best type of big capacitor to use for a low pass filter?

[SArepairman]

If I need a fairly big capacitor to use in a low pass filter in a precision circuit what kind would you recommend?
Lets say like 100uF. Ceramic? aluminum polymer? tantalum?

[Niklas]

Metallized popypropylene

[wraper]

Ceramic? aluminum polymer? tantalum?

Definitely none of the listed.

[EEVblog]

You're probably going to have to define what you mean by "precision".
Ideally that means a zero tempco, zero voltage dependency, and zero dielectric absorption etc.
A low pass filter is not generally going to be as critical as say a reference capacitor used in say a charge amplifier or ADC.
Generally you are going to use one of the film types like Polypropylene.

[SArepairman]

So where do I get a film capacitor that is 100uF but not rated for 600V and not the size of a golf ball?

http://www.digikey.com/product-search/en?pv13=67&FV=fff40002%2Cfff80010&mnonly=0&newproducts=0&ColumnSort=0&page=1&stock=1&quantity=0&ptm=0&fid=0&pageSize=25

These seem very inappropriate for my op-amp.

these capacitors seem to be intended for running heavy duty things.

Capacitance multiplier inside of a salen key low pass filter?

typically when I see a big capacitor used in a app note by a major company they seem to specify a bi-polar type capacitor when the size is +100uF, even in op-amp circuits.

even if you dial down to 47uF for a film capacitor:
http://www.digikey.com/product-search/en?pv1292=43&pv1292=47&pv1292=57&pv1293=37&pv1293=42&FV=fff40002%2Cfff80010%2C3400ba%2C34028b&mnonly=0&newproducts=0&ColumnSort=0&page=1&stock=1&quantity=0&ptm=0&fid=0&pageSize=25

10$ and 1 inch by 0.5 inches. I could live with that I suppose if its the best option. and thats polyester...

[c4757p]

You're going to have a fun time trying to find a good 100uF film cap for this purpose. Ten 10uF in parallel? Four 22uF (= 88uF) and change something else to make it work? Change the rest of the circuit by a factor of ten and use 10uF?

typically when I see a big capacitor used in a app note by a major company they seem to specify a bi-polar type capacitor when the size is +100uF, even in op-amp circuits.

Yes, for instance I've seen a Jim Williams note on measuring noise that used some magical, selected (IIRC) wet tantalum. You might want to sit down for the price.

[wraper]

@sarepairman. you did post a similar question in the beginners forum. maybe my answer there helps. gererally, you cannot circumvent physics. a nanoamp input current is generated by a semiconductor, it also generates noise. actually, due to the pretty basic question you are asking,  i do not believe that your requirements are so extreme that they cannot be met. you have not said what noise level you are seing and what you want to achieve.
generally, without being more specific of what you want to achieve, I guess everything has been said.

I think those questions are about the same schematic. Do you really need capacitor that big? Seems that you are trying to solve a problem in some obscure way.

[SArepairman]

You're going to have a fun time trying to find a good 100uF film cap for this purpose. Ten 10uF in parallel? Four 22uF (= 88uF) and change something else to make it work? Change the rest of the circuit by a factor of ten and use 10uF?

typically when I see a big capacitor used in a app note by a major company they seem to specify a bi-polar type capacitor when the size is +100uF, even in op-amp circuits.

Yes, for instance I've seen a Jim Williams note on measuring noise that used some magical, selected (IIRC) wet tantalum. You might want to sit down for the price.

well I have seen things like 47uF tantalum used in op-amp circuits around precision op amps, I know the nanovolt level noise thing you are speaking about, with the tantalum slug.

even paralleling them up, a 10uF metal foil is fairly expensive. I cannot spend 40$ on 10 parallel capacitors in my circuit, and it is still a huuge amount of area.

[SArepairman]

@sarepairman. you did post a similar question in the beginners forum. maybe my answer there helps. gererally, you cannot circumvent physics. a nanoamp input current is generated by a semiconductor, it also generates noise. actually, due to the pretty basic question you are asking,  i do not believe that your requirements are so extreme that they cannot be met. you have not said what noise level you are seing and what you want to achieve.
generally, without being more specific of what you want to achieve, I guess everything has been said.

I think those questions are about the same schematic. Do you really need capacitor that big? Seems that you are trying to solve a problem in some obscure way.

nope completely different application

[c4757p]

What are you actually trying to do? Can you digitize the signal? It'd be much easier than fairy-dust-and-unicorn-fart capacitors.

