Power Supply Resevoir Size

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Assuming a 100 watt amplifier driven at 100% duty cycle/factor what size reservoir capacitor would be required in the power supply considering that E = ½CV². A 100 watt being 100 J/s then:

If one would select a rail voltage of 50V one arrives at a reservoir capacitance of 80,000 uF.

I do not find any rule of thumb on this topic on the forum and since amplifier sound characteristic is directly affected by the power supply, maybe creating some "standard design rule" with some plausible reasoning would be most helpful to the DIY community

I would appreciate comments from our power supply specialists on the forum regarding this matter.
 
You need to assume the amount of ripple that you are trying to achieve.

It is impossible to have zero ripple.

Ripple V = I / (f * C)

Therefore lets assume I = 4A and f = 120Hz (Full Wave Rect).

If you require a ripple voltage of 100mV.

C req = 4 / (120 * 0.1) = 333000uF (unrealistic - but so is 100mV ripple at 4A)

If you are using 50Hz mains, Full Wave Rectification and allow for 1V of ripple C becomes 40000uF.
 
Last edited:
I usually use C = (It) / V where:
C = capacitance required (Farads)
I = maximum current
t = 10msec for 50Hz and full wave rectification.
V = acceptable pk to pk ripple voltage

formula comes from Q = CV and Q = It
It's a bit of a simplifcation, but seems near enough and on the safe side.

HTH,
Brian.
 
Administrator
Joined 2007
Paid Member
I do not find any rule of thumb on this topic on the forum and since amplifier sound characteristic is directly affected by the power supply, maybe creating some "standard design rule" with some plausible reasoning would be most helpful to the DIY community

Bigger isn't always best.
As you say, amplifier "tuning" or the sonic characteristics of an amp can be altered by its PSU design.

Again this is where objectivism and subjectivism might not agree.

A) The amp is better technically with the low ripple massive PSU.
B) The amp sounds better with a smaller "faster" PSU.

So which do you build ? A or B

:)

Some good general useful info here... but it doesn't tell you how godd an amp will sound with a given PSU.

The Signal Transfer Company: Power Output
 
It's also worth looking at the ripple current through the capacitors and checking versus the rating and life / temperature. Crude caluculations or a simulation in LTSpice can get values or build and measure.

A very crude rule of thumb for power supplies is 2000 uF per amp (of load current). I guess a minimum for Power Amplifiers.

A decent power amplifier should be fairly tolerant of ripple voltage. Their ripple rejection is likely to vary alot depending on design.

It also depends on what output current the amplifier is limited to or expected to deliver.

So lots of variables.
 
Do you think it is only ripple that is responsible for a better sounding amplifier, or is there more to it.

The reason for my inquiry came from the week-ends escapades. I had little to amuse myself with so I fooled around with a JLH class A that I knocked together and was discussed on one of the threads here a few months ago.

At that time I swapped between a 300VA smps and standard 300VA transformer supply with 20 000 uFcaps and a remarkable "improvement" resulted while running off the smps.

On the week-end I replaced the SMPS with 2 x12V 22.5Ah gel batteries and the"improvement" again, chalk and cheese.

The SMPS has a 220uF capacitor on a 315V rail, which equates to 10.53 Joule while the 300VA full wave rectified transformer has 20 000uF on a 24V rail and equates to 5.7 Joule.

According to the peak ripple equation the standard supply should be 0.65V at 1.3 amp load, the SMPS is measured 30mV at 1.3 amp, while there is zero ripple with a battery.
 
A very crude rule of thumb for power supplies is 2000 uF per amp (of load current). I guess a minimum for Power Amplifiers.

I have also heard of this rule of thumb, and we are on the opposite ends of the earth - is there a reasoning behind the rule (I was told this story in the 60s, but at the time it was only 1000uF/amp maybe easier calculations) Assuming all things equal, could the ripple only be responsible for "smearing" the sound. Does amps with better PSSR sound better than amps with lesser PSSR using the same power supply,.......
 
Last edited:
Administrator
Joined 2007
Paid Member
Do you think it is only ripple that is responsible for a better sounding amplifier, or is there more to it.............

