[ppl]

I myself did not take offence at tomo. like aos, I have known him for years and I am familiar with his ways, my god aos has it been over 4 years! I in no way claimed that a capacitance multiplier was an esoteric concept. This is why I suggested he search Since he did not understand my first explanation, I knew from past experience with him that he probably will not understand any future replies from me. I felt that further explanation would be futile and since I knew that lots of information on capacitance multipliers are out there on the web, I did not think it unreasonable for Tomo to seek out and find Explanations more suitable to his liking. Tomo and other old timers should know it is extremely difficult for me to communicate and I am sorry if my command of the English language is not good enough for you. In my opinion regardless of the fact that my previous explanation was unsatisfactory I believe I have Broadened tomo's horizon and he is now introduced to the concept of capacitance multipliers. Granted he had to do a little research to find answers acceptable to him, however This is a good thing I am sure. Please do not take this post as an offence because I meant no insult to anyone and I was not insulted by tomo because I understood his demanding ways. My intent in this post is to hopefully point out to people that I have limitations just lake any other mortal and have gone thought hell and back with health issues that not only make me overly sensitive to the language of others but also have impacted my thinking process a lot. I admit there are some people that flat out irritate me, this I am sure is because of ether a misunderstanding of what I say or vise versa. tomo is not in that group. the ones that are I no longer respond to. not because I like to leave them hanging, but because I do not know or understand there ways. Some people come off real offensive to my technical writings at times and since I do not understand there sensibilities and they I am sure do not understand mine, I would just as soon avoid the drama.

well there you have my 2 cents worth Peace.

 

[Voodoochile]

Rick, I think you and Carlo now need to start a dedicated warm and fuzzy group hug thread, to inversely compliment the other thread.

Nicely phrased, Phillip!

 

[ppl]

ya i heard about that on the News that the old man on the Mountain has most of his face just fall off. just proves Nothing is everlasting. not even Rock

 

[Voodoochile]

Just opened up the June issue of Audio Xpress, and though I have not read it through yet, there is an article on page 30 of interest. It's "An SE Class A Headphone Amp" by Aren van Waarde. It's a simple amp, a 2SC1327 followed by a Darlington made up of a 2SD1138 and 2SD844.

Anyway, fast forward a second...
for initial testing, Aren applied his bench supply (he gave no details on the spec), and there was no appreciable noise, but there was some hum- independent of the volume control. Must be quite the bench supply! Anyway, he started hunting for a good supply, and ended up at Rod Elliot's place, looking at the cap multiplier supply, as well as the site of Geoff Moss. He is really pleased with the results, it seems. He did make a couple of accomodations for parts substitution of a few bits he did not have, but nothing to change the real character of the PS.

I know Elliot's site was reference a while back with regard to the cap multiplier type of PS as well, but thought this to be interesting. Many of you also get AX... check it out.

Quote:

Nothing is everlasting...

You have got that right. What a bummer.

 

[ppl]

thanks for the AE tip should be an interesting read. the circuit on Rod's site has been up for quite some time and i would have thought most people would have Known about that one at least. Capacitence multipliers have been around since the Tube days. however thay were not used in power supplys alot in that era. thay were mostly used as sample and hold circuits. even in early solid state did not use capacitence multipliers alot except as sample and hold applications using op amps. Recently thay are becoming quite popular as Power supplys, for this application thay offer several advantages over Conventional regulators one being when you just want to get rid of noise and have great transient responce, and have no need for Voltage regulation or the voltage regulation is provided in a preceeding stage. one great advantage of the capacitence Multiplier is unlike voltage regulators the input voltage can varry alot and the circuit will still perform great as you have no dropout voltage to Worry about as you do with a typical voltage regulator. those long timers hear i am sure have seen some of my old Portable headphone amp schematics where i used a capacitence multiplier on the Op amp rails internaly. the reason for usinging it there was to provide a low output impedance for the op amps power supply, isolate the op amp rails from the output rails and not have to worry about dropout voltage issues as the batteries were depleated.

 

[Voodoochile]

Gaahhh... I was going to put up some links, too.
Geoff Moss
Rod Elliott
Much of Geoff's is derived from Rod's, but he dialog is different.

 

[rickcr42]

tomo ,aos ,ppl

we a ALL go way back

nothing but a thing among buds

and tomo is one of the first active posters back in the day of headwize and he is one of my first online conversationals

my first "who yoo messin with dude"

'cause I was touchy resulted in MY apology

and no , he does NOT mean it as it sounds


one of the real good guys

mess with tomo we will setup up a rick/tomo vs the world

 

[jeffreyj]

Quote:

Originally posted by morsel We are working on a capacitance multiplier power supply for the PPA. Here are a couple of links for reference: http://sound.westhost.com/project15.htm http://www.gmweb.btinternet.co.uk/jlhcapmult.htm There are 3 versions: single, dual, and single with dual darlingtons. Two current sources and a zener have replaced a resistor or pi filter from the original circuit to limit the maximum output voltage. The dual darlington versions use complimentary pairs for better negative rail performance. The bridge rectifiers are bypassed with capacitors for reduced noise.

