Rochey

decoupling caps
« on: November 03, 2010, 05:19:36 PM »
Those of us who try to design circuits have mostly been raised with the mantra of:

"0.1uF ceramic and a 10uF/47uF electrolytic should be used for most devices".

In many cases, I see my customers (pro audio and consumer manufacturers) using mostly 0.1uF only, then a few large caps on the board as "reservoirs".

My question is this:

Could you replace the elctrolytic with a large value ceramic?
If you did so, would you still need the 0.1uF?

My gut says yes - I.e you coudl replace a 10uF 'lytic with an equivelent ceramic.
However, I think you'd still need the 0.1uF.

Thoughts?
Expat Audio Home: http://www.expataudio.com


jdbakker

Re: decoupling caps
« Reply #1 on: November 03, 2010, 05:34:20 PM »
Could you replace the elctrolytic with a large value ceramic?

Generally: no. The ESR of the electrolytic cap (shared by a couple of parts) serves to dampen resonances of the power net. You could use a 10u ceramic cap with a series resistor, though.

If you did so, would you still need the 0.1uF?

Only if you want a circuit that's not ringing and/or oscillating.

JDB.
[the essence of the 0.1u cap is that it be as close as possible to the chip in question. That mantra needs to be reconsidered for anything faster than a 5532, BTW]

PRR

Re: decoupling caps
« Reply #2 on: November 04, 2010, 01:07:22 PM »
> replace the elctrolytic with a large value ceramic?

Until quite recently, a 10uFd ceramic was absurd.

Remember the power pins ARE inputs. They should be low-low-gain. But in several popular topologies the PSRR drops to nothing at a MHz. Junk on the rail is junk on the output. If another circuit is kicking the rail, that can sneak around and cause outright oscillation or worse, undetected instability and fuzzy transients.

You need a SMALL CLOSE cap to take the MHz spikes. When working at the edge you want one AT each chip. In many cases one between adjacent chips works fine.

You need a LARGE cap to absorb KHz wobble, though this does not have to be close and may be an electrolytic per handful of board.

Electrolytics have got much better and many smaller slower audio systems do OK with a few good electrolytics within a few inches. There's some super-fast "audio" chips, DSL drivers, which won't be stable without screws and nails and glue on their power pins.

As JD says, there's such a thing as "too good". The circuit between the cap's legs is finite length and thus has inductance. Inductance and capacitance is a tuned circuit, happy to ring. At VHF it does not take much series resistance to damp the ring, but ceramics have really low series resistance.

What are you trying to do? Can you save space? Is space essential? Or are you trying to build for eternity?

johnR

Re: decoupling caps
« Reply #3 on: November 04, 2010, 04:32:52 PM »
Large ceramics have an annoying tendency to short circuit and let smoke out. It's possible to get fail open types (which are the only ones I consider using) but they're not all that common.

Rochey

Re: decoupling caps
« Reply #4 on: November 08, 2010, 10:51:00 PM »
What are you trying to do? Can you save space? Is space essential? Or are you trying to build for eternity?

I'm more or less trying to ease assembly. Manufacturers I work with are moving to ceramic only caps because

1 - they are easier to pick and place. Just another reel of 5000 caps, next to the 0.1uF's.
2 - PCB size reduction
3 - Cost reduction. A 4K reel of 10uF caps is around $10 in single reels, versus chemical caps costing much much more.
4 - Lifetime. One less thing to fail.

Thats the theory i'm getting back. Just wondering what the downsides may be.

/R
Expat Audio Home: http://www.expataudio.com


 

Related Topics

  Subject / Started by Replies Last post
21 Replies
3791 Views
Last post April 17, 2008, 01:02:56 PM
by mkoijn
11 Replies
2492 Views
Last post May 29, 2008, 09:26:12 AM
by Harpo
5 Replies
1844 Views
Last post August 23, 2008, 04:00:02 PM
by Svart
11 Replies
658 Views
Last post March 29, 2020, 07:50:14 PM
by BluegrassDan