So no huge buss bars....?
Heh. Actually, maybe. If it was too difficult to have the cap array board go right across each pair of decoupling points, then maybe buss bars should be used.
Last edited:
One of the most interesting things is to place an oscilloscope onto the power supply terminals of chips and see how effective the PSU is . As Crystal pointed out years ago the quality and placement of the decoupling capacitor is important ( shortest possible wires ) . I often add between 10 nF to 100 nF to chips . I always thought it did some good . Sometimes no amount of PSU upgrades will compensate for bad decoupling . I suspect most commercial products are less than ideal in this respect . The reason being they don't allow for hand soldering espeically with SMD ( SMD in itself is great ) . The old 100 nF to serve 10 chips is still the norm in some designs . Some use very low grade ceramic caps as someone said it would be OK in a book ( yes they are up to a point ) .COG/NPO works better ( almost as good as silver mica and very cheap ) . I do break rails and add inductors sometimes . Do use a scope if you do. Often " better " is far worse .
An even more illuminating experience is to drive an audio system with the "signal" in the power supply. Using a capacitor to decouple, the power supply becomes an audio input, and what a glorious cacophany of "music" can be heard! Especially with a power amp under stress, you can "hear" the power supply regulation collapsing once you put the boot in ...
Frank
Rather than repeat myself, check out my responses to the chap who asked this question: http://www.diyaudio.com/forums/power-supplies/218862-mundorf-m-cap-supreme-bypass-cap.html. We covered a fair bit of ground on the ins and outs of things, satisfactorily I believe ...
Frank
just watched an 1979 video interview with conductor and composer Sergiu Celibidache (which btw was born in the city I live in). he basically bashed sound reproduction as unable to convey the intention of the composer (in more colorful vernacular). I translated a small portion to English:
The so-called technical advancement is a complete mutilation of the true connections between sound and movement. Sound is in movement. But you are projecting it in a completely different reference system. And it ends up not maintaining either of the human measures which gave it life. You and your new equipment... amplifiers and stereophonics, you create sonic values, sonic impressions but never the ones carrying the composer's idea, the way he heard it. Thus, by the nature of things, you are departing from the essence of music.
wonder what he would've said, had he heard a good stereo, and I'm not being sarcastic.
The so-called technical advancement is a complete mutilation of the true connections between sound and movement. Sound is in movement. But you are projecting it in a completely different reference system. And it ends up not maintaining either of the human measures which gave it life. You and your new equipment... amplifiers and stereophonics, you create sonic values, sonic impressions but never the ones carrying the composer's idea, the way he heard it. Thus, by the nature of things, you are departing from the essence of music.
wonder what he would've said, had he heard a good stereo, and I'm not being sarcastic.
Zmin - 1 ohm ...
Rather than repeat myself, check out my responses to the chap who asked this question: http://www.diyaudio.com/forums/power-supplies/218862-mundorf-m-cap-supreme-bypass-cap.html. We covered a fair bit of ground on the ins and outs of things, satisfactorily I believe ...
Frank
Thanks will give it a look over ....
Nigel,
I am just a beginner. But you might find this interesting:
http://www.diyaudio.com/forums/power-supplies/216409-power-supply-resevoir-size-169.html#post3320547
And this:
http://www.diyaudio.com/forums/atta...3-power-supply-resevoir-size-zoom3a_33kuf.jpg
Highest regards,
Tom Gootee
As I understood it from an admittedly brief lookover (no time now, will dive in later on), you got 53,050 uF per channel for a nominally 100W/8 Ohms amp?
Tom, as I see it, no amp ever made satisfied this criteria, or perhaps a few wildcard products did, but I never heard about them.
By such values, even in monetary terms, it's time to think about fully electronically regulated PSUs. These days, power devices are cheap, so the real penalty one has to pay is the added space requirements for the regulator heat sinks.
It would also be very interesting to compare two identical amps (in terms of the audio circuit), the only difference being that one is fed off capacitor filtered lines only, and the other used split supplies, with the input stage and VAS being fed off separate, slightly higher but fully electronically reguated lines.
This is a concept I have been using for well over two decades now. I adopted it after maing two very simple, low power (app. 30 W/8 Ohms) amps. The one with separate power supply lines was very clearly the better one, more focus, more definition.
Having said that, I must point out one very frequent caveat - even when people do use full regulation, i.e. electronically regulated power supply lines, but only one set of + and - for each channel, for resons I fail to understand they seem to think they can start skimping on the electrolytics preceeding the regulator. In my experience, this is wrong. You can safely SLIGHTLY reduce the cap size, from say 27,000 to say 22,000 uF, but not more as this will become audible,subtly at first and progressively more and more.
A friend, one Dan Banquer of Boston, who used to own a company named R. E. Designs, was (for family reasons, he is no longer in manufacturing) one of the few people who understood this. He used fully regulated power supply lines, developed his own regulator and so forth, but still used 33,000 uF caps for a nominally 80W/8 Ohms amplifier.