Here are the problems with this circuit, w/r/t your application, as I see them:

1) the Darlington configuration, and the Szlikai, are both extremely slow. The minority charge carriers in the driven transistor cannot be swept away by the driver transistor so much of the advantage of a capacitance multiplier , i.e. - increased stiffness of the emitter voltage - is tossed out the window because of storage time.

2) Such configurations prevent the driven transistor from saturating, and therefore result in at least a 2 diode drop loss of supply voltage.

3) Zener diodes are noisy and have terrible dynamic resistance even when driven with current regulator diodes.

4) Speaking of... you have too many current regulator diodes to begin with. The ones to the left of the zener are somewhat better than a resistor, though in a capacitance multiplier there is some argument against them, while the one's post-zener are inexplicable and/or redundant. The capacitor in a c.m. should go directly between base and ground for maximum effect.

5) Why the Sziklai configuration in the ground return of the single supply? To preserve circuit board uniformity across models? You are losing a diode drop in the ground return for whatever reason.

6) Where does leakage from the driver transistor go? Why, into the base of the driven transistor! You need a resistor between base and emitter of the driven transistor for each leg to drain off leakage current and help sweep some of those minority carriers out of the base junction when the driven transistor wants to turn off (or, more off, if that's even remotely grammatically correct).

Finally, and this one gets no number, c.m.'s are no substitute for a good linear regulator with feedforward (compensation against input level variations, that is). In my experience, an LM317 regulator with it's ADJ pin bypassed with a small capacitor (say 0.1uF) makes about the quietest, quickest transient response regulator around.

 

[aos]

Zeners are indeed noisy and not particularly stable and should be replaced with modern references, or with LEDs if possible. Darlingtons can indeed be slow, though I thought it mostly relates to monolithic Darlington transistors, not the discrete Darlingtons. But LM317 is not the best regulator around. TPS79101 for example is much faster and much quieter. And there are many others.

I really miss old times, when there used to be several enthusiastic people who'd do a lot of Spice simulations whenever there was a suggestion to change something.

 

[Tomo]

Ok,

So we know the problems lets try to solve them.

Remember Elliot's site doesn't have enough stuffs. I would go GOOGLE and find more links. You can try TNT-Audio.

Yeah, get some pspice results. Me? I am doing simulations. It's my true job actually, simu. computer jockie, another name for computer lab rat. My boss wants data ASAP or I am toast.

So I am sorry. I don't so simulations for ya. Barely got time for my own ...

T

 

[Tomo]

Jeff -

1) So don't use Darlington. Elliot is using darlington because he needs a large gain. Note that with Elliot's setup, he measure 1mV ripple. Rather large. It doesn't do good enough go solo.

2) Elliot has predicted that himself.

Quote:

We must assume that at least 3 Volts will be lost across the capacitance-multiplier filter, to ensure that the DC input (including ripple component) always exceeds the output voltage.

3) Frankly, I am not sure. There are many people don't mind Zener setup. I am starting doubt it has harmful effect. But then I don't mind using LED's. They make a nice voltage reference according to my experience.

4) Yeah, I must agree a little too many things dangling off of base. Couldn't we start with something simple? After all, it is easier to debug simpler systems especially when the system is un-tested. Besides, 4 current source per channel is rather ... complex.

5) The diode bypass is to prevent the current flow into the output of c.m. Kinda strange looking. But I think both them diodes are reverse biased. ...

6) Ooo nice point. Current musta go somewhere. But then does it matter? Do you have data? I don't believe in simple theory. I like data a whole a lot better.

Elliot states that c.m. is not a regulator.

Quote:

Because there is no regulation, the power amplifier must be capable of accepting the voltage variations from the mains - every standard power amplifier in existence does this quite happily now, so it is obviously not a problem. Note that the output power is affected, but this happens with all amps, and cannot be avoided without a regulator.

In terms of LM317, I have no real complaint. It worked nicely for me for bunch of years. But then it would be wrong to say "the quietest, quickest transient response regulator." There are better. You probably mean, "the most suitable, most cost-effective yet high performance regulator." Again, "is it worth it?' is a difficult question to answer.

Tomo

 

[jeffreyj]

Quote:

Originally posted by aos Zeners are indeed noisy and not particularly stable and should be replaced with modern references, or with LEDs if possible. Darlingtons can indeed be slow, though I thought it mostly relates to monolithic Darlington transistors, not the discrete Darlingtons. But LM317 is not the best regulator around. TPS79101 for example is much faster and much quieter. And there are many others.

Monolithic or not, if something isn't done about the driver's leakage current then disaster awaits when said leakage brings the driven transistor into conduction! A resistor from the driven transistor's emitter to base helps shunt the leakage (producing an error at the emitter, mind you - tanstaafl) with the added benefit of helping remove current from the driven transistor's base when it needs to shutoff.

