decoupling caps on a big pcb

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

radiance

Well-known member
Joined
Jun 4, 2004
Messages
3,228
Location
the Netherlands
I'm designing a pcb for an active summing mixer. It's quite a big project and the pcb meassures 35cm by 20cm.
All opamps have the usual 100nF decoupling caps but would it be wise to add some bigger caps as well (electrolytic maybe) because of the size of the pcb?
....and what value should they be? I read somewhere that it should be 100 times the value of the small (in this case 100nF) decoupling cap. So that makes 10uF....hm, I was more thinking of bigger caps, more like 220uF or so..

Any thoughts??

Thanks.
 
Should be at least 100 times. But hey, that's just a rule of thumb. I seriously doubt you will actually see any issue using caps within the x100 range. That RUT is really based on the inductances in surface mount parts so that there is no strangeness in decoupling, see Dr. Johnson's work on this.

With that being said, I would look up radius effect for capacitors. It would be something like 100nf between both rails to ground, then 100nf between the rails themselves then bulk capacitance within 1-2 inches. You then want bulk cap every few inches along the rail traces and then lots of bulk cap (1000uf per amp) at the power supply.

You can't go wrong with lots of bulk cap in a big design. Too many designs don't have enough and get bad reps because of it.
 
Without getting too esoteric, keep in mind that "ALL" pcb traces have real impedance. Hangings a lot of capacitance between PS rails and ground is not magically shunting PS noise to some infinite current sink but coupling the PS rail at that node to the ground rail at that node, so PS noise voltage is coupled into that ground node as a current and voltages can be developed based on current flow and trace resistance.

If you keep a power ground system separate from signal ground traces you can ecounter situations due to compliance between the two. I once had a console take off with a significant 960kHz signal between the two ground systems when the console was installed in a studio about a half mile from an AM tower in the main lobe. :oops: If you use multiple ground systems keep them well coupled at HF and keep in mind that opamps are 5 terminal devices.

In general when routing signals around inside consoles which can be 6' long, ground is a local or relative concept anyhow. The differential amp is your friend. :grin:

JR
 
Thanks John. I think I don't have to worry THAT much since this summing mixer will fit in a 4 unit 19inch rack case.

As it is now I've a lot of extra cap space on the pcb. Once they're finished I'll start only with the 100nF decoupling caps and see how everythin behaves.
I can always add more caps later on when I feel like it.

It's just that I want to be really sure about every aspect of mixer design before I send the pcb files to the developer. It's also my first real pcb design and since it's quite big it feels like a gamble....
 
Well you always want bulk cap too, not just the 100nf decoupling. You might hear people say "muddy" or "smeared" a lot when referring to cheap units. A lot of times that is lack of proper decoupling schemes and not the opamps or DC blocking caps at all..

Why does an SSL full of 5534 opamps sound good and a Behringer unit full of 5534s sound terrible?
 
Back
Top