|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
24.6.199.104
The last week has been a major revelation to me. I added MV recifiers to my amps, it significantly improved some aspects but the added RF chokes and caps ("snubbers" to get rid of RF hash), added HF roughness and glare.I tried a whole bunh of things but they didn't seem to make sense, the change in the sound was not correlated to the reduction in RF hash. Then I started playing around with PSUD with a small stepped current rsponse and tried modeling the different snubbers I'd been trying. What I found was the filters I'd been using were under damped, they "rang" when a transient hit, the snubbers were exacerbating the ringing (it was already there but not as bad). The situations with the most awful HF problems were the ones with the greatest ringing of the filter. Aha, I'm on to something!!
So I started playing around with PSUD trying different choke inductances, DCRs and ap values. What I ound out was that my cap value was quite a bit too small. If I went from 100uf to 500uf I wound up with a 'critically" damped filter, more caused the voltage to slowly change, less caused ringing, right at the critical value the step caused the voltage to quickly change level with out ringing or overshoot, this looked like the place to be. With this cap value adding the snubbers made no differentce.
So I tried it out, I tried a number of different cap values and listened, wow I was quite surprised at the radical change in sound from a fairly small change in cap value. At 600uf the sound was dull and lifeless sounding, at 300 it was bright and glary, at 400 the glare was gone and the sound was very alive sounding, at 500 everything just sounded right, natural, not unbalanced in any way. Its kind of a toss up to which I prefer, the 400 "sparkles" a little more, its just a little more "wow" sounding, but after a couple hours of listening it starts to get fatiguing. The 500 I can listen to all night, with 600 I get up and leave after 10 minutes because its too boring.
I tried the same thing with a couple other amps and got similar results. Changing the PS caps changes the sound, I cn "tune" it with a fairly small change in cap value. Strangely enough with SS rectifiers the effect is not as strong as it is with MV rectifiers. I'm not sure why this is so. The peroperly tuned MV amp sounds quite a bit better than the best tuned SS amp, but with the SS amp you can make big changes without changing the sound that much.
Another interesting aspect of this is how choke DCR and cap value interact. A number of people here tout low DCR, but I found that when you use low DCR you need quite high cap to keep the filter at the critical damping point. If you have high DCR you can use much lower cap values. So it seems to be an interesting trade off. With low DCR you need large caps, which usually mean electrolytics, which conventional wisdom say don't sound as good as film caps. But if you use the lower value film caps, you need to go with high DCR chokes, an interesting balancing act.
All this has been with a simple LC filtere, as you start cascading stages it starts getting very complicated! With an LCLC I could eventually come up with a good set of values, but it was much more time consuming. The amp I put the MV in originally had a LCL filter, but it sounded better with just LC, no wonder, the second stage was ringing like mad! It woul be interesting to try a proprly setup two stage, but I don't have any room left, the amp is completely full right now.
So there you have it, I'm now a believer in power supply tuning.
John S.
Follow Ups:
Instead of a speaker crossover tuned to 500Hz with a 8 ohm load, it's more like a 3Hz crossover with a load somewhere around a few K. Surprisingly, when you like at it in filter terms, a lot of amps have overshoot and ringing around 2-3Hz - which guess what, is going to interfere with the sense of timing in the music, with 70 to 110 beats/minute.
Whats wierd is that these low frequency resonance effects were primarily noted in the HF range. Maybe the RF from the MV tube was exciting the resonances up high? I don't know, I don't have a good explanation, I can change the HF characteristics by changing the very low resonances of the PS.
What you say matches somrthing I have been thinking about for some time now, The Q and fr for the LC circuit can't help but have a big effect on the sound of the amp, at least if its single ended or class AB/B.As you say, most common power supplies you meet on the net are often underdamped and have a high resonant frequency. I wonder if some of them sound good "because" of the power supply ringing, and if the ringing its not adding a certain life to the sound. But when the supply is critically damped I believe you have a more integrated sound. I wonder if some times the low frequenct resonance is used (maybe unintentially) to provide some needed bass boost when using smaller output transformers.
I can understand and work out what happens with a single LC circuit, but it gets beyond me when a second stage is added. It should be simple, two LC stages both damped should work together, but every time I try and model it, I either end up with a overdamped supply, or the fr of one stage being a lot higher, so I get ringing, or if the fr of the two stages are the same, then you seem to end up with a almost chaotic system.