[SArepairman]

What are you actually trying to do? Can you digitize the signal? It'd be much easier than fairy-dust-and-unicorn-fart capacitors.

not interested in digitizing anything. I want to filter the ever living crap out of a DAC that adjusts the offset voltage in a circuit after range change is done.

i guess I want to know what is the least evil of the polarized capacitors. I only need one so I can splurge on some kind of weird polymer thing, if its any better.

What about capacitance multipliers for a small film cap to turn it into a big film cap? I would rather do a RC filter -> buffer if I had a really large C, I think.

[wraper]

What are you actually trying to do? Can you digitize the signal? It'd be much easier than fairy-dust-and-unicorn-fart capacitors.

not interested in digitizing anything. I want to filter the ever living crap out of a DAC that adjusts the offset voltage in a circuit after range change is done.

i guess I want to know what is the least evil of the polarized capacitors. I only need one so I can splurge on some kind of weird polymer thing, if its any better.

What about capacitance multipliers for a small film cap to turn it into a big film cap? I would rather do a RC filter -> buffer if I had a really large C, I think.

wow, why do you need 100uf mentioning precision circuit for that.

[SArepairman]

What are you actually trying to do? Can you digitize the signal? It'd be much easier than fairy-dust-and-unicorn-fart capacitors.

not interested in digitizing anything. I want to filter the ever living crap out of a DAC that adjusts the offset voltage in a circuit after range change is done.

i guess I want to know what is the least evil of the polarized capacitors. I only need one so I can splurge on some kind of weird polymer thing, if its any better.

What about capacitance multipliers for a small film cap to turn it into a big film cap? I would rather do a RC filter -> buffer if I had a really large C, I think.

wow, why do you need 100uf mentioning precision circuit for that.

because I recall seeing some app note some where that used a rather large C for a very similar purpose (nulling)

I wanna say they used a 47uF for a low pass filter,.

[wraper]

Aren't you going to make a single pole RC low pass filter with a huuuge cap?

[SArepairman]

Aren't you going to make a single pole RC low pass filter with a huuuge cap?

yea, a RC filter followed by a buffer. If the capacitor is big enough that should be all it ever needs.


oooo I just realized the voltage can be positive or negative. So I need a non polar capacitor.

so that either means electrolytic BP, film, ceramic or two unipolar capacitors...

[codeboy2k]

Go for a larger resistor and smaller capacitor and buffer the output of the rc filter before you send it anywhere else. This is the proper way.

[c4757p]

Why does the DAC output need to be filtered so much? It's just a fixed offset, right? Go with a DC DAC like an R-2R rather than a delta-sigma or something, and you shouldn't need so much filtering.

[T3sl4co1l]

Your error due to dielectric absorption, even on the best possible capacitors, will completely and utterly swamp your attempt at measurement.  Find a different way.

Tim

[SArepairman]

Your error due to dielectric absorption, even on the best possible capacitors, will completely and utterly swamp your attempt at measurement.  Find a different way.

Tim

hmm? please elaborate.


I intend to adjust the output of the DAC while measuring the offset that I want to null (after the RC and buffer, which is connected to a summer). what problem will dielectric absorption cause?
The offset is only intended to change once per range switch, I don't care if it takes a few seconds to auto calibrate itself.

[T3sl4co1l]

Dielectric absorption looks like an R+C across your ideal capacitor, with the R relatively large (giving a long time constant, minutes to weeks) and C some small fraction of the bulk capacitance (usually 1% or less).

Consider what happens if the capacitor sits on a shelf.  Over time, the two capacitances equalize in voltage.  Then you pick it up and apply a voltage: you get two time constants, one dominant, another extremely slow and weak.  The gain difference (between, say, 1Hz and 0.001Hz -- at least, one frequency well above and one well below the absorption's cutoff frequency) between these is very small, but since you're measuring the difference between signal and a reference, this is an error in your measurement.  And even from the best capacitors, 0.01% is a whole lot of LSBs at 20 bits or more.

Tim

[SArepairman]

Dielectric absorption looks like an R+C across your ideal capacitor, with the R relatively large (giving a long time constant, minutes to weeks) and C some small fraction of the bulk capacitance (usually 1% or less).

Consider what happens if the capacitor sits on a shelf.  Over time, the two capacitances equalize in voltage.  Then you pick it up and apply a voltage: you get two time constants, one dominant, another extremely slow and weak.  The gain difference (between, say, 1Hz and 0.001Hz -- at least, one frequency well above and one well below the absorption's cutoff frequency) between these is very small, but since you're measuring the difference between signal and a reference, this is an error in your measurement.  And even from the best capacitors, 0.01% is a whole lot of LSBs at 20 bits or more.

