Replace overkill caps with even more overkill caps?

[hansibull]

Hi!
I got a subwoofer amplifier (3x LM3886) that's driving a relatively large and long stroke 10" woofer. I got 6x 10000uF Nippon Chemicon caps for the power supply.
I also got 6x 22000uF Nippon Chemicon caps lying around collecting dust.

Should I replace the small caps with the big ones? I know 6x 22000uF is overkill, but will it provide better/worse sound in any way? what about combining them?

 

[hangar18]

naah 6x10000uF is already enough, keep the other caps for other projects, generally speaking the rule of thumb is to use 1000-2000uF per amp. Also if there isn't any hum or lack of bass, the power supply is good, the sound largely depends on the amplifier itself, the power supply plays a crucial role until some point when afterwards the sound quality remains the same.

 

[sreten]

Hi, Overkilling overkill is utterly pointless, and probably worse, rgds, sreten.

 

[hansibull]

I've planned to upgrade the amp some time and maybe include amplifier modules for a pair of stereo speakers too (using 1x LM3886 for each speaker). Is 6x 10000uF still enough?

I've read somewhere that these amplifier IC's have a peak Amp rate at ~9Amps or so.

 

[hangar18]

peak not continous current, that's why caps are put in power supplies. to help the PS withstand those peaks, otherwise the sound will get distorted. anyway 6x10000uf are more than enough.

 

[jcx]

I've read somewhere...

you mean the datasheet?

 

[hansibull]

you mean the datasheet?

I read that somewhere at the Chipamps forum some time ago. I can't find the post on the go, but I can't seem to find it it the datasheet, only peak wattage.

 

[jcx]

look at the graphs, SOA shows current*time limits, pulse power limits graph too

 

[ticknpop]

I disagree with Hanger 18, I currently run over 700,000 uf in my power amp with chokes between the first 600,000 uf and the final 100,000 uf. It sounded better (more bass depth and control, greater smoothness to the sound each time I added a couple of hundred thousand uf.

Real cahones kind of bass - direct coupled DAC , amp, and preamp all help too.

 

[jcx]

you could try applying engineering principles to the design - find the I*t capabilities of the chip amps from the mentioned datasheet graphs, allow for the ps C dv/dT between supply charging pulses...

 

[googlyone]

You can work out for yourself what rail droop you will experience in a mains powered linear supply.

If you are in the USA you have 1/120 second between charge cycles from the mains. Your load / amplifier Ilimit tells you the maximum current draw, which does need you to apply some common sense in terms of peak currents as opposed to the current the power supply needs to deliver long term.

The relationship C = Q/V. Is your friend

Differentiate this and you get dV/dT = I/C

You know dT as this is as we have established 1/120 seconds.

You know I (max) as you have this from your design

Therefore you have the voltage droop - notwithstanding your need to consider transformer and diode drops.

Stick it in a spreadsheet and play.

How low does it need to be? Your choice really, but getting your head around how "diminishing" the diminishing returns can be just makes sense.

Then I usually use a bigger cap, just to make me happy. But at least I know it is "ok"..

 

[Michael Bean]

In addition to the "point of diminishing returns" thing, larger amounts of capacitance have a "dark side" too. The rectifiers may need to be upgraded to handle the larger charging currents. And the charging spikes, while being shorter in duration, will be much greater in amplitude. Too much in this case could be worse than not enough.

Mike

 

[jcx]

eventually too big C requires inrush protection staged/timed turn on circuitry as well - although cheap toroid xfmr core saturation may be the bigger inrush problem

 

[hangar18]

I disagree with Hanger 18, I currently run over 700,000 uf in my power amp with chokes between the first 600,000 uf and the final 100,000 uf. It sounded better (more bass depth and control, greater smoothness to the sound each time I added a couple of hundred thousand uf.

Real cahones kind of bass - direct coupled DAC , amp, and preamp all help too.

As i said 2000uf per amp is already an overkill and wont lead to any improvments in sound. Nowadays it's possible to make amps with 0.001%THD at full power, im pretty sure no one can hear a diffrinence between 0.5% and 0.001%, but speakers aren't so good, and that deep bass you are talking about is coming from the speakers not the amp itself. Most of the speakers have atleast 2% distortion and even more, they are inefficient, and any change in the amp can lead to an overall change of sound.