripple voltage to float around a fairly constant level even with varying current draw. The magnitude of the ripple increases with current draw, but the average value remains relatively constant. This is in contrast to a conventional C-input filter where the average value must necessarily drop as ripple magnitude increases; the maximum voltage is limited to the peak output voltage of the transformer minus and diode drop.
Anyone interested should read Henry's posts themselves.
Again, although his motivation to explore this circuit was undoubtedly spawned by the ranting of Jeff Medwin and Dennis Fraker, I don't see his findings as being confirmation of their ideas. Jeff in particular has been pushing low DCR, but that is not an important element of Henry's results.
-- Dave
Henry has been looking deeply into first LC sections composed of a low value input choke followed by a low value capacitor. We're talking about in the neighborhood of a couple hundred millihenry and a couple of microfarad. What Henry found is that it is possible to achieve very good regulation with a filter section like this; MUCH better than a simple C-input filter and perhaps as good as or better than an L-input filter that meets critical inductance.
In a nutshell, what he found is that the ripple voltage across the cap is not limited to the peak output voltage of the power transformer as is the case with an ordinary C-input filter. Instead, the small input L allow the average ripple voltage to float around a fairly constant level even with varying current draw. The magnitude of the ripple increases with current draw, but the average value remains relatively constant. This is in contrast to a conventional C-input filter where the average value must necessarily drop as ripple magnitude increases; the maximum voltage is limited to the peak output voltage of the transformer minus and diode drop.
Anyone interested should read Henry's posts themselves.
Again, although his motivation to explore this circuit was undoubtedly spawned by the ranting of Jeff Medwin and Dennis Fraker, I don't see his findings as being confirmation of their ideas. Jeff in particular has been pushing low DCR, but that is not an important element of Henry's results.
-- Dave
All I see are some simulations done on PSUD which mean little or nothing. Apparently he never actually answered the question of whether he had tried the design out. Furthermore, he seems to be persona non grata on that forum. What's that all about?
There have been some strange ideas about power supplies floating around out there due to the idiosyncratic nature of that program.
John
The strange ideas came first, the sims after. The ideas evolved over the period of discussion from low L to low L&C. Henry's hard science textbook approach crashed into Jeff's 'ears only' methodology and it wasn't like chocolate and peanut butter. Henry just gave up for a while because there literally appears to be no reasoning with Jeff about the underlying principles. To his credit once he saw it had merit his intellectual curiousity brought him back for a short while to discuss. If you follow rec.audio.tubes this isn't uncommon behaviour for Henry, talented though he is.
BTW Dave, I won't quibble with your historical outline except to say it gives too much credit to Henry for where they're at. Regarding low DCR though it's essential to maintaining what they're currently considering a tank circuit between transformer, rectifiers, first L and first C. Raise the transformer secondary winding impedance and regulation collapses. In my sims tuning also appears to be more critical with higher R, an unintuitive result.
We'll see how the sims hold up. I have a 200 Volt torroid on the way for a little 6CW5 SE amp. Always wondered how I was going to use that .07 H 3.0 ohm choke laying around. Incidentally, it doesn't seem to have dawned on the AA guys yet but with the values they're talking a common mode power choke would work as the first L. If this topology works it's the perfect solution: extremely wide bandwidth, high saturation, MHz self-resonance, cheap as chips and filters the ground leg at the same time.
BTW Dave, I won't quibble with your historical outline except to say it gives too much credit to Henry for where they're at. Regarding low DCR though it's essential to maintaining what they're currently considering a tank circuit between transformer, rectifiers, first L and first C. Raise the transformer secondary winding impedance and regulation collapses. In my sims tuning also appears to be more critical with higher R, an unintuitive result.
We'll see how the sims hold up. I have a 200 Volt torroid on the way for a little 6CW5 SE amp. Always wondered how I was going to use that .07 H 3.0 ohm choke laying around. Incidentally, it doesn't seem to have dawned on the AA guys yet but with the values they're talking a common mode power choke would work as the first L. If this topology works it's the perfect solution: extremely wide bandwidth, high saturation, MHz self-resonance, cheap as chips and filters the ground leg at the same time.
The only gross filtering you will get with one of the input common mode chokes is from the leakage inductance between the two choke windings. This is enhanced somewhat in a lot of CM chokes by using multisection bobbins for the windings. This also reduces the intrawinding capacitance and raises the self-resonant frequency. A leakage inductance of between 2-5% of the common mode inductance is par for the course, though it's easy to measure if you have an LCR meter. The combinatiion of relatively high differential mode inductance and high common mode inductance makes it possible to pare an inductor off for an input EMI filter and take out some cost.
Three sentences, and three factual errors.
If you read all of the postings, you will find an objective technical analysis, with both simulation results and discussion, of a particular aspect of a particular power supply design. Nothing less, nothing more.
If you find the results interesting, feel free to use them. If not, then disregard them.
-Henry
I have moved on from my initial interpretation of the input section as a resonant tank; see my follow-up posting on the Asylum.
The input choke provides some energy storage that delays the peak of the input current waveform relative to the secondary voltage waveform. This is what allows the voltage on the input cap to exceed the peak secondary voltage.
Presumably, secondary resistance will bleed off some of the energy stored in the inductor, reducing the beneficial effect. To compensate, you need a larger inductor. Experimenting with the simulation, this constraint makes it harder to get optimal tuning out of the circuit. To me, this is also an intuitive result.
It occurs to me that you might be able to interpret the effect of the input choke as a power factor correction for the interface between the rectifier and the filter.
Rec.audio.tubes is ruined, a sad loss for the community. Can you say anything good about it?
-Henry
Old usenet just aint what she used to be; aint what she used to be; aint what she used to be.- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.