Author Topic: Non microphonic, 1uF SMT cap in a high Z circuit?  (Read 2577 times)

0 Members and 1 Guest are viewing this topic.

Online dmillsTopic starter

  • Super Contributor
  • ***
  • Posts: 2093
  • Country: gb
Non microphonic, 1uF SMT cap in a high Z circuit?
« on: March 22, 2020, 11:24:15 pm »
So I have a dc removal servo integrator (1M ohm, 1uF caps), and I am after building this thing SMT, question is, what to use for the integrator cap?

I need something non microphonic (critical, which rules out class II dielectrics) and low leakage, and I don't know which polarity will be applied.

Current thinking is PPS film from Wima or Kemet, but I am seeking the brain trusts input on this one.

I would rather avoid reaching for banks of 10 of the (Awesome) 100nF C0G parts that are now available on cost and space grounds if nothing else. The other downside of these is that they are not generally available as soft term parts and in 1206 that is not a good thing.


 
 

Online Conrad Hoffman

  • Super Contributor
  • ***
  • Posts: 2078
  • Country: us
    • The Messy Basement
Re: Non microphonic, 1uF SMT cap in a high Z circuit?
« Reply #1 on: March 23, 2020, 02:46:52 am »
Not sure but PPS is probably OK. Is there any way to raise the impedance further, say 10X and use a smaller cap?
 

Online dmillsTopic starter

  • Super Contributor
  • ***
  • Posts: 2093
  • Country: gb
Re: Non microphonic, 1uF SMT cap in a high Z circuit?
« Reply #2 on: March 23, 2020, 05:23:39 pm »
10 Meg is getting annoying from a offset current perspective even with the excellent chopper opamps now available, also getting into a space where board cleaning is becoming critical.....

I could, but I would very much rather not as a gut reaction, might be worth some spice modelling tonight to see what I get however, as 100n I use elsewhere on the board in C0G.

Regards, Dan.
 

Online Conrad Hoffman

  • Super Contributor
  • ***
  • Posts: 2078
  • Country: us
    • The Messy Basement
Re: Non microphonic, 1uF SMT cap in a high Z circuit?
« Reply #3 on: March 23, 2020, 07:51:53 pm »
I think dielectric absorption is the thing to look at, and it's also the thing with the least data available.  >:(
 

Offline DaJMasta

  • Super Contributor
  • ***
  • Posts: 2393
  • Country: us
    • medpants.com
Re: Non microphonic, 1uF SMT cap in a high Z circuit?
« Reply #4 on: March 23, 2020, 08:25:38 pm »
Two back to back 2uF tantalums?  Even if you don't know the polarity, can you bias it in a way that whatever the eventual signal is, you have a set polarity, and you can just use a small electrolytic or tantalum?


While there are plenty of film caps and stuff that would fit the bill, I think the ones small enough to be smaller than the proposed array of C0G caps probably are pretty pricey (though maybe not in comparison of 1uF worth of C0G caps).  Would stacking the C0Gs be viable/give you the footprint you were looking for?
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 15157
  • Country: de
Re: Non microphonic, 1uF SMT cap in a high Z circuit?
« Reply #5 on: March 23, 2020, 08:36:57 pm »
The tantalum caps would be really poor dielectric absorption and leakage is usually also not that great, though hard to tell apart from DA.
So I don't think this is still an option.

In some case, when the DC range actually used for the DC servo is small, one could add a divider at the servo output. This effectively acts like a capacitance multiplier and even reduces some noise sources.

The PPS caps are probably a good option. The DA may not be as good as with C0G, but still quite good. The C0G caps can also differ between brands and types.
 
The following users thanked this post: Someone

Offline Zero999

  • Super Contributor
  • ***
  • Posts: 20363
  • Country: gb
  • 0999
Re: Non microphonic, 1uF SMT cap in a high Z circuit?
« Reply #6 on: March 23, 2020, 09:13:29 pm »
10 Meg is getting annoying from a offset current perspective even with the excellent chopper opamps now available, also getting into a space where board cleaning is becoming critical.....

I could, but I would very much rather not as a gut reaction, might be worth some spice modelling tonight to see what I get however, as 100n I use elsewhere on the board in C0G.

Regards, Dan.
What sort of op-amp are you using? J-FET/MOSFET should be no problem with 10M, but not for BJT.
 

Online dmillsTopic starter

  • Super Contributor
  • ***
  • Posts: 2093
  • Country: gb
Re: Non microphonic, 1uF SMT cap in a high Z circuit?
« Reply #7 on: March 24, 2020, 03:39:22 pm »
It is a fet input chopper 'zero drift' part.

