I'm thinking about pro and cons of different bypass caps. In audio equipment lots of people prefer film caps over ceramics - even for bypasses. But looking at the size and ESR ceramic caps seems to be the better option. Smaller size comes with lower inductance. The only argument pro film caps seems to be linearity - which seems not critical for me in this use case.
Any arguments for the use of film capacitors as bypass caps?
Any arguments for the use of film capacitors as bypass caps?
The esr of good film caps is likely to be low, maybe too low.
This would require a proper R+C snubber to be used for the decoupling.
It is far simpler and far cheaper and probably less inductance to use the internal esr of the X7R as the R of the snubber.
A higher esr ceramic is the correct choice for HF decoupling.
This would require a proper R+C snubber to be used for the decoupling.
It is far simpler and far cheaper and probably less inductance to use the internal esr of the X7R as the R of the snubber.
A higher esr ceramic is the correct choice for HF decoupling.
Surely for a bypass capacitor, the lower the ESR the better. Both X7R and film of sufficient capacity have low ESR at audio frequencies. X7R are probably better for higher frequency decoupling due to their smaller size and lower inductance that some op-amps require. I always use a bulk aluminium or tantalum capacitor and a smaller film or X7R close to the decoupling point to remove any high frequency noise or feedback. I have never seen a snubber used to increase the ESR of a bypass capacitor whose main aim in life is to provide a low impedance path from any power supply noise to ground. Yes a resistor upstream of the bypass capacitor will provide additional decoupling but one in series.....No No No.
Film caps are better in signal circuits as they do not exhibit a lot of the "funnies" of ceramic capacitors that can be a pit fall for the unwary. Temperature coefficient, voltage dependance and microphony are just a few of the issues with ceramics if the correct type of dielectric is not chosen carefully.
Film caps are better in signal circuits as they do not exhibit a lot of the "funnies" of ceramic capacitors that can be a pit fall for the unwary. Temperature coefficient, voltage dependance and microphony are just a few of the issues with ceramics if the correct type of dielectric is not chosen carefully.
cap + trace inductance equals ringing.
Cap+snubbing resistor + trace inductance equals no ringing, when you get it right.
An X7R has roughly the right amount of resistance to prevent that ringing.
If you adopt a low esr capacitor then adding the cap may introduce ringing that was not there before. The only way I know to prevent that ringing with a low esr cap is to add resistance to damp the ringing.
Cap+snubbing resistor + trace inductance equals no ringing, when you get it right.
An X7R has roughly the right amount of resistance to prevent that ringing.
If you adopt a low esr capacitor then adding the cap may introduce ringing that was not there before. The only way I know to prevent that ringing with a low esr cap is to add resistance to damp the ringing.
John Curl uses bulky and chunky MKP and MKS REL caps as PS decoupling caps near the gain blocks he designed. Even near PS pins in his IC based design Parasound JC-3 phono preamp.Old Threshold preamps had giants MKP 100n Wonder caps for PS decoupling. Erno Borbely prefers Wima 100n MKP-10 in the same position.
All the mentioned caps are bulkier than 100nf stacked foil polyester or ceramic multilayer X7R caps., and have larger inductance. We have only better dielectric in MKS and MKP or FKP capacitors for decoupling. How better dielectric quality influence PS decoupling? More HF noise filtering ? Less garbage to gain blocks? Any measuring proof? What about inductance? Or we just have"It simply sounds better" unreliable argument.
All the mentioned caps are bulkier than 100nf stacked foil polyester or ceramic multilayer X7R caps., and have larger inductance. We have only better dielectric in MKS and MKP or FKP capacitors for decoupling. How better dielectric quality influence PS decoupling? More HF noise filtering ? Less garbage to gain blocks? Any measuring proof? What about inductance? Or we just have"It simply sounds better" unreliable argument.
What are you bypassing?
Most electrolytics do not need bypassing unless you are working at radio frequencies.
If you are bypassing electrolytics with a high ESL then be warned that this can make a resonant circuit (not a good thing) which can colour the sound. At audio frequencies it's not necessary. For power supplies you only need it if there is a considerably long length of wiring in between the amp and the reservoir cap. Are you thinking of snubbing the rectifier?
Most electrolytics do not need bypassing unless you are working at radio frequencies.
If you are bypassing electrolytics with a high ESL then be warned that this can make a resonant circuit (not a good thing) which can colour the sound. At audio frequencies it's not necessary. For power supplies you only need it if there is a considerably long length of wiring in between the amp and the reservoir cap. Are you thinking of snubbing the rectifier?
I'm talking about bypass caps for the power supply of low power audio ICs like opamps, analog switches, instrumentation amplifiers and similar devices.
I'd go for ceramic in that case - it's not too critical so long as there's a capacitance. Very cheap and reliable. There's a lot of bad press for ceramics out there but they're perfect for this application. You usually don't need anymore than 10nF. I use 22nF ceramics and haven't had any problems whatsoever.
Hope that helps!
Hope that helps!
X7R ceramics make excellent bypass caps - just watch out for their voltage coefficient. Often not shown in the datasheet, some parts lose up to 80% of their capacitance when charged close to their full rated voltage. Kemet's Spice program is an excellent method of exploring the changes that occur under bias - whilst the capacitance reduces, so does the ESR.
The change in capacitance does not matter.
The voltage applied to the cap when the supply is up to operating voltage is supposed to stay constant. That means the capacitor value stays constant at that applied voltage.
Now draw a fast current spike. The capacitor supplies the current for a tiny instant of time. The voltage will deplete very slightly. The capacitors job is done.
The MF decoupling now tries to recharge the voltage lost in the HF decoupling and also supplies the slower rise time current demands.
Once you have selected you ceramic capacitor, finds the biggest value in that package size. That gives you the lowest inductance for the package and the maximum capacitance for for holding the voltage as nearly constant as possible.
The voltage applied to the cap when the supply is up to operating voltage is supposed to stay constant. That means the capacitor value stays constant at that applied voltage.
Now draw a fast current spike. The capacitor supplies the current for a tiny instant of time. The voltage will deplete very slightly. The capacitors job is done.
The MF decoupling now tries to recharge the voltage lost in the HF decoupling and also supplies the slower rise time current demands.
Once you have selected you ceramic capacitor, finds the biggest value in that package size. That gives you the lowest inductance for the package and the maximum capacitance for for holding the voltage as nearly constant as possible.
its counterintuitive I know, but Andrew is 100% correct. x7r is the most suitable choice for localised PS bypass, BECAUSE of the small amount of ESR. being slightly lossy allows the cap itself to damp any ringing/resonance. this is a reason why tantalums are good too. the most important component of the impedance in a HF bypass is ESL, so with that still at a minimum, the small amount of resistance is of no great bother. they are also able to pack more capacitance into the smallest packages, lowering inductance again. I like the 0508 size reverse aspect ratio caps, easily solderable by hand, but with the inductance of an 0402
if you use the lowest impedance type you can, like a c0g or film you can easily cause more problems than you solve as they can be very high q, causing them to be quite picky for the value chosen as well.
- Status
- Not open for further replies.