Designing for PSRR

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Having spent some time on active regulation circuits (after Jung et al) I've decided that improving such designs significantly is too much like hard work at present.

The dearth of suitable error amps and the complexity of discrete design makes this a task for the back burner, at present.

My mind is now turning to DIY amp design, and I've found plenty of useful informative articles about design covering all of the normal parameters one looks for, low distortion, dynamic behaviour etc.

Designing circuit for improved PSRR though, to further enhance supply-induced errors, has drawn a blank. It's obvious that those who consider PSU-induced distortion a fallacy have in their own designs some huge susceptibilities to rail-induced errors. A look at Douglas Self's 'blameless' amplifier articles shows that the current sources to one rail give high PSRR across the audio band, but the negative rail is frankly rubbish. and barely attenuates any noise that appears there.

The classic degradation of this parameter with increasing frequency is happening in conjunction with the PSU's degrading performance too - further compounding the effects.

So my question to the forum learned, is do you have any pointers to articles, design ideas and resources that can I can start to investigate?

Thanks all,

AX tech editor
Joined 2002
Paid Member
PSRR etc


Sorry, don"t have any pointers right now, will look over the weekend.
But, having an obvious interest in the Jung/Didden regulators, I wonder what your experiences were trying to improve them. If you have any interesting info, I would like to hear it, off-thread probably.

Cheers, Jan Didden

"RC or LC filters in the supply work wonders....."

But have a very unfortunate side effect in that they raise rail impedance at LF. I'm trying to reduce rail impedance across the board, and LF is where the most significant current drains will be.

My experience with LC and audio is not good either, although their use in HF / digital is something I'd support.

Unless drain is absolutely constant (an almost impossible thing to acheive) I feel a lower rail impedance rather than higher is likely to be better, if my experiments thus far are anything to go by.

Using the Jung reg's I've removed the RC filters in my preamp stages, and the difference in rythmic and low end performance is stunning.

Increase the caps

Fair comment, but then it starts to look inelegant, gets expensive and bulky and gives performance degradation with time, since polarised caps are necessary.

I'm also not convinced that it's sonically beneficial, and in my experience there's a lot to be said, subjectively, for achieving a balance.

From your experience what sounds best - rising PSU impedance at HF, flat response across the audio band, or falling impedance with increasing frequency?

I'm curious because I've found that for the music to remain intact, what one intuitively feels is right doesn't always work. There could be other factors at work of course, so it's useful to discuss such things and find out if the blanket statements often made are based on universal truths, or an indivuals limited experience. No criticism intended, just an attempt to get a bit more than an 'increase the caps' statement ;)

I find many of the raw amp power supplies that we often see pictures of on this forum fascinating - it seems that 'more capacitance = better' is a mantra set in stone, but my experience is that there is an optimum value, related to transformer / rectifier characteristics and that it is always much lower than the values most people use here.

Have you experimented, for example, with capacitor size in the unregulated supply for a preamp or power amp? I have a PSU based on a 600VA transformer that sounds noticeably worse if more that 15 -22,000 uF is placed after the rectifiers, irrespective of configuration (single cap, multiple caps, different types etc).

Larger caps slow the sound rythmically by a significant amount, and result in all the important timing clues and a significant amount of upper bass detail being lost.

The main difference is suspect is I have a different set of rules for assessing musical replay than many, IMHO the pace, rythm and timing integrity of a system is far more important than the HiFi froth of 'space', 'imaging' and other things so often talked about as improvements. I'm not saying these things are uniportant, but that they are icing on the cake, which has to be right first.

Disabled Account
Joined 2002
Hi alw...
Self discusses this method in his's commonly used in i.c's.....cascode the diff. pair.....and connect your miller comp. capacitor to the emitter of of relevant cascode transistor, rather than the base of the VAS, (correctly called the Transimpedance stage)... Chigwell..:cool:


PSRR can most definitely be simulated in SPICE - it's easy to add modulation to a DC source and see the result.

This is, in fact, what I'm currently doing in the design phase.

It's easier than actually measuring it - you require some serious amp performance to swing AC over a wide bandwidth into lots of capacitance and maintain stability.

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.