Any experience with that, Tom? I'd love to hear about it, please.
Its no longer just for family reasons he's no longer in manufacturing, medical reasons have taken precedence : Daniel Banquer Obituary: View Obituary for Daniel Banquer by Stanetsky-Hymanson Memorial Chapel, Salem, MA
In the reservoir size thread linked by gootee we established that required cap size depends on the amount you can allow the voltage rail to droop between charging pulses (assuming infinite PSRR in the audio circuits). The amount of droop which can be tolerated depends on how high you started, so more voltage from the transformer secondaries means you can use smaller caps and vice versa. Thus specifying X amount of cap with Y amount of output power is, by itself, mathematically meaningless. It only gains its apparent practical usefulness because we typically think in terms of a transformer off-load voltage which will allow us so many percent voltage droop before we get clipping on peaks.
This is not an attempt to transfer the discussion from that thread into here, but just a note for those who have not read that (rather long) thread. Perhaps given this taster, people will find the time to read the thread as it contains much useful information.
This is not an attempt to transfer the discussion from that thread into here, but just a note for those who have not read that (rather long) thread. Perhaps given this taster, people will find the time to read the thread as it contains much useful information.
Last edited:
Abraxalito, many thanks for the link. It is as I suspected, Dan and I exchanged mayn messages over the years, and then it suddenly stopped. I knew his wife was not well, but somehow, naively, I believed he was strong.
I am much saddened by these sorry news. Even if I hoped against hope it wasn't that bad. To me, this is a serious personal loss, Dan was so much more than just kind to me - he actually transferred all his manufacturing rights to me, just like that. He also passed on all his past and future projects to me, which I suspect may have been a way to insure they got made, even if not by him.
Well, it's up to me now. All I can do is to make what he didn't and name it after him, then put it in public domain as "The Dan Baquer Headphone Amplifier". And this I have to do so as to make Dan proud.
They don't make 'em like that any more.
Agreed.
But then, it doesn't have to be so, if one invests time and work and materials into it.
In my view, the key problem with regulated current stages is speed of delivery, i.e. rise and settling time of the regulator.
The input stage and VAS stages are infinitely easier to do, given that they don't require more than 20 ... 100 mA of current. I have seen it done with your plain vanilla 3 point regulators (7815, 7915) with zener diodes lifting the ground by whatever is required. This may surprise you, but that sounded quite all right, simple as it may be.
The added benefit is that one can lower the supply line voltages to the current stages (but still keeping them appropriately stiff), thus pusihing the output stage more to the left of their SOAR curve and reducing the number of variables we have to deal with.
Point taken.
Just one little note - ever since I bought my first transformer in 1975 (or '74? don't remember) I have ALWAYS ordered them custom made for me quoting full power on voltage. So, if I want say 56V, I ask for 40V taps at full load on.
The nominal tolerance from my transformer makers is +/ 5% for toroids up to 300 VA, and +/- 3% for 400 and above VA toroids. Furthermore, their toroids' leakeg current is not 17...20 mA as for British ILP, for example, but rather 12 mA or less, indicating quality cores and wiring. Two days ago, I measured a 500 VA toroid and it had 11 mA of leakage.
I know this can be reduced to below 5 mA, but for that, one would need a sintered core transformer, which is EXTREMELY hard to find, and when you (if you) do, you get slugged by the whopping price. Crazy prices, even if one knows a quality sintered core is capable of delivering around 2.4 times its nominal value without bad saturation. Economically, it's a better deal to simply do not use say 300 VA, but say 500 VA instead.
In a nutshell, AC designs using a fully complementary topology. N and P input differential pairs, followed by single transistors for the VAS. One driver, and two pairs of Motorola MJ 250W trannies.
In between, there's a two transistor circuit which does a floating bias, in an attempt to mimic pure class A operation.
All transistors strictly off-the-shelf MPS, MPSA, TIP and MJ 15xxx types.
Single rails for both halves, but the rails are electronically regulated to +/- 40.5 V. The regulators use some chips, but the actual output to the amp is performed by 250W Motorola Darlington transistors, one per each supply line. The regulator is preceeded by two 33,000 uF caps, I think they were by Mallory or Jensen, not sure which.
The sound was of the "vintage" type, meaning warm and soft, his objective was for the amp to deliver the Big Picture as it should, rather than being "analytical", "detailed in the mids and highs", "powerful bass", and the rest of that garbage. Mind you, he made no attampt to have one over another thing, quite to the contrary, he tried hard to make it come out all together now, in proper relationship.
If the program material was harsh, it would sound harsh, if it was smooth and sweet, the amp would follow. In other words, that amp had an adorable way of mostly disappearing as a singular factor.
And it had what I love to hear most in any amp, it somehow suggested that it had infinite power, as if you simply couldn't overload it anyhow. An illusion, of course, but quite a number of American amps pull this off, and I daresay nobody pulls it off like the Yanks. Well, maybe Naim a little bit.
- Status
- Not open for further replies.