What value of cap for opamp PS bypassing?

[overmind]

For opamp power supply filtering and bypass, what value of electrolytic cap should be used? I had done a search, but was not fruitful, thus posting the question here.
Considering the opamp is to be used in CDP filter/output stage, pre-amp, filter & phono stage.

I have come across recommendation of using 1000uF for this application. I have seen designs that use 4700uF of electrolytic cap. There are datasheets that indicated 2.2uf Tantalum caps to be used. In most of the CDPs, 100-220uF of electrolytic caps is usually used.

I presumed that the value of the cap should be as large as possible. Besides the limitation of space, cost, etc, what would the optimal value of electrolytic cap to be used for this application? As in further increase in capacitance value does not bring about much significance improvement in performance. Of course, a film cap/s by-passing should be added on top of the E cap.

What if the bypassing is done without the Electrolytic cap, a large film is used instead? What value of film cap should be use?

Does anyone with experience in this aspect that can share with me?

Thanks in advance
Jack

 

[paulb]

It depends on your power supply. If it is already well filtered, small 0.1 uF caps right at the op amp terminals is fine. Make sure the grounds from these caps go by a separate route to the power supply (don't connect them directly to the signal ground).
If it is a power op amp, or driving a low impedance (< 600 ohms?) load, you may want some higher capacitance in parallel, still fairly close to the op amp. But I really don't think there is any good reason to go over 100 uF per cap. If the power supply needs filtering, that is a separate issue.

 

[tvi]

UltraCAD have a page you might want to look at <a href="http://www.ultracad.com/esr.htm">ESR and Bypass Cap Self-Resonant Behavior</a>
They have a small programme to work out the impedance response of a large number of parallel RLC circuits (just 3 values in the demo).

Hope this helps

James

 

[Chucko]

I'm POOGEing some musical instrument gear that contains about a dozen dual op amp chips. Unfortunately there's only a ground plane around the first stage, and elsewhere no convenient way to run bypass caps to ground without dirtying up the signal grounds, so for the moment I've got .01 uF caps from rail to rail (the standard +/-15V) at the socket pins. I do have some 4.7 uF tantalums from the rails to ground where the power supply enters the board.

Does anyone see any harm to this arrangement? It doesn't display any obvious signs of oscillation, which actually surprised me because I'm using a 90 MHz GBW amp for the variable gain and tone control stages!

When this project is done, I should have enough high quality chips left over to do a proper ground-up design (pun intended!).

 

[paulb]

Chucko,
I've seen some discussion somewhere (on this forum, maybe?) that says your arrangement is actually superior. But I'm hardly an expert.