Tim

all I could think then is to use a smaller capacitor in a gyrator? no other solution...?

can you use a capacitance multiplier connected to a foil capacitor maybe?

how about this?


In this circuit, the capacitance of capacitor C1 is multiplied by the ratio of resistances, i.e. the circuit gain and the effective capacitance C = C1 * R1 / R2.

When using this circuit, care must be taken not to exceed the current capability for the op-amp. This should not be an issue with low current applications, but where higher currents are required, this must be taken into account.

[EEVblog]

not interested in digitizing anything. I want to filter the ever living crap out of a DAC that adjusts the offset voltage in a circuit after range change is done.

You generally don't need a precision cap for something like that. Because even if you get a horrible that that varies by +/-50%, you design your system so it doesn't matter.
Are you able to chose a different DAC? I assume oyu are talking about a PWM based DAC here?

[SArepairman]

not interested in digitizing anything. I want to filter the ever living crap out of a DAC that adjusts the offset voltage in a circuit after range change is done.

You generally don't need a precision cap for something like that. Because even if you get a horrible that that varies by +/-50%, you design your system so it doesn't matter.
Are you able to chose a different DAC? I assume oyu are talking about a PWM based DAC here?

it is a presumably dirty DAC connected to a micro controller. It is not PWM. I thought the same as you did until I read the post

Dielectric absorption looks like an R+C across your ideal capacitor, with the R relatively large (giving a long time constant, minutes to weeks) and C some small fraction of the bulk capacitance (usually 1% or less).

Consider what happens if the capacitor sits on a shelf.  Over time, the two capacitances equalize in voltage.  Then you pick it up and apply a voltage: you get two time constants, one dominant, another extremely slow and weak.  The gain difference (between, say, 1Hz and 0.001Hz -- at least, one frequency well above and one well below the absorption's cutoff frequency) between these is very small, but since you're measuring the difference between signal and a reference, this is an error in your measurement.  And even from the best capacitors, 0.01% is a whole lot of LSBs at 20 bits or more.

Tim

I initially thought that a capacitor capacitance changing would just effect the Fo of the low pass filter, meaning the vRMS noise would change only, not the DC level, however I think that Tim is saying that the DC level will drift like an RC circuit because of the dielectric absorption, or at least that dielectric absorption is the main detail to worry about while designing at this level..
I suspected there would be problems like this, because there always are with precision , which is why I asked.

***From my personal experiences with op amp RC filters I found that they were pretty damn stable. I am fairly sure a regular old one as I have used in the past would be used acceptably in this application however I have never attempted to build anything this precise (or have been this interested in stability) and I would like to know the fine details behind the operation, so I can write nicer specifications on a piece of paper...


*tim, are you essentially saying that the dielectric absorption leads to a particularly nasty source of low frequency noise that would often be lumped in with 1/f noise? I am a bit unclear.

[EEVblog]

it is a presumably dirty DAC connected to a micro controller. It is not PWM.

Without knowing exactly what type of DAC you are using, it's pointless looking into all this.

[TerminalJack505]

Dielectric absorption looks like an R+C across your ideal capacitor, with the R relatively large (giving a long time constant, minutes to weeks) and C some small fraction of the bulk capacitance (usually 1% or less).

Consider what happens if the capacitor sits on a shelf.  Over time, the two capacitances equalize in voltage.  Then you pick it up and apply a voltage: you get two time constants, one dominant, another extremely slow and weak.  The gain difference (between, say, 1Hz and 0.001Hz -- at least, one frequency well above and one well below the absorption's cutoff frequency) between these is very small, but since you're measuring the difference between signal and a reference, this is an error in your measurement.  And even from the best capacitors, 0.01% is a whole lot of LSBs at 20 bits or more.

Tim

all I could think then is to use a smaller capacitor in a gyrator? no other solution...?

can you use a capacitance multiplier connected to a foil capacitor maybe?

how about this?


In this circuit, the capacitance of capacitor C1 is multiplied by the ratio of resistances, i.e. the circuit gain and the effective capacitance C = C1 * R1 / R2.

When using this circuit, care must be taken not to exceed the current capability for the op-amp. This should not be an issue with low current applications, but where higher currents are required, this must be taken into account.

By introducing an op amp into the mix you are essentially creating an active filter.  You might investigate that line of thought and see if any of the active filter topologies meet your requirements.