A noise of ceramic capacitors?

Leaky or cracked capacitor? With a 200k impedance, leaky caps will cause issues. Could you have a contamination problem on the pcb? I have come across ceramic caps and pcbs with low megohm leakage problems. Which reminds me, I gotta install the leaky cap in my Tek

Mark

Maybe air currents causing temperature variations, which vary the capacitance. If you have a capacitance variation, and a bias voltage, you have voltage variation, since the charge remains constant. A C0G type should be much better, but not cheap or small.

Best regards, Spehro Pefhany

I know you already said - but are you absolutely sure it is not microphonic? I have just had to replace some X7R's with tantalums, in critical areas of a circuit. They were definitely badly microphonic - e.g. mV-level pulses when the circuit board was tapped. I think this extends to down DC, i.e. a strain in the body of the capacitor would produce a level shift (in absence of any other currents).

Other than that - X7Rs have quite a large temperature coefficient the capacitance changes so will the voltage, at frequencies above 1/2piRC. If there are fast temperature fluctuations I guess this could be a problem too.

For what it's worth...about 15 years ago, I was working on the design of a circuit where I used a ceramic cap, 1206 SMT 0.1uF X7R, for AC coupling, but in a different arrangement than usual. I discovered a 1/ f noise that was not apparent until you got below 10Hz or so, due to the capacitor. It would not have been microphonics, and unlikey due to temperature variations, as the spectrum was too cleanly 1/f. (It takes quite a while to get statistically-significant spectra at such low frequencies...) Polyester film cured the problem. It only happened with a bias across the cap, but the amplitude of the noise was nearly independent of bias voltage above a low level, around a volt as I recall.

Cheers, Tom

Hmm, I did a laser diode controller earlier this year where caps were in a PID loop. Provided extra pads for film caps just in case. Found no difference between ceramics and film. And this thing is ultra-sensitive down to the uA level, resp, uV level at the caps. Noise in the sub-10Hz range would have killed things but it didn't happen. Could there be a bad lot or brand of caps involved here?

What was the p-p amplitude of the noise? Are you talking about 10s of mV or uV level?

If you're talking about 10S of mV level then I'd scrutinize the measurement method. Is the voltage being measured directly across the suspect capacitor? The fact that the film or tantalum cap only helped but didn't competly cure it seems to eliminate the micrpphonics as a possible cause. Possible low frequency coupling into the high impedance node?

If your in the uV range there a re a variety of issues that could cause that. rapid temperature swings of a few degrees C. Low frequency coupling (electro or magnetic) into the high impedance node. Ceramic capacitors have been found by some of the analog gurus at my company to have relatively high levels of "popcorn" noise when used in high gain amplifiers. Glass or teflon solved the problems.

Ceramic caps vary their capacitance with applied voltage which may cause you some issues and might show up as phase noise, which would be more apparent with low frequency, large signal swings. Much depends on the rated voltage of the device. A typical X7R will have 50% rated capacitance at rated max voltage.

Cheers

PeteS

What bias voltage did you have on the caps? The OP has a huge bias voltage-- 2,500,000 uV, so a mere 10ppm change in capacitance means

Best regards, Spehro Pefhany

Nope, does it with different lots, different brands, all at about the same level. But you would never notice it in most circumstances. I can't imagine you'd see it in a controller.

Cheers, Tom

You basically have metal on ceramic junctions in that thing, it must be a low frequency distribution on the leakage current. Switching to metal on film eliminates the troublesome physics altogether.

Roughly a volt, so in the same ballpark. Temperature changes are really slow. Got to make sure this isn't in any area of turbulent airflow. But ok, since mine was in a PID loop slow changes are muffled. However, the host can hold this board in software lock where the PID is disabled and that's much slower. No problems there either. We noise-tested the heck out of this design.

We would at least see it while it's in software lock or hold mode. With a laser diode that is wavelength-steered and in this application a few microvolt change on that cap would have thrown us out of the range where the whole thing worked at all. IOW a red light would have come on.

If the noise was inside a loop correcting down to DC and there was a lot of gain between the integrator and the error detector, it would have been suppressed by the feedback.

Yes, but as I said I had designed in a feature that allowed the SW to stop the loop and SW-control the whole thing. Which my client ended up using more extensively than we all thought, they kind of got a free production tester out of this whole design. No noise problems there either.

None of this is unusual. Ceramic caps are cheap and have relatively high capacitance for the size but the ceramic dielectric is NOT the most stable stuff in the universe. They are prone to lots of noise especially when there is considerable leakage. The dielectric in the higher value ones are worst. If you want a pure capacitance rather than a capacitance in parallel with a noisy leakage resistance use a capacitor with a decent and stable dielectric. These include mylar and other films, mica, glass, or any other stable dielectric. Note that this does NOT include tantalum. Though apparently your tantalum was better than the ceramic ones. The trade-off of course is that capacitors with better dielectrics than ceramic or tantalum are much larger and often more expensive too. Sometimes MUCH more expensive. If you want more details about this sort of thing, talk to some of the analog guys who design high end mic pre-amps and the like.

series

capacitor?

Microvolts are large compared to what I saw. The circuit in which I saw the effect put the capacitor in an unusual electrical environment in which the effect was observable; it would not be in any common circuit configuration I can think of. I regret that I'm not at liberty to tell you more about the particular configuration. It's possible that Spehro's comment about thermal variation would explain it, but as I recall there were reasons I thought that was not the answer (like, the amplitude didn't vary nearly enough with changes in DC bias).

Cheers, Tom

WHAT tantalum one?? -- Leakage current is an interesting suspect for the noise I saw in the X7R ceramics. If I get a chance, I might check that out. On the other hand, lumping all ceramics together as "bad" is not very useful. C0G caps are available in reasonably large values these days, and are better in many ways than Mylar/polyester film caps. Maybe I can start some leakage tests on various ceramic dielectric caps, to go along with the ones I have running on metalized polyester and polypropylene ones. Both of those are showing longer self-discharge time constants than you would suspect from the manufacturers' ratings.

Cheers, Tom

series

capacitor?

Awww, now you held a delicious piece of crispy bacon under all our noses and then pulled it away ;-)

Joerg,

Didn't the shaman tell you that gluttony is one of the sins?

VLV