Most of these loops are actually just stripping offset and bias currents off other bipolar opamps to avoid switch clicks and pot wiper noises.
It is a little annoying as for myself I would probably just throw 1,000uF 6.3V electrolytics in place of all the servos and call it good, but 'no electrolytics' is a marketing point (Audio, what can you do?).

Regards, Dan.
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 15157
  • Country: de
Re: Non microphonic, 1uF SMT cap in a high Z circuit?
« Reply #8 on: March 24, 2020, 04:02:58 pm »
If the offset to compensate is small and the integrator thus only needs to provide a small voltage like a  < 100 mV, one can add a divider after the integrator and use this as a capacitance multiplier. So no more 1 µF needed and some 50-100 nF can be enough. There should also be no problem with using more than 1 M for the resistor. Noise of the resistor would effect mainly the transition region, so usually not a problem. So just using 10 M would be possible too - though possibly tricky with the bias current.
Just for audio one usually does no want a cross over frequency too low - so 1 M and 22 nF should already be sufficient. So I see no need for so much capacitance.

For the amplifiers I am not so sure an AZ OP is such a good Idea with audio, as there can be artifacts from the chopper. At least not all types would be suitable.  There are some precision FET types that are good enough (e.g. OPA196). Remember that some 100 pA of bias with 1 M resistance also give 100 µV of offset.
 

Offline Yansi

  • Super Contributor
  • ***
  • Posts: 3930
  • Country: 00
  • STM32, STM8, AVR, 8051
Re: Non microphonic, 1uF SMT cap in a high Z circuit?
« Reply #9 on: March 24, 2020, 04:26:27 pm »
1uF mylar/MKS, MKT, MKP whatever foil is fine. Just do not use ceramics (or electrolytics).  Tantalum does not like any kind of negative voltage. They are not as tolerant as electrolytics.
 

Offline Vovk_Z

  • Super Contributor
  • ***
  • Posts: 1478
  • Country: ua
Re: Non microphonic, 1uF SMT cap in a high Z circuit?
« Reply #10 on: March 24, 2020, 07:27:52 pm »
I would like to try C0G. It's can be almost as good as film with microfonic effect (and more temperature stable).
You dont need 0.1 uF parts, today we can buy 0.22 uF 1206 parts easily (even I have 20 pieces).
 1 uF needs only 4-5 caps in parallel and they take very few space.

And yes, cheap (I mean usual) tantalum doesn't fit integrator circuits. I don't know about wet tantalum (haven't worked with) but it's price and parameters are not good enough too.
« Last Edit: March 24, 2020, 07:36:46 pm by Vovk_Z »
 

Offline SeanB

  • Super Contributor
  • ***
  • Posts: 16385
  • Country: za
Re: Non microphonic, 1uF SMT cap in a high Z circuit?
« Reply #11 on: March 24, 2020, 07:55:47 pm »
Wet tantalum is not at all bad with leakage, but will need a permanent bias to keep it that way.  I did take 100 that had been in storage for around 2 decades, and measured them, half with no reformation, and half with them connected to 20V via a 1k resistor for a night. No difference really in capacitance, ESR or other parameters, just some were off the bat shown as near ideal capacitor, and others were shown with slight leakage of around 100k-1M. they were all 10uF 25v units. Do not use rubber seal ones, they only last 5 years like electrolytics, while the glass seal ones last till they corrode through from the outside. Rubber seal is cheaper, but not worth it if you are paying for tantalum. I was changing them by the hundreds.

But for low noise use polyputthekettleon, and use those in sealed housings, as the bare ones are a little sensitive to environmental parameters. Bare are cheap, and often available in SM forms, but temperature and humidity swings will cause havoc with leakage, plus PCB cleaning and solder flux residue as well.
« Last Edit: March 24, 2020, 07:58:24 pm by SeanB »
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 15157
  • Country: de
Re: Non microphonic, 1uF SMT cap in a high Z circuit?
« Reply #12 on: March 24, 2020, 08:23:33 pm »
100 nF C0G is available with soft termination.
If really needed one could consider a well placed slit to reduce the board stress.
With audio I see no absolute need for 1 µF: 100 nF should be plenty to get a cross over below 2 Hz and thus only marginal effect at 20 Hz.

If audio noise is critical one may want the extra attenuation behind the AZ OP alone for noise reasons. So one may not even need 100 nF.
With just a few 10 mV to compensate even 10 nF could be sufficient.
 
