Guard Rings on PCBs

For DC leakage, probably not.

We just did a bunch of boards as test cases for various no-clean fluxes. The worst one so far was 0.3e12 ohms between all the alternate pins on a d-sub, after a 100% humidity soak.

What part is involved? How much does leakage matter?

I do sometimes actively guard a part or a pad *from below* with an inner-layer pour, to reduce capacitance.

You can bend the critical pin of a DIP package out into mid-air.

It's a NE5532; using it as a preamp for a dynamic mic instead of the previous choice of IC, which as you may recall was widely denounced here.

The datasheet does actually mention this technique. I'd have thought 'guard rings' and capacitance issues were the domain of RF, not LF audio!

Absolutely no reason for guard rings on that part.

THanks, that'll save me a huge amount of swearing. Now, these DILs have two op amps per package. Since I probably won't be able to screw enough gain out of just using one of them, how do you feel about using both in series to end up with ~0.5V p-p audio signal out of the 2nd one? Will there be a problem with cross-talk?

The signal won't cross-talk to itself.

Balance the gains maybe, as in 10*10 = 100.

Screw that. I could get more out of a single BJT! I'm expecting a good

4000+ *minimum* out of a pair - a hundred's no good to me. 100 out of both in series with each other?? I'm hoping you missed a couple of zeros off your figure here. Please tell me it was a typo.

That was just an example of how N*N = N^2, which maximizes bandwidth. I have no idea how much gain you want.

Don't be obnoxious to people trying to help.

Cursitor Doom wrote: =================

** Dynamic is too vague, nominate the actual mic. Balanced mic line or not ?

...... Phil

THAT was being obnoxious?? <baffled>

Anyway, you've already helped so don't sweat it.

Cursitor Doom wrote: ==================

** Listen you fucking NUT CASE !!!

Cascading the halves of a 5532 with a total gain of 4000 times at 20 to 30 kHz will *guarantee* HF oscillation. Even guard tracks will not help.

WTF do you need so much gain for?

Wanker.

..... Phil

Yes, and I'm not surprised - even though you often call others out for being unnecessarily unpleasant.

/This/ time anyway. Next time?

Oh, you still have many ways to screw it up.

It would if Cursitor Doom did it. You do have to very careful to deal with all the possible feedback paths. At those frequencies there isn't a lot of power supply rejection, so you have to decouple the power rails very carefully, otherwise the current drawn from the power rail by output device will feedback into the input stage of the input device though it's power supply pins, with enough gain and phase shift to guarantee oscillation.

But if you are careful you can do it, and get there by design. I've done it.

They aren't enough, on their own. A ground plane underneath the parts can be helpful, if you do it right, but you do have to kill feedback through the power rails too.

Cursitor Doom may aspire to be a wanker, but he probably couldn't even get that to work.

All covered comprehensively in the layout suggestions in the datasheet, Bill. I read it. Did you?

My, someone's a little cranky today! I'll just put it down to the side-effects of the shots you had. :-D

As always, an inaccurate and inappropriate conclusion.

Many years ago. Not that I've ever used the 5532. The proposition that you could cascade both halves of one 5532 is obvious nonsense, since you are stuck with one pair of power supply pins for both devices and the devices don't have the power supply rejection ratio at high frequencies to let it work. Cascading two 5532 amplifiers from two different packages could work if you decoupled the power rails carefully enough - but that does mean that you have to undertstand what the data sheet is telling you.

Low equivalent series resistance by-pass capacitors close to the power supply pins probably wouldn't hack it. Separate capacitors for each device, and ferrite beads to minimise any high frequency current fed back into the power rail would be closer to what it takes, and the 5532 layout suggestions don't go quite that far.

They do have to be read by somebody who understands what they are saying, and you don't seem to be able to do that.

Neither the ONSemi nor the TI, nor the original Philips datasheets includes a graph of power supply rejection ratio against frequency. A professional device, such as the Linear Technology LT1028 does, but the audio market doesn't get that kind of service. The audio amplifiers don't perform any better.

It's only a newsgroup. It ain't life.