Multiple parallel caps in a PS is there a downside?

[steven2583]

I'm building a new power supply for my sound processor's dac daugher card. I have an extra transformer around and some caps. I have a few at 1000uf, 2200uf, and 470 uf. I would like to bet about 3000-6000uf since the ps will support about 5 DAC AK4395 dac chips. Is there any downside in having multiple caps in parallel feeding the regulator. Is it better to go out and purchase one larger cap per regulator?

 

[DF96]

Caps in series add ESR, and usually need voltage balancing resistors. You will need a lot of 1000uF if you are going to use them in series (and parallel) to get 3000uF. Remember, caps in series reduce their net capacitance.

Did you actually mean caps in parallel?

 

[steven2583]

I misstated. I want to use them in parallel.

 

[DF96]

Caps in parallel are fine. Some people do that as a matter of habit, as you can get reduced ESR and inductance. Just be careful where the ground currents go.

 

[nigelwright7557]

Caps in parallel are fine. Some people do that as a matter of habit, as you can get reduced ESR and inductance. Just be careful where the ground currents go.

Valve amplifiers often have 2 or 3 stages of RC to get very low voltage ripple.

 

[steven2583]

Is it better to have one larger cap or multiple small ones? I'm thinking using ELNA Silmic II and a Nich Muse cap.

 

[AndrewT]

How about asking the Forum to change the title of your Thread?

 

[tinitus]

title changed to concern paralel caps instead of series connected


moderation

 

[Andersonix]

Try some inductance CLC in there for low ripple.
Downside? Running out of space for more caps inside the chassis.

 

[abraxalito]

What he said plus all the time it takes soldering them all together. The upsides in SQ are generally more than an adequate reward for the time spent IME.

 

[thoglette]

Try some inductance CLC in there for low ripple.
Downside? Running out of space for more caps inside the chassis.

Most important to know that L will exist. If not planned for you can get resonances which can be very hard to damp. If you deliberately plan for the L (ie add L of a known value and use it or put R in as well) then you can get good results

 

[TonyTecson]

there is just one downside, you need more space....