Posted by John Swenson (A) on November 02, 2006 at 13:46:49
I have done a fair amount of looking at the supply with signals going through the amp. The first problem is what to use for the signal! You can use real music, but since it is varying all over the place its tough to really see whats happening on the scope (unless you have a good storage scope, which I don't at the moment).
So I decided to analyze the music that seems to have the largest sonic difference between supplies. What I found was large transients (ie they look like spikes) even on music that you wouldn't think of as having "transients", I would expect this on a piano, but not so much with choir, but the choir recording has more and bigger spikes than the piano. The recording that don't seem to be have as much sensitivity to the supplies I was playing with don't have nearly as many or as high spikes. (BTW this wasn't just commercial recordings, I tried some of my own recording made with high quality mics had the same sort of spikes)
So I decided on several sine waves, different frequencies from very low, low couple mids and a couple highs and some "spike" tracks. Looking at the music I found that the width and rise time of the spikes was actually fairly consistant, so I made a track with just the spikes. It consists of 10 spikes at sort of an average timing between spikes, then a fairly long time of silence, then 10 more spikes etc. This allows me to see how the supply responds ot these type of trransients and also whether it recovers fast enough so its "ready to go" when the next one shows up.
I've run this suite of tests on a couple different amps with different supplies in each.
I didn't do a complete rigorous testing regime with all possible combinations and run it multiple time and take rigerous notes on every single result so its not going to pass muster as an official scientific exploration of the subject but there are some common results. BTW I was using a non-inductive resistor as a load in all these tests, not a speaker, take that as you may.
The "traditional" designs tend to have PS modulations that vary with frequency, and they are not flat or simple curves. There are "humps" in the modulation VS frequency plots, thus the output impedance varies with frquency, but its not a smooth, variation. The latest low DCR designs I've been playing with aren't flat eaither but they don't have the "humps". The amount of modulation is also less than the traditional designs, but its still definately there.
The transient response is where things get interesting. All the designs have significant changes when the transients hit, but the low DCR designs recover much faster, they have essentially completely recovered by the time the next transient hits, but the traditional ones do not. You can see the B+ drooping from one transient to the next, the first hits, it goes down and starts recovering, then the next hits but its not all the way up, then the next etc, with the average level of the B+ getting lower each time.
This transient response is probably one of the major sonic differences IMO. The smoother impedance curve probably also has a sonic impact, but I don't think its as big a difference as the transient behavior.
I have not tried out the latest designs on a preamp yet so I can't comment on how that sounds.
I think its pretty obvious that the amp will be happier with less sagging on the B+, I can't imagine how it would LIKE more sag. The impedance issue is a lettle less obvious. The lowest impedance of the traditional was lower than the lowest of the low DCR and the highest was also higher. The low DCR kind of stradled the middle ground in this regard. My guess is that the amps like this better, but that is just a guess, no hard evidence one way or the other as to how PS impedance relates to sonics.
John S.
최근 tube DIY에 실린 글인데, Low H Low DCR Low C가 소리가 좋다는 근거를 댄글임
몇번에 나누어 전원부 설계 및 PSUD simulation에 대해 올리겠심다.
전원부에 대한 output impedance 보다 regulation(그기 그기 입니다)과 transient response(이양반이 중요하다고 하는 점)에 관해서...
좋은 전원부
1. 리플이 적을 것
2. regulation이 적을 것(low Zo)
3. rapid transient response
4. not expensive
5. not heavy
밧데리가 젤 좋은디.... 자동차 밧데리 200개 이상 사용해야 그를듯한데...
|