I wouldn't even hazard a guess at that... probably more to it I would say.

Could it be that the audio modulates the rails and some of that gets into the mix somehow. If the combination of amp suceptibility and just the right amount of hash is present then who knows.... maybe it is more pleasing to the ear.

Any effect must be vanishingly small but then again so are the differences between many opamps in the normal sense and they can all sound totally different as we know.

Batteries are a strange one. Have you tried measuring the impedance of them at hf and as they discharge ?

I have come to the conclusion that audio design is often as much an art as it is a science.
 
Administrator
Joined 2007
Paid Member
Mooly, you built a few amps in your time, don't say maybe this or that, say what you experienced as being "better", we could be on to what the real difference are between amps.

OK then. In no particular order.

Best allrounder for technical performance, SNR etc etc was a Pioneer A80 that I bought new in the early days of CD. (Maybe it doesn't look quite so stellar now but it was back then :)

How did it sound. Super clean, silent background, detailed and revealing but ultimately clinical and boring. I would call it "glassy". It was the amp that had me believing all I read about how it is the source material that is at fault and that only a few recordings sounded really good and involving.

------------------------------------------------------------------------------------------------------

The first amp I made was based on AD149 germaniums (a kit) and used 1000uf rail cap running around 30 volts or so from memory. Whatever it's failings it made music. 470uf speaker coupling caps.

------------------------------------------------------------------------------------------------------

A Practical Wireless design from 1978 (the Europa) that used TIP3055's in a quasi output. That was good and the unusual feature was the LTP input stage. The inverting input for some reason was a darlington (BC109C's) against a single BC109C for the non inverting. Probably an interesting distortion spectra.

------------------------------------------------------------------------------------------------------

Doug Selfs blameless Class B built as blamelessly as possible on the official PCB's to the exact original design and layouts :) 600VA PSU and 10,000 uf rail caps.

A bit like the Pioneer in musical presentation. Good HiFi but not so involving to listen to. Years later I discovered a major error in the official PCB layout and the recommended wiring scheme that introduced noise and hum into the audio. I could never get the amp totally silent hum wise but in the end sussed the error out (in disbelief after all the pages and pages of text on addressing all the distortion mechanisms) and tried a simple fix which was 100% successful. The amp was no longer in daily service then though.

------------------------------------------------------------------------------------------------------

Best sonically (and your all fed up of this one) is the lateral MOSFET amp. Smallish PSU (300VA and 4700uf rail caps) and it's single ended input topology. If it wasn't the best I had heard I wouldn't be still listening to it daily after all these years. It does things sonically that others don't. The ripple rejection isn't particularly good... does that have any bearing... I don't know. I also run the laterals at around 70ma quiescent (but don't tell anyone :)) which is below the 100ma "lateral standard". Listening tests all confirm its sonic abilities.

What this can do when fed from a 64kb WMA file is unbelievable. Sure it's not as open or revealing as the original CD source but it's in no way unpleasant or lacking "musicality".

-----------------------------------------------------------------------------------------------------

Heard a few Sugdens in the last few years, have been round the factory and have a friend with two Sugden setups (amps and CD players) and I kind of used those as an initial benchmark for quality.
 
I don't think that we are at opposite ends of the earth. The crude rule of thumb I quoted is for general purpose power supplies, not for power amplifiers. It is only a starting point that will hopefully give a reasonable level of peak to peak ripple voltage and acceptable ripple current for a typical electrolytic capacitor. So at least the power amplifier will work.

The JLH Class A amplifier which I assume is the 10 Watt 4 transistor design might have poor power supply ripple rejection. The Wireless World article (High Fidelity Designs, compilation) includes a ripple filter circuit.

I am not sure that switch mode power supplies are well suited to audio power amplifiers because the output load transient response is usually worse than a linear power supply and the output capacitors are small. I would expect significant modulation of the power supply voltage by the audio particularly at higher frequencies. The simple JLH Class A design is likely to have poor rejection of this.

Batteries have output resistance so the voltage is being modulated at high frequencies where an amplifier power supply rejection is likely to be worse than at 100 Hz so they aren't a perfect solution. At high audio frequencies a large electrolytic capacitor might have a lower impedance than a battery.
 
