I've never considered X8R, but in recent searches for a ~2µF X7R bypass for 17V rails, I've found that different chip sizes can have radically different 'capacitance loss' at bias. Some 50V chips have 80% of their nominal capacitance when biased to 17V, and some have only 20-30% of their nominal capacitance at 17V.
Many vendors have online simulators that can give you a pretty close estimate of how the part will perform at a particular bias, so I suggest using those. I also suggest trying again with X7R, unless you really need the 200ºC rating that X8R can do - there are X7R parts that can be used at relatively high voltages without losing most of their capacitance.
Many vendors have online simulators that can give you a pretty close estimate of how the part will perform at a particular bias, so I suggest using those. I also suggest trying again with X7R, unless you really need the 200ºC rating that X8R can do - there are X7R parts that can be used at relatively high voltages without losing most of their capacitance.
TDK has capacitance vs bias data on some of its X8R caps - for example here : https://product.tdk.com/en/search/capacitor/ceramic/mlcc/info?part_no=C3216X8R2A104K115AA
Certainly they do quite a bit better than X7R in this respect - on the ones I looked at, about 50% of the 'ticket' capacitance at full bias.
Certainly they do quite a bit better than X7R in this respect - on the ones I looked at, about 50% of the 'ticket' capacitance at full bias.
That seems to be excessively bad !
losing 20% of capacitance when voltage is @ 34% of rating is poor.
losing 70 to 80% @ 34% is incredibly bad.
Are these really X7R?
For general decoupling especially of digital, the size of the capacitor is critical.... and the resonant frequency, capacitance value is a lesser concern, choose the smallest available package.
Automotive, since the Toyota incident has become very safety conscious and here the flexi terminal higher temp X8Rs are excellent, also other high reliability situations where the elevated temp range is a benefit. Whether you will see a difference in decupling performance is another matter as 0603s are rather large these days for a lot of digital circuit decoupling with 0402s and 0201s being a better choice with todays fast rise and fall times...
Automotive, since the Toyota incident has become very safety conscious and here the flexi terminal higher temp X8Rs are excellent, also other high reliability situations where the elevated temp range is a benefit. Whether you will see a difference in decupling performance is another matter as 0603s are rather large these days for a lot of digital circuit decoupling with 0402s and 0201s being a better choice with todays fast rise and fall times...
"Typical applications include decoupling, bypass and filtering in extreme environments such as down-hole oil exploration, under-hood
automotive, military and aerospace."
I'm thinking we're not the target market for these things! The Kemet datasheet says the dielectric is CaZrO3.
automotive, military and aerospace."
I'm thinking we're not the target market for these things! The Kemet datasheet says the dielectric is CaZrO3.
The COG capacitor long predates any audiophile concern for dielectric quality and wasn't the last time I checked designed specifically for audiophiles which is what my comment addressed. I never stated that it was not good for audio applications, just that it was not designed for audiophiles...
It is clearly a much better choice in the unfortunate scenario where one must use an smd ceramic coupling capacitor for an audio application.
And I was really just trying to be funny, so apparently you missed my intention.
Modern Calcium Zirconate C0G capacitors are actually extremely high quality for audio applications. The dielectric exhibits no nonlinearities. It has no microphonic or piezoelectric effects, and the breakdown voltage for these parts is usually extremely high compared to their data sheet rating. They perform better than the older, more expensive C0G formulations that used precious metal electrodes - Calcium Zirconate is simply a better dielectric than the older varieties of perovskite ceramics used for NP0/C0G in the past.
Since the SMD end terminations are very reliable, MLCCs are even better than many film caps, where problems with the end terminations can create measurable distortion for some fraction of a given lot of otherwise high quality parts.
So, don't dis' ceramic capacitors. Specific types are actually quite good, and not all NP0/C0G caps are the same either. Again, the old, more expensive ones are actually not as linear as the modern, Calcium Zirconate and base metal electrode varieties. The remaining drawbacks are still mechanical size, low capacitance values for coupling uses, and cost, but not performance.
I went back and checked my notes, and here's an example of an X7R that loses lots of capacitance with bias: the Taiyo Yuden UMK212BB7225KG-T Here's a link to the product page, from which you can download the PDF data sheet. UMK212BB7225KG-T CERAMIC CAPACITORS TAIYO YUDEN Co.,Ltd.The DC bias graph shows the massive C loss with bias, and yes, it is an X7R!
Maybe there are different ceramics used for X7R, but whatever the cause, not all X7R behave the same under bias. It's good to check the data sheet, or run a simulator if the vendor has one, and specify the exact bias you'll use it with.
Yes packagesize vs capacitance voltage coëfficiënt, higher capacitance classII in smaller package drop more capacitance with voltage. Kemet uses package even bigger than cog for xr8 and maybe someone can tell about cog above temperature rating, because I remember seeing very linear graphs also above 125 degrees for 125 degrees part, maybe just a convention difference how classI is specced and how classII is allowed to be specced, for 150 degrees the cog might be more stable too, I don't know.
Ceramic Capacitors | Capacitors | DigiKey
Ceramic Capacitors | Capacitors | DigiKey
X7R MLCCs in small SMD packages have their uses, the main one being digital decoupling where they are one of the best choices, especially for the smaller values and case size allowing the decoupling cap to be as near to the pin as possible... Critical in reducing parasitic inductance.
I have no problem using a 10uF/50V/X7R/1210 MLCC as an input coupling cap for a class "D" amp such a TI TPA3116. You would never have to change it. Pretty hard to hear or measure the difference when i used a film type in the same app. Do not think I'd use a MLCC in a good class "A-B" type amp however.
I have to look at what sizes are available in X8R and compare prices to see if is an alternative to what i an using already.
I have to look at what sizes are available in X8R and compare prices to see if is an alternative to what i an using already.
Last edited:
But these X8Rs only really make sense if you need the high temperature rating. For HiFi there are other more suitable/cheaper choices
A diversion:
I have bought some of these and removed the output cap and muting.
IT002 Digital to Analog Audio Converter DAC Converter (EU Plug) - Free Shipping - DealExtreme (or eqal)
Original they are quiet, but with the 0604 output cap they have a lot of distortion at max out. (guess it is class 3 cap?)
As you see the distortion is nicely spread. Maybe it sounds i little warmer?
'At -15dB the distortion is at -80dB which is OK.
(Modded I now remove the DC at input of a multichannel amp. The input is a MiniDSP nanodigi)
(The SPDIF receiver and DAC are from firms that has been in the game for a while.😉)
I have bought some of these and removed the output cap and muting.
IT002 Digital to Analog Audio Converter DAC Converter (EU Plug) - Free Shipping - DealExtreme (or eqal)
Original they are quiet, but with the 0604 output cap they have a lot of distortion at max out. (guess it is class 3 cap?)
As you see the distortion is nicely spread. Maybe it sounds i little warmer?
'At -15dB the distortion is at -80dB which is OK.
(Modded I now remove the DC at input of a multichannel amp. The input is a MiniDSP nanodigi)
(The SPDIF receiver and DAC are from firms that has been in the game for a while.😉)
Attachments
Last edited:
- Status
- Not open for further replies.