Power Supply Resevoir Size

[Kindhornman]

fas42,

It looks like they do have a series that does not require a huge increase in size. The UL30 series is touted as a direct replacement for electrolytics in a comparable size. Thanks for the reference.

[gootee]

Originally Posted by jameshillj

Thanks Frank,
the various transformer capacitances and the effects of the current harmonics generated in both the sec and pri windings plus the effects of the load/filter capacitances reflected thru the diode bridges - perhaps the SIMs just gets too complicated to include these other mechanisms?

For me, it's usually just a problem to find what values to use, and where to place them in the model. If I have that information, it's usually no problem to include it in the simulations. And the simulator doesn't have a problem with it, either way. It can simulate much more complex models than I have ever even attempted.

[Krisfr]

Out of curiosity, has anyone tried hooking up several batteries to form a power supply for an amplifier? For instance 3 or 4 motorcycle batteries per rail, to a known working amplifier to see if there is an audible difference. Something in that direction to be able to say that a more perfect power supply is worth the work and or effort.

[Kindhornman]

Krisfr,
I think if you look here on one or more forums that has been talked about. I don't think that a battery has been found to be a perfect power source either as they have their own electrical inductance and capacitance parameters to think about. Do a search, I know I have seen multiple references to this.

[Terry Given]

yes, if for no other reason than large batteries make it really, really easy to wire up gigantic loops.

here's a good example of large loops in a PCB:

the EL cap placement means they reduce the loop area by very little - only the bit above them. At higher frequencies where the cap ESL becomes significant, the loop area is as shown

[danielwritesbac]

Originally Posted by Terry Given

The EL cap placement means they reduce the loop area by very little - only the bit above them. At higher frequencies where the cap ESL becomes significant, the loop area is as shown

Terry, do parallel caps at power supply boards reduce both overall loss and inductance similar to inductors in parallel?
The performance is nice. The idea seems to improve standard caps.

Also, I'm not quite familiar with how to assure filtering occurs, since the worlds best specs capacitors at power boards tend to send noise signal as intact as possible rather than flattening/ruining the noise signal like standard/econo caps do better and there's quite the difference in bass between those two choices, all of which favors the caps with the inferior specs and none of which favors the caps with the best specs (if used on linear unregulated power boards).

The only place I've been able to use the high zoot (excellent specs) capacitors as an enhancement, is as a shield for the amplifier board, where we do probably want to send the noise, intact, somewhere else. That actually favors low ESR varieties as much more important than low inductance varieties.
And, the wide body (big size) low inductance varieties are prone to having the amp sound as bad or worse than a cheap standard cap used at the amp board. Except for caps in parallel, I can't seem to implement low inductance caps to do any good anywhere in an audio amp as some fairly powerful caveats seem to accompany the wide body low inductance specialty caps, especially poor results with those that aren't low ESR.

P.S.
My testing capacity is crude; however, those caps that succeed best in a contest of peers, as an output cap and other signal duties, tend to be the excellent specs varieties (mainly low ESR), which, of course work great on amplifier board power circuit. But that variety are very weird on power boards and there's no bass. Help?

[gootee]

Does this look reasonable?

(I only used 10000 uF to get some idea of the avg voltage available. That's not really the point of the data here, yet.)

[Terry Given]

Daniel,

look at the PCB layout above. see the highlighted loops? pick one - lets say the white loop.

Break the white loop into 3 separate areas:
1 - the area above the electrolytics (up to the rectifier)
2 - the area directly beneath the electrolytics
3 - the area below the electrolytics (down to the O/P transistor)

these three areas correspond to three sets of stray inductance (which is proportional to loop area):
L1 - Inductance between the caps & rectifier
L2 - Inductance between the caps themselves
L3 - inductance between the caps & the O/P transistor

Its not too hard to physically calculate these inductances (F.E. Terman's Radio Engineers Handbook has pages of relevant formulae), nor is it hard to measure them given a blank PCB.

But without calculating or measuring anything, we can make some observations:

1. L3 >> L2 & L1 - the bulk of the loop area is between the electrolytics and the O/P transistor. It doesnt matter what you do to/with the capacitors to reduce ESL if this loop remains untouched. Its the 400lb gorilla thats defacating on the carpet and kicking holes in the walls during subjective listening tests.

2. think about the mutual inductance with the 1uH air-core output inductor - thats not going to help now, is it (hopefully they are at least orthogonal).

3. L2 adds to the capacitor ESL (and is probably bigger).

4. L1 doesnt really stop at the rectifier - the inductance (loop area) of the long dangly wires to the transformer need to be included too (although if cable tied together this is still a lot less than area 3).

A DC bus is supposed to be exactly that; a sizeable chunk of this thread has been demonstrating that good DC bus layout is important. However many of them are far from low impedance, and as shown there can be all manner of exciting inductive couplings present. "normal" wiring and PCB layout practises in audio amplifiers are horrendously bad, for no apparent reason I can see (other than applied misunderstanding).

So before you can attempt to compare the effects of various caps you really need to:
- ensure all xfmr wiring is short & tightly twisted
- ensure the DC bus PCB layout is extremely low inductance (regardless of how you do it, no single-sided layout will ever be good in this regard)

I'd also make sure the output leads are tightly twisted, and ensure the L,R output inductors are orthogonal and as far away as possible from each other and the AC input (this has the extra advantage of increasing immunity to external fields).

and then I'd measure the impedance of the DC bus (seen from the O/P transistor) to see how good my cap bank was. its not that hard to keep the impedance in the mOhm range up to a few MHz, and with careful design that can extend to a few hundred MHz.

[danielwritesbac]

Thank you! That was totally understandable and contains some good tips as well!

The other question was, although the loss of series filters and cables could increase the need of a larger power supply reservoir, what sort of parallel filters (other than caps, capmulti, kmulti, regs) could decrease the need of large power supply reservoir size?

[Kindhornman]

danielwritesbac,
I have been trying to keep up with this thread and I must admit that much of the technical discussion goes over my head. But at the same time when you questions whether any other type of filtering would reduce the requirements of the power capacitors I just don't follow that. There does obviously seem to be a balance between the allowable maximum voltage for a particular circuit with particular devices staying within the SOA and the required capacitor reservoir size to limit to some agreed upon % of ripple currents. What am I missing here that changes that? If you have a transformer with sufficient voltage output, high enough VA rating, and a minimum capacitance, how is that affected by any additional non-capacitance type of filtering?

Steven