The following users thanked this post: dmills

Online dmillsTopic starter

  • Super Contributor
  • ***
  • Posts: 2093
  • Country: gb
Re: Non microphonic, 1uF SMT cap in a high Z circuit?
« Reply #13 on: March 24, 2020, 08:33:53 pm »
Been doing some spice sims, and 1M/100n plus the usual non inverting passive pole at the integrator input from the same values as an RC network, an RC network on the output of the integrator a decade above the 1M/100n time constant to remove the chopper artefacts up at 800kHz and a 100:1 divider into the inverting node of whatever I am trying to correct looks good in ltspice.

And I can do that with all C0G! There is NOTHING wrong with ceramics in audio as long as you stick with type 1 dielectrics.

Just got to build the damn thing now and we will see how we go.

Thanks for the input guys.
 

Offline Yansi

  • Super Contributor
  • ***
  • Posts: 3930
  • Country: 00
  • STM32, STM8, AVR, 8051
Re: Non microphonic, 1uF SMT cap in a high Z circuit?
« Reply #14 on: March 24, 2020, 08:51:30 pm »
Iam looking forward for your 1uF and 100nF in C0G.

Expensive nonsense, if you could find any.

1uF MKS will likely be cheaper.
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 15157
  • Country: de
Re: Non microphonic, 1uF SMT cap in a high Z circuit?
« Reply #15 on: March 24, 2020, 08:56:20 pm »
100 nF C0G is not so bad anymore. Mouser has quite a few of them, starting at some 20 cents (Qt. 10).
 When limited to SMT it becomes more tricky with MKS and PPS caps may be more expensive.
 

Offline floobydust

  • Super Contributor
  • ***
  • Posts: 7681
  • Country: ca
Re: Non microphonic, 1uF SMT cap in a high Z circuit?
« Reply #16 on: March 24, 2020, 09:03:17 pm »
Why speculate when you can just listen to the capacitor microphonics. I use the circuit here:
https://www.eevblog.com/forum/projects/gravity-wave-detector-radio-electronics-april-1986/msg1103545/#msg1103545
I've had surprising results, some parts are very microphonic at high (vibration) frequencies, like a drum compared to a low thump.
 
The following users thanked this post: Someone

Offline Yansi

  • Super Contributor
  • ***
  • Posts: 3930
  • Country: 00
  • STM32, STM8, AVR, 8051
Re: Non microphonic, 1uF SMT cap in a high Z circuit?
« Reply #17 on: March 24, 2020, 09:13:26 pm »
100 nF C0G is not so bad anymore. Mouser has quite a few of them, starting at some 20 cents (Qt. 10).
 When limited to SMT it becomes more tricky with MKS and PPS caps may be more expensive.

Interesting. I have just looked and stay being corrected. Couple years back I have designed a space constrained lowpass filters for an 8 channel audio codec, where I have used C0G caps. The largest value required I think 22nF and cost a pretty penny.
 

Online dmillsTopic starter

  • Super Contributor
  • ***
  • Posts: 2093
  • Country: gb
Re: Non microphonic, 1uF SMT cap in a high Z circuit?
« Reply #18 on: March 25, 2020, 12:32:08 am »
Yea C0G has finally gotten useful for audio in places other then opamp stability caps.

 

Offline Zero999

  • Super Contributor
  • ***
  • Posts: 20363
  • Country: gb
  • 0999
Re: Non microphonic, 1uF SMT cap in a high Z circuit?
« Reply #19 on: March 25, 2020, 12:08:13 pm »
It is a fet input chopper 'zero drift' part.

Most of these loops are actually just stripping offset and bias currents off other bipolar opamps to avoid switch clicks and pot wiper noises.
It is a little annoying as for myself I would probably just throw 1,000uF 6.3V electrolytics in place of all the servos and call it good, but 'no electrolytics' is a marketing point (Audio, what can you do?).

Regards, Dan.
What? You're not allowed to use electrolytic capacitors, even in the power supply?  :palm:

What's the:
Budget?
Volume?
Size constraints?
Surface mount of through hole?
Tolerance?

Digi-Key sell 1µF 16V capacitor in 1206 size, but they're only 20% tolerance and are not cheap, even in quantities of 2k.
https://www.digikey.com/product-detail/en/cornell-dubilier-electronics-cde/FCA1206A105M-H3/338-4076-2-ND/4431988
 

Online dmillsTopic starter

  • Super Contributor
  • ***
  • Posts: 2093
  • Country: gb
Re: Non microphonic, 1uF SMT cap in a high Z circuit?
« Reply #20 on: March 25, 2020, 01:54:17 pm »
Elcos in the power supply are acceptable, but not in the "Signal Path" whatever the fuck that is supposed to mean (I mean if a component does not impact the signal, why is it there...).

As I say, tweaker audio, what can you do?
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf