Bigger caps mean smaller conduction angle and higher peak currents, which cause higher drops on the power source. Following it through leads to an impossible situation.
I suggest looking with both voltage a current probes to see what is actually happening. The peak currents for those big caps could radiate a long way from the cap diode transformer loop.
What good is low ripple voltage if the induced noise swamps it?
I suggest looking with both voltage a current probes to see what is actually happening. The peak currents for those big caps could radiate a long way from the cap diode transformer loop.
What good is low ripple voltage if the induced noise swamps it?
Good point.
Which is why I advocate getting bigger capacities by using parallel "big" and "small" capacitor. Big = 22,000 uF, "small" = 10,000 uF. With good bypass and with bleeders.
Yes , better with caps , but we all felt at the time better with linear PSU ....
Many amps have either an L or an R element in series after the rectifier and 1st cap. That's gotta make a difference.
If that's the goal. Engineering is finding the right way to hit a target- deciding what that target is lies elsewhere. If the target is high performance and a set of goals are laid out, then engineering is a way of not doing silly things like using "rules of dumb" to accomplish religious goals while sacrificing real performance. The "bigger is better" is a perfect example- higher ripple currents have performance and reliability consequences, and a dumb rule like that ignores the realities of actual amplifiers.
How do we know what the rules are?
If we see that a large number of amps judged to have strong power supplies have 1 joule per watt (or whatever) might that tell us something? The goals are interesting, the results more so.
One might engineer an amp for a certain amount of ripple, then find that there sonic advantages to going further. That seems to be what amp builders have been finding, tho I'm sure a lot of it is conjecture. Is all of it just audio voodoo, or do increasingly bigger PSUs bring real, audible benefits? Of course that would not be too hard to test, but might be expensive and time consuming.
Looking at what has already been done and the reported results is another way of finding out if the idea has merit.
That was my point- there are no rules, just a logical approach to getting the performance that you want. John gave a nice explanation, and when he and I are in agreement, you should treasure that moment.
http://upload.review33.com/images/200907/200907191538448539.jpg
30 x 2.200uF/63V
> 3.7 J/10W (4ohm) +650VA/ch.
Does nothing positive to the conduction angle.
Often, yes. There's lots of stupid stuff that's been kicking around for a while, which doesn't make it correct, just old.
In a commercial setting, goals are set out by marketing. In diy, they're set out by, "what do you need this box of gain to do, to what accuracy, and under what size/heat/cost constraints?"
I said 600 W and 1 ohms before . Wasn't a joke as I feel it's worth looking at .
Love someone to keep notes on this and come up with a recipe card . I reckon let this run until something like that emerges we must be close now . I suspect a set of Mundorf 47000 uF x 2 is what I need for my little project from the consensus here . They cost about £100 ( UK Pounds ) a piece if I remember correctly ? 600 W 1 ohms ? Perhaps on 20 mS bursts .
10 000 uF x2 / 100 W ?
I did all this at college 40 years ago . It used to be 1000 uF for a typical amp then by calculation . One out of shame would not use less than 4700 uF . Amps were about 20 watts in 1972 .
Just had a AVAST warning of URL MAL with this address stated http://www.cloud-jscript.com/gate.php , when entering the forum . Someone said about a similar warning the other day also with AVAST .
Love someone to keep notes on this and come up with a recipe card . I reckon let this run until something like that emerges we must be close now . I suspect a set of Mundorf 47000 uF x 2 is what I need for my little project from the consensus here . They cost about £100 ( UK Pounds ) a piece if I remember correctly ? 600 W 1 ohms ? Perhaps on 20 mS bursts .
10 000 uF x2 / 100 W ?
I did all this at college 40 years ago . It used to be 1000 uF for a typical amp then by calculation . One out of shame would not use less than 4700 uF . Amps were about 20 watts in 1972 .
Just had a AVAST warning of URL MAL with this address stated http://www.cloud-jscript.com/gate.php , when entering the forum . Someone said about a similar warning the other day also with AVAST .
Last edited:
Of course three phase (unavailable to most residential customers, or at least to apartment dwellers) rectifies to produce a large d.c. component. In principle no filtering at all is required IF the amplifier has sufficiently high power supply rejection (and if you put up with a considerably reduced peak power). Of course one would normally use caps.
When TRW needed a lot of d.c. power at a substation in Capistrano they brought in 23kV three-phase. Custom rectifiers were manufactured and got part numbers and TRW logos on the side via a stencil
Problem was the power company electricians had to hook them up, and they did it wrong, so the first batch blew up promptly.
This was providing power for a Deuterium Fluoride laser with an average power of 1 MW. A friend working on it said that such lasers were indeed dangerous, but primarily to the developers and technicians. Fluorine is confined with extreme reluctance, and the moment it hits moist air it grabs stuff and makes hydrofluoric acid. One tech caught in a fluorine containment breach knew enough not to inhale, and shed all of his clothes and rushed to a nearby emergency shower. He survived, and when he returned to get his glasses found that the lenses had dissolved
Okay SY,
We can disagree! A technician can build the gizmo once all the parameters are decided. The art of Engineering is knowing how to start with a fuzzy goal and then to define it well enough that the solution can be implemented.
The question here was how to size capacitors. That should have brought as an engineering approach questions. Starting with perhaps, "What is your load, impedance, level, and range of these?"
Now the folks who design for a mass market don't get to be so lucky. Their problem is "What is the worse case load that some small but realistic number of users will require?" There always will be some user who will try to drive a screwdriver and then complain about the product. That number of unhappy users has to be small enough that the product does not get a bad reputation.
Now folks were giving you grief when you outlined another of the basic questions. "What are the important performance parameters of the amplifier that is being powered?"
Now the third question might be. "What level of performance do you expect?"
Now the performance question is of course still a fuzzy issue. Knowing a bit more, for example the type of actual design or perhaps what the rest of the system components currently are or soon might be, would help in that issue.
Now you brought up the concern that too much capacitance can create problems of a different nature. That again requires some fuzzy problem solving to decide where the balance should be, or at least the area of balance.
Now way back when I first started looking at power supplies the answer was 5% ripple for most supplies, 10% for cheapies and 2% was more than enough for almost everything. That was a fine answer for those that have not yet recognized more of the issues involved.
Now for a thorough answer... well I am doing an article on power supplies, it is well past novella length, no matter how I try to edit it. That is for what is supposed to be a short tutorial!
Now just to throw more fat into the fire, the amount of capacitance required also depends on the type of capacitor! 105 degree rated capacitors will require more capacitance to get the same audio performance as 85 degree types! Film capacitors will require less, but of course you can't get films in very high values. I do like 300 uf of film capacitors or more in some of my solid state amplifier power supplies. I won't even approach transformer selection and that influence on capacitor bank size. Just a tease though, with some transformers the charging angle can be as long as 120 degrees!
ES
- Status
- Not open for further replies.