I have found it useful in a number of situations to add a good lo-ESR capacitor of 1-10uF between each rail and ground. The trick is where you place it physically in the amplifier. The "rail" should be close to the output devices. The "ground" should be at the speaker return, again, close
to the output devices. This assumes the amp has been laid out so that power supply and speaker runs are not sharing vulnerable ground or return points.
The idea is to consider the capacitor as your only power supply, and to minimize the physical loop area that the output current encompasses. By minimizing loop area, you are minimizing inductance. More importantly you are reducing the loop antenna effect.
I have seen a number of amps that the output circuitry couples high frequency transients back into the input stages by induction. Careful choice of added bypass capacitor and its placement can reduce those effects. Poor choices of location can cause added injection of transients.
You can test it by running a high frequency suare wave (10-20 kHz), at full power (so you get roughly max current), then checking various nodes in the circuit. You'll need to run your 'scope with differential inputs, and keep the probe wiring closely paired so the probes don't act as a loop antenna. It helps if you are comfortable with measuring RF.
Does it make any difference? Well, it might help with high frequency stability. I had one amp that had a couple of percent high frequency feedback because of induction back into the first stage. Adding polypropylene "snubber caps" from the output transistors rail connections to speaker return line on the PCB reduced it by a factor of about 5.
In small amps you can often strategically place the rail electolytics to do this job, but often the main electrolytics are far enough away, and the physical loop will cause inductive effects at high frequencies.
Does it make any difference to the sound? Too hard for me to tell. Although I have an A/B switching box (and randomization) I'd need two amps that are identical except for the modification. My experience with amps (that are decent) is that it take a lot of careful and not so careful listening to pick up the differences reliably. The amp that had the problem (Linn LK100) had been panned by reviewers for being "grainy", whatever that is. I have made a number of changes to that amp, and compared it to others, and found it to be no worse than any other DECENT amp I've heard in listening tests. It is however, better now, as far as my test equipment can tell!
Paul
to the output devices. This assumes the amp has been laid out so that power supply and speaker runs are not sharing vulnerable ground or return points.
The idea is to consider the capacitor as your only power supply, and to minimize the physical loop area that the output current encompasses. By minimizing loop area, you are minimizing inductance. More importantly you are reducing the loop antenna effect.
I have seen a number of amps that the output circuitry couples high frequency transients back into the input stages by induction. Careful choice of added bypass capacitor and its placement can reduce those effects. Poor choices of location can cause added injection of transients.
You can test it by running a high frequency suare wave (10-20 kHz), at full power (so you get roughly max current), then checking various nodes in the circuit. You'll need to run your 'scope with differential inputs, and keep the probe wiring closely paired so the probes don't act as a loop antenna. It helps if you are comfortable with measuring RF.
Does it make any difference? Well, it might help with high frequency stability. I had one amp that had a couple of percent high frequency feedback because of induction back into the first stage. Adding polypropylene "snubber caps" from the output transistors rail connections to speaker return line on the PCB reduced it by a factor of about 5.
In small amps you can often strategically place the rail electolytics to do this job, but often the main electrolytics are far enough away, and the physical loop will cause inductive effects at high frequencies.
Does it make any difference to the sound? Too hard for me to tell. Although I have an A/B switching box (and randomization) I'd need two amps that are identical except for the modification. My experience with amps (that are decent) is that it take a lot of careful and not so careful listening to pick up the differences reliably. The amp that had the problem (Linn LK100) had been panned by reviewers for being "grainy", whatever that is. I have made a number of changes to that amp, and compared it to others, and found it to be no worse than any other DECENT amp I've heard in listening tests. It is however, better now, as far as my test equipment can tell!
Paul
Exactly.
Do it at very high frequency for the nF at the power pins.
Do it again for the small electrolytics @ medium to high (Audio) frequencies.
Finally do it for the main smoothing bank @ lower frequencies.
If an additional bypass capacitor will make an improvement depends on your amplifier. If the power supply capacitors are inside the feedback loop and you have moderately efficient speakers and listen at a reasonable level so that the output voltage level is a few % of the rail voltage, you most likely will not notice any difference.
If you have a low global feedback class A amplifier using say 105 degree rated capacitors from the lowest cost producer, then a few hundred uf of film capacitors on each rail will make quite a difference.
As to the example of Altec and Crest amplifiers. Crest puts in parts when field failures tell them they are needed. As to Altec 9440A amplifiers engineers, maybe! The capacitors are used to keep some units from oscillating.
The best use of a bypass film capacitor is, as the larger aluminum electrolytic capacitors age they do not filter higher frequencies as when new. So a small film capacitor will reduce warranty claims.
In your amplifier I doubt any added film capacitor will make much of a difference.
Oil filled capacitors are one of my favorites for vacuum tube amplifiers as the power supply input capacitor. They tend to handle surges better than other types, they last longer and they resist vibration induced noise.
If you have a low global feedback class A amplifier using say 105 degree rated capacitors from the lowest cost producer, then a few hundred uf of film capacitors on each rail will make quite a difference.
As to the example of Altec and Crest amplifiers. Crest puts in parts when field failures tell them they are needed. As to Altec 9440A amplifiers engineers, maybe! The capacitors are used to keep some units from oscillating.
The best use of a bypass film capacitor is, as the larger aluminum electrolytic capacitors age they do not filter higher frequencies as when new. So a small film capacitor will reduce warranty claims.
In your amplifier I doubt any added film capacitor will make much of a difference.
Oil filled capacitors are one of my favorites for vacuum tube amplifiers as the power supply input capacitor. They tend to handle surges better than other types, they last longer and they resist vibration induced noise.
We also should remember, that high NFB and high PSRR designs are not good sounding in most cases, since the NFB is good only in theory, in practice it causes sound harshness, in particular, by trying to fulfill the PSRR task.
NFB is good in those cases, were a subject for its action is very small, i.e. initial design before NFB is very linear and non distorting.
If we put bad cap in PS and propose PSRR to fix the problem, we have standard Hi-Fi sound of low and medium priced equipment.
I believe, main secrets of the good sound of some expensive models are based on deep investigations of how PS must be arranged.
NFB is good in those cases, were a subject for its action is very small, i.e. initial design before NFB is very linear and non distorting.
If we put bad cap in PS and propose PSRR to fix the problem, we have standard Hi-Fi sound of low and medium priced equipment.
I believe, main secrets of the good sound of some expensive models are based on deep investigations of how PS must be arranged.
Last edited:
"NFB is good only in theory, in practice it causes sound harshness"
Let's not go there, as it could easily end up as a thread hijack. Stick to caps.
Let's not go there, as it could easily end up as a thread hijack. Stick to caps.
- Status
- Not open for further replies.