Carlos' snubberized Gainclone Power supply

[jackinnj]

Jan:
Is that a Shakti apple on the power supply ? My article on using aluminum-foil wrapped bricks on top of power line was rejected.

It was a "quickie" experiment -- there are times of day (like 5 in the morning when neither heat, air conditioners, or microwave ovens etc. are radiating) when you can see the periodicity -- but we really have crappy power here in New Jersey -- not only is it unreliable but there are really untextbooklike harmonics. I first became aware of the problem when I started doing serious black and white photography...gotta crank up the real spectrum analyzer.

 

[peranders]

Surely you are not talking about my schematic, because I don't use caps across diodes.

Yes I did and if you check closely you have written 100 nF where you mean uF. This did make me a bit confused.... EDIT: I _was_ right.

To be clear:

10000 uF // 100nF(C5) // 100 nF(C7)+1R0 // 100 uF // 100 nF (the two latest at the IC pins)

Ok, we are back to square one. Why have 100 nF(C5)// 100nF(C7)+1R0?

Schematic here.

 

[Greg Erskine]

Occam, a memebr of another forum kindly gave me these two links. I found them very informative and worthwhile reading.

www.calex.com/pdf/3power_impedance.pdf

www.designers-guide.com/Design/bypassing.pdf

Peranders,

Did you get a chance to read these to articles? I think they answer all the questions in this thread simple enough for even me to understand. 😉

And I believe in my GC a 100nF+1R||10,000uF makes a significant difference. The articles also explain why with careful selection of caps the 100nF+1R may not be necessary as mentioned by Jan.

So I guess like most audio arguments we are ALL right.

 

[ThorstenL]

Konnichwa,

But isn't it more interesting to have the whole circuit?

MAYBE. But you asked a specific question, namely:

Looking at the schematic above Carlos have a 100 nF in parallell of 1R+ 100nF. Does this make any sense? What would the 1 ohms resistor do? Why not 10 ohms or 2.2 ohms?

The simulations I have done illustrate the answer to this question.

Had you asked for a complete analysis of a given PSU Circuit, including all parts, parasistcs, mains parasitics et al. I might have answered:

"But Kuei Yang Wang declared that would never agree
With the plans he had made for the day: -
Simulation was always a difficult art,
Though with only one capacitor and one bypass:
And he feared he must really decline, for his part,
Undertaking another as well. - "

(With apologies to Lewis Caroll - "The hunting of the Snark")

Sayonara

 

[ThorstenL]

Konnichiwa,

Ok, we are back to square one. Why have 100 nF(C5)// 100nF(C7)+1R0?

Schematic here.

Whereby you have just neatly intiated an infinite loop.

You forgot to insert the following:

If Bored = True OR Ignorance = Enlightened Then
Exit Loop
Else GoTo Post #1

Instead of actually answering I direct you to my first answer, from where you may follow the whole lot again (saving me the time to answer again....

http://www.diyaudio.com/forums/showthread.php?postid=545272#post545272

Sayonara

 

[carlosfm]

Do while (refused to test + unable to read a schematic)
open threads in anguer;
post one thing on one forum, another thing on another;
Enddo;

Result:
Infinite loop😕

 

[jam]

I take it as Peranders chasing his tail again.................😀 😀 😀


http://www.flashfoundry.com/details/no-491-swf-0/

Jam

P.S. Watch it for a few minutes and see what I mean.

 

[Jaka Racman]

Originally posted by Kuei Yang Wang

Could you suggest which specific capacitors offer as low an ESL and how you took the measurements?

Hi,

here is Hitachi and here is Epcos (page2 of the PDF). I never bothered to check lower voltage cans, but according to my measurements modern electrolytics are usually much better than large polyproplyne film caps (WIMA FKP1 for instance). All measurements were done with HP4192A impedance analyzer at 10 Mhz.

Best regards,

Jaka Racman

 

[peranders]

There is much talk about simulation and no talk at all about real measurements. How does the simulations correspond to the real world? Not interesting, I'll suppose.

 

[demogorgon]

Offcource you'll suppose P-A, you are never going to accsept this snubber thingy as a positive improvement to the GC design simply because you dont want to. noone can tell you differently, because you have made up your mind, so why bother?

 

[ThorstenL]

Konnichiwa,

here is Hitachi and here is Epcos (page2 of the PDF).

Looks good on paper. Measurements of various sources of more "normal" electrlyics suggest that these are more the exception than the rule.

I never bothered to check lower voltage cans, but according to my measurements modern electrolytics are usually much better than large polyproplyne film caps (WIMA FKP1 for instance).

