Power Supply Filter Question(s)

i'll try to give a brief description:

my circuit uses lots of discrete parts, and i wish to remove some where i can. after looking at my current board i think i can do w/o some caps on the supply side.

power is automotive power, comes into a muRata filter and then through a 7805 +5v regulator. within 1mm of the 7805 output is a 220uF electrolytic and a 1k pF ceramic for filtering, however, my current circuit uses more caps near each IC to help reduce EMF from automotive environment. within 1mm of each IC +Vcc i have a 1k pF ceramic and a 10k pF poly cap to reduce Vcc noise. and when i say "noise" i mean to guard against noise. i didnt actually see noise on the scope, etc.

1. is having the extra caps next to each IC really needed? or perhaps i only need the 1k pF ceramic? any suggestions?
2. my initial design failed to include a -V for the op-amps, so i plan to use a very simple circuit with a Intersil 7660 to get my -V. 7660 has a inernal oscillator and currently runs at 10kHz. should i also cap the -V near each IC to reduce noise? the 7660 +Vcc is fed from the 7805.


thanks in advance for any replies....

I think it's standard practise to use capacitors on all power pins for IC's to protect them from voltage surges. Lets say your circuit gets affected by a nearby magnetic field then you definitely want capacitors on your power pins so the IC's dont momentarily fail due to voltage level fluctuations.

The caps at each IC guard against much more than "voltage surges". As you appear to know, auto power is notoriously "noisy". Some of it comes in on the power, some of it is radiated through the air, and some of it is actually produced by the chips on your circuit board. I think some of it is actualy voo-doo.

Some of it is easy to see on a 'scope, some of it is not. The noise you need to worry about the most is the noise that you can not see, because I guarantee you, it is there. And your chips are better at seeing it than you are.

If you wind up having a problem with noise, you can easily spend endless hours, exploring paths you would never expect, and in the end not being able to fix it. Or it will appear "fixed" until you need it the most. Then when you try to troubleshoot the problem, it disappears...

The only way to fix it is to take more precautions than you would have thought were necessary. Filter your incoming power (you do); Filter your power supply (you do); Put in a sheilded box (do you?); Bypass at each chip (you need to keep doing this); Then, maybe say a little prayer when you mount it in an auto/plane/boat.

Bypass caps are cheap insurance. Use them.

All the capacitors are necessary - especially the 10 uF caps on the 7660. The one on the input side furnishes power to the charge pump, and the one on the output is the filter for the charge pump.

Get data sheets for both those devices. They will explain the function and need for the capacitors.

i built 7660 supply as noted in the datasheet. however, there are lengths of traces across the board to the from 7660 to the op-amp IC's, etc. so i should perhaps also cap the -V pin of the op-amps just as i am doing to the +V pin?

Yes, the negative voltage is generated by a switching process, so there is going to be some residual noise that the 10 uF tantalum can't entirely pass off. A .1 uF ceramic is good.