Multiple Op Amps - decoupling caps?

[kanurys]

Hi guys, this is my first post here. I'm very much a novice electronics nerd compared to many of you so maybe you can answer this noob question kindly for me.

I'm working on an analog signal conditioning circuit for some RTD temperature sensors. I have 3 sensors and 3 conditioning circuits, each containing an LM324N quad op amp dip package for a current driver and some signal scaling. All 3 sections are powered from the same regulated 12v rail and ground.

The question:
Should I put  0.1uF and 10uF decoupling caps near the power pins on each dip package or just somewhere on the 12v and ground rails - location importance not withstanding? If each LM324N had a set of decoupling caps, wouldn't the capacitance all add up since they're in parallel on the 12v supply, resulting in filtering an unintended frequency range?

Thanks for the help!

[caffeinatedbard]

Decoupling caps should get physically near the power pins of each DIP package.


Don't think so much of capacitance adding up because they are in parallel but rather  think about as that you are providing a low resistance path to ground for high frequency noise, so that's why you should have at least 0.1uF on each power pin.

Hope that helps.

-bard

[c4757p]

If each LM324N had a set of decoupling caps, wouldn't the capacitance all add up since they're in parallel on the 12v supply, resulting in filtering an unintended frequency range?

Couple of points:

- The point of putting a cap on each is that the power/ground connections aren't direct connections per se, they have enough resistance and inductance to be poor conductors at high frequency. The capacitors aren't in parallel because this inductance separates them. Location is very important - it's the whole point.
- The point of using multiple values on each is that higher value capacitors typically have worse performance at high frequency. 0.1uF paralleled with 10uF is indeed 10.1uF, but with increased high frequency performance because of the separate 0.1uF cap.

[kanurys]

That all makes good sense. Thanks for the responses. I understand about the inductance and resistance of traces, so I see your logic. I'll design the boards with that and proximity in mind.

[KJDS]

Lay the board out to put both caps on each op-amp. In practice it may work better with just the smaller cap fitted by each op-amp and a single large one on the board. It's much easier to not fit a component on a board that is laid out for one, than it is to try and fit a component when the board isn't laid out for it.

[kanurys]

Thanks everyone. The design came out great and clean scaling of the input signal is what I got!

[David_AVD]

Usually I'll put just the 100nf (0.1uf) caps right next to each op-amp and a pair of 10uF per few chips.

[miceuz]

I'd say LM324 is not a very good choice for this application. I suppose you are using it as a current source and your readin precision depends on the offset voltage. You can calibrate that out in software, but to do that for three different sets seems to be a bit tedious. If you choose an opamp with low offset voltage, you might get more consistent results.

I've made a PT100 thermosensor conditioning circuit according to microchip app note http://www.microchip.com/stellent/idcplg?IdcService=SS_GET_PAGE&nodeId=1824&appnote=en011700 I've just slapped in any opamp I had in a bin (324 or 358 don't remember now) and later replaced it with low Vos one to notice the huge difference.

Here are the schematics and board files if you are interested - https://github.com/Miceuz/reflow-oven-controller hope this is still relevant to you.