Interesting comments Mooly, I remember the days when 1000 uF filter cap was a monster cap, maybe that is where the thumbs started ruling I guess we collected the same magazines from around the same time.

Now regarding the lateral mosfet amp you talk about, I fully support you view of one of the nicest sounding amps and also remains one of my favorites with 55V rails and 4700uF.

Until I built a modified version of Anrej's (Lazy Cat) SSA it was my everyday office amp of choice. I think it is time to reconnect.
 
So KatieandDad (how did you dream up this handle) and PChi what is the verdict which power supply sounds better the mega farad PSU or the 1000uF (regardless what we should be saying to audiophiles.)

I agree with PChi, that battery impedance does change with frequency in fact most things do. I did decouple the board with 470 uF/1uF/10nF but this I would have done in any case regardless of power source.

Why I used the JLH is that it presents a constant load and therefore ripple unlike class AB that would be way to loud to listen to by the time the rails start bouncing up and down.
 
Last edited:
I haven't got a definitive answer.
I believe that reasonable amplifiers sound very similar because I have found it very difficult to hear any differences in A/B comparisons (unlike Loudspeakers).
I think that the power supply and amplifier need to be looked at together. My guess is that reasonable power amplifiers with good power supply rejection are insensitive to the capacitor size so there is no difference until clipping between a mega farad PSU or 1000 uF. Whereas some amplifiers are sensitive so it will make a difference.

The JLH design is a simple Class A amplifier so I don't know how constant the current draw is. Also if the speakers have some low impedance dips then a Class A amplifier might drop into Class B.
The JLH bias currents are not well defined and it appears to have no frequency compensation making high frequency stability marginal. The distortion measurements are also quite high. I can believe that it is possible to hear a difference between this and a modern amplifier. With large loudspeakers the bass response needs to be within a fraction of dB down to tens of Hz for no differences to be heard. That is difficult to achieve with a realistic sized output coupling capacitor so again a difference might be audible.
 
Nico,

What makes those mosfet amps special ;) ? Guys I told you I have the formula ;) .Make the SYMEF or any other HAL amplifier and become happy. I posted a small funny pic about power supplies here http://www.diyaudio.com/forums/solid-state/211977-quick-guide-grounding-4.html#post3029294

A good amplifier should not be affected by small variations in power supply, especially if it uses feedback ;)

A switchmode supply is likely to have stiffer supplies than the normal, and batteries are batteries
 
Administrator
Joined 2007
Paid Member
Thanks Nico :)
Happy days back then.

For anyone interested this was the Pioneer A80 (sorry going a bit OT here).
I was deeply impressed at the time by the complexity of it and yet something like the JLH69 would eat it alive for sound quality.

I'd forgotten how scary those rails were too.

-/+ 82 volts for the main amp. The outputs ran on a much more sensible (lol, you can not be serious) lower supply of around -/+65 volts only switching to the higher rails when needed... which was probably never.
 

Attachments

  • Pioneer A80.jpg
    Pioneer A80.jpg
    196.9 KB · Views: 2,470
  • Pioneer A80_2.jpg
    Pioneer A80_2.jpg
    194.8 KB · Views: 2,406
  • Pioneer A80_3.PNG
    Pioneer A80_3.PNG
    98.8 KB · Views: 2,377
Mooly, is an interesting comment and one I have also mentioned on Harrison's (OnAudio) thread. You say the front end runs on its own supply, remember the thread regarding the Goldmund about a year ago, it also had separated the supplies.

Hugh Dean decouples his front end from power stage with diodes. All of these amps are said to have "exceptionally clear sound with very defined and deep sound stage" could the ripple be the primary cause of smearing the sound and is it primarily affecting the front/driver end.

Correct me if I am wrong, but could an amp with good regulated low ripple rails on the front end be the main requirement for a good sounding amp while the output stage need not have a very sophisticated supply at all.

In other words should we spend less on mega farads power supplies when 4700 uF (or less) will do and pay more attention to the front end driver stage power supply ?
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.