When using low values (0.1uF, 0.22uF, 0.47uf) which type should I use?
I know I should use Monolithic for bypassing (opamp rails, power stiffener bypass) but what about other applications? Can I safely replace poor ceramic disc caps with Polyester / Monolithic, and which one is better?
And what about output caps? I'm talking about 0.1uF, 0.47uF and 0.22uF that can be found here: http://www.e-ele.net/DataSheet/TA2020.pdf
P. 5 Right Side of Diagram (goes into the speakers).
Thanks!
I know I should use Monolithic for bypassing (opamp rails, power stiffener bypass) but what about other applications? Can I safely replace poor ceramic disc caps with Polyester / Monolithic, and which one is better?
And what about output caps? I'm talking about 0.1uF, 0.47uF and 0.22uF that can be found here: http://www.e-ele.net/DataSheet/TA2020.pdf
P. 5 Right Side of Diagram (goes into the speakers).
Thanks!
ceramic change capacitance with changing voltage.
i.e. vary the signal voltage and the capacitance varies.
They are very good if the voltage is held steady. eg decoupling the supply pins of an opamp.
All ceramics suffer from this.
In addition tolerance and capacitance stability changes with with temperature. NP0/C0G are the tightest tolerance and the most stable, the Xs and the Ys are very dependant on temperature.
MKT/PES/Mylar generally have a higher ESR than MKP/PP and this makes the MKT slightly more resistant to oscillation when there are fast signals and inductance around.
i.e. vary the signal voltage and the capacitance varies.
They are very good if the voltage is held steady. eg decoupling the supply pins of an opamp.
All ceramics suffer from this.
In addition tolerance and capacitance stability changes with with temperature. NP0/C0G are the tightest tolerance and the most stable, the Xs and the Ys are very dependant on temperature.
MKT/PES/Mylar generally have a higher ESR than MKP/PP and this makes the MKT slightly more resistant to oscillation when there are fast signals and inductance around.
for filtering Class D you want the caps to have both low parasitic R and L, cap construction details such as electrode termination and lead length are critical, dielectric type is less important
surface mount caps are the best on the low L, provided you've done the PCB correctly with gnd plane and kept filter components close together
surface mount film types need to be safe at soldering high temps because the whole body heats up so PEN and PPS are the most commonly used plastic film dielectric types
NP0/C0G ceramic caps are not V sensitive like the higher "k" types- in fact their performance in high impedance filtering situations is better than some plastic films and mica caps
Class D filtering is not a high impedance application and "audiophile" dielectric "quality" is largley irrelevant - just avoid high k ceramics which do have large Vcoefficients
surface mount caps are the best on the low L, provided you've done the PCB correctly with gnd plane and kept filter components close together
surface mount film types need to be safe at soldering high temps because the whole body heats up so PEN and PPS are the most commonly used plastic film dielectric types
NP0/C0G ceramic caps are not V sensitive like the higher "k" types- in fact their performance in high impedance filtering situations is better than some plastic films and mica caps
Class D filtering is not a high impedance application and "audiophile" dielectric "quality" is largley irrelevant - just avoid high k ceramics which do have large Vcoefficients
Last edited:
In my daytime world we use X7R's for general de-coupling usualy 0402 to 0805 depending on value (0402) preffered due to inductance.
For critical/RF NPO/COG are used but cases tend to be larger due to the construction. Despite rumors NPO/COG is a stable dielectric, and even X7R's arn't going to cause you problems, they are used to de-couple circuits that are a lot more sensitive and critical than audio.
There has been another thread that has covered this topic with a link to a good paper on MLCC capacitors, the different dialectrics etc, I'll look through my stuff and get a link.
The problem is that some people class ALL MLCCs as the same, this is not true, and they do have there place especially in de-coupling digital circuits where the small packages available and low impedence makes them the right choice for high speed de-coupling. If you use multilayer boards capacitor position and via placement become more critical as circuit speed increases.
For critical/RF NPO/COG are used but cases tend to be larger due to the construction. Despite rumors NPO/COG is a stable dielectric, and even X7R's arn't going to cause you problems, they are used to de-couple circuits that are a lot more sensitive and critical than audio.
There has been another thread that has covered this topic with a link to a good paper on MLCC capacitors, the different dialectrics etc, I'll look through my stuff and get a link.