I have been trying to consider the LC systems in terms of the enery transfer between the two parts, Energy is transfered from the flux in the choke to the electrostatic charge in the cap and vice versa. But adding the second stage just gives the sytem too many places for energy to go. It can move from the first choke, into the cap, then from the cap, some into the second choke, and the rest into the first choke, while thats going on the energy from the second cap is being transfered into the second choke. And so on.
What I am going to try over the next few weeks (I have a 1300v PSU thats just LC at the moment, 50H into 50u, sounds good, and is just about damped with a Q of about 0.6, but I need a bit more smoothing so I am going to add another LC stage) is a LRLC setup, the R providing a bleeder resistor to maintain the choke input supply above its critical current, but not to produce a LC circuit, Then a LC circuit to supply extra smothing, but that way I haven't got two LC stages to interact. It won't have the level of smoothing that a LCLC would have, but it may have enough for my needs.
Yo!LCLC is a hard nut to crack. What I have been doing is building a simple LTSpice model with,say,SV572-3. I used two AC sources - one at the PS simulating PS tranny, the second simulating strong (1V below max bias) signal. Then I was FFTing the signal on the last cap, trying to get the least PS + signal IM as possible (just judging by eye). What seemed to work best is the final LC cell with the lowest Q, calculated as in the series LCR. This is something that really suprised me as I would expect that it behaves more like a parallel LC circuit....unless I screwed something up of course.
> I wonder if some of them sound good "because" of the power supply ringing
I've wondered that also - but it's what sounds good that matters, right?> when the supply is critically damped I believe you have a more integrated sound
Every time I've pursued something approaching 'critically damped' - it ended up sounding near death.Re: modeling L-C-L-C filters: a really tough thing to do...ears work best!
With this amp the best spot is just slightly less damped than critical damping. It definately does sound flat, dull and "dead" when much past critical.I agree that tuning by ear is the best way, but using PSUD to find out where it was critically damped gave me a good head start on zeroing on the area to try. Signficantly away on either side was just not working.
> > I wonder if some of them sound good "because" of the power supply ringing
> I've wondered that also - but it's what sounds good that matters, right?Oh yes, I fully agree, but it would be nice to narrow down the search a bit.
I wasn't suggesting that there was any problem with that If thats part of why they sound good,then great.
> Re: modeling L-C-L-C filters: a really tough thing to do...ears work best!
Yes, no doubt, but thats not so easy to do without a big stock of chokes and caps, if you are like me starting out in this game, and having to decide on things to purchase and get wound, it doesn't hurt to have some ideas on paper first. And I would like to hope that understanding whats going on won't prevent me from using my ears as the final judge.
Two thoughts. Push-pull amps do not need high capacitance with a LCLC. Set amps do require high capacitance on the last or output cap. On a LCLC, the last cap should be at least 100uF & many prefer upwards of 500uF.
Agreed about PP, well at least if they are not running into B, then they need the supply to be able to cope with the chnaging load.I am just not sure about "SET's requiring large caps", I don't think it can be put as simple as that, as if its a fact of life. Its a real double edged sword. Ok, A large cap means the supply can cope with a big peak demand, but the LCL stage before that has then to charge that big cap up again before the next peak load comes along. And whats going on to the amp while the cap is charging up? It never seems to be as simple as that. Ok, a large cap is good, but then you have to have a low DCR choke before it to allow it to charge, that means you start to have the resonance and damping problems that this thread started about. I have heard some wonderful sounding single ended amps that have only a small PIO cap after the choke, 8u or so. And voltage is another factor. There is no way I am going to charge 500uf up to 1300v, as it is, the 50u I have contains about 50 joules of energy when the amp is running.
SET amps have a large current swing thus is best with low DCR transformer high voltage winding and chokes plus a large electrolytic cap.My PP quad set 6V6 amp is operating Class A, thus has a 5HY to a 40uF to a 5HY to a 25uF oil cap. First choke is 150 ohm DCR & second choke is 40 ohms DCR. It is the best sounding non triode amp I ever heard.
Another option. A shunt regulator in-between the LCLC or LC-shunt-LC also provides great sonics. I like the less than highly efficient shunt type regulator as it does not seem to color sonics. In fact, I believe the shunt regulator provides best quality sonics, but the location in the filter is important.
Interesting post. I use either a CLC or for the most part a LCLC filter. When I tried adding 100uF oil cap to an existing 25uF oil cap on the output of an LCLC, I heard no difference in the sonics. The amp is a PP type.I also believe a LC filter is more subject to careful tuning than a LCLC filter or may I state the LCLC is more forgiving.
This post is made possible by the generous support of people like you and our sponsors: