Capacitor Question - Amplifier/Power Supply

[thylantyr]

I'm in the thinking stage of the Holton AV800 amplifier
project.

My printed circuit board size looks like it will be
between 6"x10" up to 8"x10" with 28 output mosfets
per channel. (one pcb = 1 channel).

Question:

Would you rather use 14 Panasonic FC series 680uf (100V)
low impedance capacitors per rail mounted on the circuit
board, capacitor diameter is .71" x 1.57" height.
Total of 28 caps per board, per channel.

9520 uf per rail.

or

Would you rather use six larger diameter generic computer
grade caps per rail mounted on the circuit board, perhaps
3000uf, 1.5" dia. x 3" height caps.
Total of 12 caps per board, per channel.

18,000 uf per rail.

These capacitors mounted on the circuit board are extra capacitors, the power supply will have large can type
capacitors inside the amplifier chassis, perhaps 40,000 uf to
100,000 per rail.

What are the pros/cons of having these extra ones near
the output mosfets mounted on the pcb ?

or

Do you have another other suggestions ?

 

[thylantyr]

Look into my eyes .......
You are getting sleepy.......
You will say something ......

 

[jan.didden]

Distributed cap

Generally I would favor distributed cap at each output device. That way you keep wiring impedance low, and you improve the hf performance as all the ESR's and ESL's are in parallel, though this may be a marginal difference in practise.

How you run all the caps to a start gnd point becomes an issue of course. As far as the primary supply cap is concerned, I would not go over 50mF or so. You get really nasty sharp charging current pulses, which can wreack havoc with your s/n ratio. The benefit for the amp of such massive capacitance is marginal at best.

Can I go now?


Jan Didden

 

[thylantyr]

Thanks for the comments.

You get really nasty sharp charging current pulses
Is this caused by any type of capacitor or do these
low inductance type of capacitor cause more problems than generics?

I was reading the pass labs power supply article
taking about using 100,000 uf of capacitance (minimum)
for the power supply and I was also reading the
owners manual of the XA200 which claims to have
200,000 uf of caps.

My printed circuit board may be 4 layer... maybe like this.

top layer - signal routing
inner layer 2 - split planes, gnd/+v/-v
inner layer 3 - split planes, gnd, speaker out
bottom layer - signal routing

Also, if needed, there seems to be room on the top
and bottom layers to add a 250 mil bus for +v/-v for
extra redundency..

Do you think this is a bad idea using planes?

 

[Jeff R]

Thanks for the comments.

You get really nasty sharp charging current pulses
Is this caused by any type of capacitor or do these
low inductance type of capacitor cause more problems than generics?

It will happen with any type of capacitor. As the capacitance increases, the voltage drop due to discharge between charging cycles will be less, thus the current will have discharge in a smaller amount of time. So, you may get large current spikes. For example, if you use a lower amount of capacitance, the voltage might discharge by 1 volt over 1ms, and charging current will thus flow for about 1 ms. If you up the capacitance such that the capacitor only discharges by .1 volts, then all of the charging current will have to flow during a time period of around 100 us. So, you get higher currents flowing for shorter times, and this can lead to EMI problems.

Also, when you have lots of capacitance and you first turn your amp on, all that capacitance will look like a dead short for a while. So, in addition to your house lights dimming a bit, you will be stressing your power transformer and any fuses you have, as well as possibly your interconnect wiring. In fact, you will need to be sure to use larger gage wire to prevent dropping extra voltage in the leads during those brief, but high current pulses.

In short, higher capacitance in your power supply is not always a good thing - there are limits.

I agree with janneman - 50,000uF is probably way more than enough, unless possibly you have a very high powered amplifier. I do, however, like the idea of distributed capacitance on the PWB, though thousands of uF's may be a bit excessive. A few hundred uF is plenty for the voltage gain stages, and you might want to connect the collectors/drains of the output transistors directly to the main filter capacitors, thus reducing the need for high amounts of distributed capacitance on the PWB.

My printed circuit board may be 4 layer... maybe like this.

top layer - signal routing
inner layer 2 - split planes, gnd/+v/-v
inner layer 3 - split planes, gnd, speaker out
bottom layer - signal routing

Do you think this is a bad idea using planes?

Leach's original amplifiers, up to Version 3, used a ground layer. The top layer was ground and the bottom had the signal traces and rail voltages. When I got my version 4.4 boards from him that didn't have the plane, I asked him about it and he said it didn't seem to make any difference, so he stopped using a ground plane.

Theoritically, I suppose planes would be a good thing. Anything to lower impedances is good, though I do not know if the results would be audible. If you have the money to do it, why not? Can't hurt (unless you route sensitive input traces near noisy traces - just be careful with your layout and things should be fine).

 

[thylantyr]

unless possibly you have a very high powered amplifier

The Holton 1kw (AV800 modded) amp was tested at 1600w @ 2 ohm short term with 20 output transistors per channel using
+110v/-110v, but I read in the archives, Holton said
.....loaded -> the power was closer to +96v/-96v I think,
pushing the transistors to the edge of ratings.

I want to try 28 output transistor if the buffer
stage allows..... ambitious goal. Don't know
if it will work until I try it. (hehe).

The reason I'm using 2 planes is for current delivery.
I don't have enough room on the PCB to run big fat
traces for power distribution. I dedicated room on
the top layer of the pcb to accomodate a 1/4" x 1/4"
copper bus bar bolted to the pcb, one for each supply,
extra redudency not really needed and perhaps,
for the "cool" factor like old school Pheonix Gold amps.
.. looks perty ...

I'm trying to layout a board that minimized point to point
wiring so it's clean like a car audio amplifiers where
everything is mounted on a single pcb.

I found a phat PCB mounted relay from Magnecraft,
30A rated contacts for speaker out.... 2"L x 1.35"W x 1"H
and found some Molex 50A (per contact) power connectors for my
PS connections.


Hobbies are bad....bad... bad...

 

[Jeff R]

unless possibly you have a very high powered amplifier

The Holton 1kw (AV800 modded) amp was tested at 1600w @ 2 ohm short term with 20 output transistors per channel using
+110v/-110v, but I read in the archives, Holton said
.....loaded -> the power was closer to +96v/-96v I think,
pushing the transistors to the edge of ratings.

Well, dang, that certainly meets my definition of a "high powered amplifier" !

I would use some large computer grade main filter capacitors then add in a bunch of distributed capacitance all along the bank of output transistors. I think bus bars would work better than power planes, as you say. The planes might be good for the low power VA section, but your idea to use big copper bus bars for the high power parts is excellent. And cool.

Have fun.