The problem is that some people class ALL MLCCs as the same, this is not true, and they do have there place especially in de-coupling digital circuits where the small packages available and low impedence makes them the right choice for high speed de-coupling. If you use multilayer boards capacitor position and via placement become more critical as circuit speed increases.
I should thank you mister Andrew for this
seven years ago that you wrote this I had no idea this would be my problem someday
+1
For decoupling X7R is all good for audio and much more. It's the physical aspects of both the capacitor and pcb layers / layout that are important - and SMT essentially removes the inductance of the capacitor leads. And the correct value of course- given that a smaller value will tend to be effective upto a higher frequency.
If you think there's still a problem then you probably need to look at inductors / ferrite beads in line. But it's unlikely unless you are dealing with 'fast digital / rf' signals - and if they are 'in' your audio signals or supplies then they probably shouldn't be
The 'X2Y' type of capacitors are worth a look
For signal path caps - Ceramic NPO/COG ; Polystyrene and Polypropylene (PP) are all good.
This (and similar questions) is often asked. There is no answer (apart from "it depends") because it seems to assume as true something which is false: that some caps are "better" than others.LosNir said:
All caps are good at something, otherwise they would not be made. For supply rail decoupling at high frequencies you often want small size (so low inductance) and a little HF loss to dampen resonances: so use high-k ceramics (their nonlinearity does not matter in this application). At lower frequencies an electrolytic may be the best bet. For signal frequency filters (including some, but not all, coupling capacitors) you need stability and linearity, so use film caps (or foil if you are fussy).
So the question must always be "better at what?".
Indeed. Although to be fair some types are basically just good at packing a high capacitance into a smaller package for lower cost.
But we often read than the ESR of caps, in serie as in shunt towards gnd are calculated in speaker filter for optimal results?
Is it true ? For instance in some Kef speaker of the high end line they use some Low Loss bipolar caps (said to be 5% loss at ??? : which is high vs a film cap but better than some basic elec bi-polars) !
Akways read than the Mylar (MKT) film are close to the elec cap for behavior in a speaker filter (ESR, tangent angle loss), then only the polypropylen made a huge improvment (less distorsion) and of course the better polystyrene and teflon dielectric !
But the dielectric some brands seems to calculate perfectly the resistivity of their speajer filter and it matters in the XO behavior ! Dunno if a myth or not as we often read : don't change this elec cap in your speaker driver for a polypro or it will waste the time constant, etc, etc, asf !
Is it true ? For instance in some Kef speaker of the high end line they use some Low Loss bipolar caps (said to be 5% loss at ??? : which is high vs a film cap but better than some basic elec bi-polars) !
Akways read than the Mylar (MKT) film are close to the elec cap for behavior in a speaker filter (ESR, tangent angle loss), then only the polypropylen made a huge improvment (less distorsion) and of course the better polystyrene and teflon dielectric !
But the dielectric some brands seems to calculate perfectly the resistivity of their speajer filter and it matters in the XO behavior ! Dunno if a myth or not as we often read : don't change this elec cap in your speaker driver for a polypro or it will waste the time constant, etc, etc, asf !
Naively changing caps which have been carefully chosen by the original designer can lead to degraded performance. As I often say, in order to genuinely improve a circuit you first have to understand it (in detail) even better than the original designer. This fact seems to be ignored by most component swappers.
Yes, but we can also assume on basic products (and sometimes on the very good ones) the designer had not the time budget to improve the choice of the caps.
Some mix gives better results than others and as you said it may also worst the whole result... hence the questions we always see ! One should indeed understand few things before playing with caps... but some very famous designers here does, with good results... but they firtstly made a good design & layout...
With some dacs, amps, pre, here that were improved by some caps and better supply as well : despite all the adverts, auto self proclamed best stuff of the year, decade, century, it's only showing it's a poor design : the engineers are not all good : experience matters as well, as the good teaching and informations exchanges ! hard to sort out for the enthusiasts or swappers as you design them !
Some mix gives better results than others and as you said it may also worst the whole result... hence the questions we always see ! One should indeed understand few things before playing with caps... but some very famous designers here does, with good results... but they firtstly made a good design & layout...
With some dacs, amps, pre, here that were improved by some caps and better supply as well : despite all the adverts, auto self proclamed best stuff of the year, decade, century, it's only showing it's a poor design : the engineers are not all good : experience matters as well, as the good teaching and informations exchanges ! hard to sort out for the enthusiasts or swappers as you design them !
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.