Hmmm. Funny. That wasn't the case all that long ago, have suddenly all electrlytic capacitor manufacturers changed their processes? Try measuing a generic ultra-cheap Jamicon, Samwa or such a "brand", 10,000uF/50V type as you would find in the kind of supply being discussed here....

Sayonara

 

[peranders]

noone can tell you differently, because you have made up your mind, so why bother?

Yes, they can if I see some proof.

 

[Upupa Epops]

My conclusion of all this debate is, that LM's ( or chips amps in general ) is extremly sensitive to hf disturbing. It's probably caused by nonfiltered supply voltage on input and VAS . So we must wait for chip, which will have these stages separated or go back to discrete amps, where isn't this problem . 😉

 

[ThorstenL]

Konnichiwa,

My conclusion of all this debate is, that LM's ( or chips amps in general ) is extremly sensitive to hf disturbing.

This is rather true, any Op-Amp (not just super fast types) and ANY DISCRETE OP-AMP is extremely sensitive to RF and impedance probelms on the raisl, due to a virtually zero PSRR at high frequencies.

It's probably caused by nonfiltered supply voltage on input and VAS .

The single ended VAS and the compensation that is looped around it are solely responsile.

So we must wait for chip, which will have these stages separated or go back to discrete amps, where isn't this problem .

Well, we could simply design good powersupplies, now could we. Nah, that's too much work, certainly for PA.... ;-)

Sayonara

 

[Upupa Epops]

Good PS ? From this look is only good acid or alcalic battery - no hum, no noise, no snubber ... 😀

 

[hjelm]

Is the debate if the increasing of the ESR helps to avoid ringing or if the chipamps are sensitive to HF garbage on the supply.

Just to verify that the ringning is there i rigged a measurement and yes it was there. I did purposly have a long wire between a 100uF electrolyte, nothing fancy, and a 0.1uF ceramic cap but this was just to see the effect, it should be there at a higher frequency if i shorten the wire. Yes there was ringing at 1.7 MHz and the pp voltage was 0.2 V. I brutally shorted the load for a short instance to trigger the ringing. To take the supply out of the equation i had a 10k resistor in series with the supply.

A one ohm carbon resistor in series with the ceramic cap took care of the ringing totally and i can hardly see any ripple on the voltage.

Now the question i have is, what could trigger a ringing or is it just the regulation at HF that is bad and we have garbage leaking in to the system. If i have a digital system it is obvious that the load will be pulsed but a chipamp does it consume current way outside the audio band? Does a class (a)b stage cause high frequency pulses in the load that could cause ringing in the supply?

 

[jan.didden]

Hjelm,

IMO the ringing in such cases is the result of a step change in current in the system. You did that by the short and subsequent release. Because of capacitance and inductance in the wiring (and components), the fast delta-I/delta-T causes the ernergy stored in the system to die out exponentially while oscillating between charge in the cap and current in the L.
That is the (damped) ringing you see. But the important point is that it is totally dependent on the current slew rate, the mentioned delta-I/delta-T. With bandlimited signals like in audio systems, this will not be seen in your setup.

Jan Didden

 

[ThorstenL]

Konnichiwa,

Good PS ? From this look is only good acid or alcalic battery - no hum, no noise, no snubber ... 😀

Batteries do not allow bypassing and/or snubbers to be eliminated, they have too much inductance....



Sayonara

 

[ThorstenL]

Konnichiwa,

Is the debate if the increasing of the ESR helps to avoid ringing or if the chipamps are sensitive to HF garbage on the supply.

Neither of course. PA and CFM have interpersonal issues, PA seems to be unable to take criticism of his design approaches and seems to be the kind of "Hardcore Pseudo Objectivist" who refuses to accept all but most basic design techniques.

His objection that "R & C in series CANNOT DO ANYTHING" was the start of the thread. He made this objection without analysing the issues (which I eventually had to do for him) and since has dodgedly refused to aknowledge his fundamental error in considering the facts.

Now the question i have is, what could trigger a ringing

A signal applied to the amplifier fed from the supply? And remember, the ringing is merely a symptom, the actual problem is a high and highly reactive PSU Impedance at certain high frequencies, which wreak havoc with a looped feedback amplifiers stability margin.

Add a speaker cable with a bit of capacitance and a speaker with a good deal of reactance and nice RFI noisepickup via the various antennae (oops, make that speaker cables - yes they are nice aerials) and you have plenty of potential to reach a state where the assembly of PSU and Amp becomes instable on signal peaks, despite looking fine on the testbench, even with squarewaves.

Sayonara

 

[Upupa Epops]

Inductance ? Is this problem in motor starter ? Or is this inductance bigger that inductance of transformer ? 😉