Mea culpa: I should not have written that the LM317 is the best, rather, that is the best I have had experience with. I've not used too many of the newer, and better looking spec-wise, linear regulators mainly because I have learned the hard way that not using common/industry standard parts can bring about serious sourcing miseries later on in a product's life when that esoteric part is no longer available

. Still, I won't argue at all that there aren't discrete and monolithic designs out there that'll kick the LM317's proverbial posterior. But, are they necessary in an amp that might demand a whopping 50mA peak??? Not hardly.


Quote:

I really miss old times, when there used to be several enthusiastic people who'd do a lot of Spice simulations whenever there was a suggestion to change something.

The incentive to do such is inversely proportional to the potential for commercial application of the effort.

BTW - ever tried to design an oscillator in some of those early spice implentations... almost never worked because not enough parasitics were modeled in (everything was too "perfect"). I keep that in mind whenever I fire up a simulation program.

 

[Tomo]

Nope,

The one I got only does first order calculations. I don't think it knows the phrase "second order."

T

 

[jeffreyj]

Quote:

Originally posted by Tomo Jeff - ... 3) Frankly, I am not sure. There are many people don't mind Zener setup. I am starting doubt it has harmful effect. But then I don't mind using LED's. They make a nice voltage reference according to my experience.

There are also people out there who think putting a rock on top of their solid state amplifier will help "reduce resonance effects" (i.e. - Jonathan Scull of Stereophile)

. But seriously, a buried zener makes an extremely low noise reference - so low that Johnson noise in your resistors becomes more of an issue! Regular zeners, though, are the noisiest things around - noise pedals for guitars often use zeners as the white noise source?!?! As far as unusual application of LEDs goes, I like their use as a mildly voltage-dependent resistor in DC servos. That's even better than the guy who figured out that varying the reverse bias on a diode varied its junction capacitance. But I digress... my suggestion for the reference in battery powered equipment, if one must go discrete, would be a bandgap. Operating current in the 10''s of microamps instead of the 10's of milliamps and lower noise than a zener (not lower than a buried zener nor heated zener, but that's the consequence of getting away with 3 orders of magnitude less current requirement!).

...

Quote:

5) The diode bypass is to prevent the current flow into the output of c.m. Kinda strange looking. But I think both them diodes are reverse biased. ...

They are reverse-biased, but that's one thing that's okay with the circuit, if a bit over the top. They prevent the base-emitter junctions from being driven into breakdown by a reverse voltage. The most likely cause of such would be hanging (yet another) big capacitor on the output of the c.m., which could subject the emitters to breakdown when the power is turned off.

Quote:

6) Ooo nice point. Current musta go somewhere. But then does it matter? Do you have data? I don't believe in simple theory. I like data a whole a lot better. Elliot states that c.m. is not a regulator.

For data, look no further than the temperature dependence of Vbe, which is -2.1mV/C. A 10C temp increase will cut the forward voltage drop from approx. 0.6v to 0.4v. This very much matters in that the circuit loses control of the emitter voltage, rendering the transistor a moot element. Would that degrade the sound quality of the amp? Well, if it were just a c.m., not likely. As long as there was a decent amount of "real" capacitance somewhere in the chain, but this is also a regulator - a poor regulator, mind you - and if the driver's base can no longer control the driven's emitter, whither the regulation??? Tis no more.

Quote:

In terms of LM317, I have no real complaint. It worked nicely for me for bunch of years. But then it would be wrong to say "the quietest, quickest transient response regulator." There are better. You probably mean, "the most suitable, most cost-effective yet high performance regulator." Again, "is it worth it?' is a difficult question to answer. Tomo

Indeed I did, and I apologize for getting a tad too opinionated on that one

The point was much like the one you just raised: is it worth it for so easy a load? Especially in contrast with this c.m./regulator circuit, which will likely compare poorly in all aspects with a bypassed ADJ pin LM317 (and 337 for the neg. rail) setup!

jeffreyj

 

[Voodoochile]

Not to suggest that the LM317/337 are or are not appropriate for this PS, but rather to address them on their own...

I have built several power supplies using both the 317/337 regs (with bypass caps, some with both small electrolytic and film bypass), and with the Linear Technology 1085CT/1033CT regs, again bypassed. Though the specs indicate that the LT regs should perform much better, and there is a slight difference in measured N/R, they both perform very well, and I surely cannot hear the difference between the two. Yes, I'm fussy, and have acute hearing. My eyesight is also 20/23, if that matters.

The LT regs offer lower dropout, but again, I'm not going for the ultimate in energy efficiency here, and we are talking about very low consumption either way. Either reg has plenty of overhead for regulation, and both respond with ease to the load. I do run a resistor current source on the PS as a whole to keep it warmed up, and they both perform nicely. I realize that this is probably the root of al audiophilia, but if a difference exists, and no one can hear it, is it a difference in all practicality?

****, I can't belive I used the word 'practical' in this forum. I hope I don't get banned for that.

Now back to the more relevant discussion amongst those far more qualified to comment.