Solid polymer capacitors for decoupling and bypassing

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Not always, sometimes a little ESR in the decoupling caps can help damp resonances when multiple value caps are used. These will be reservoir caps in the decoupling scheme to large for local pin decoupling, there MLCC in the smallest package size is best especially for digital circuitry (X7Rs are a good choice).
 
I read mixed reports about OSCONs for this duty although in theory they should be outstanding because of their ultra low ESR.
Anybody have any experience they are willing to share?


Sent from my iPhone using Tapatalk
Oscons are good for digital 'dec' duty...and they don't have 'ultra low esr'.

Use Panasonic fc,fm series or Vishay Roe for after regulator dutys.
 
These kind of general statements are false in this field. Good or not? This is not simple as this. Every applications are different. MLCC is good for the highest freq, Oscon and other polimer elcaps are good for about 10 kHz to 1 MHz at low voltage, and traditional electrolytics are good for about DC to 200 kHz.

Ultra Low ESR is not an absolute term, it means much lower than one of a normal electrolytics of similar size, but it is still higher then one of a MLCC! They don't resonate with typical parasitic inductance, MLCC do.
 
Last edited:
Not always, sometimes a little ESR in the decoupling caps can help damp resonances when multiple value caps are used. These will be reservoir caps in the decoupling scheme to large for local pin decoupling, there MLCC in the smallest package size is best especially for digital circuitry (X7Rs are a good choice).



Do they make good noise filters and reservoirs not right at the power supply pin but close by for audio frequencies ( less than 100khz but particularly lower frequencies)? Obviously the very high frequencies need to be dealt with by the smallest physical size possible ceramic right at the pin.


Sent from my iPhone using Tapatalk
 
I was experimenting with Oscons in digital circuits - input and output C in voltage regulators and around digital ICs in DACs. They work very well in these positions and help render a very fluid sound - this is probably because of their efficiency in high frequencies. However, I found the treble to still be a bit grainy (I also used them for local filtering of analogue voltages in DAC IC) and I found tantalums to work better for me. Have no idea how Oscons work in analogue circuits but I wouldn't use them for signal decoupling.
 
I thought nothing better than trying for myself so I put a 180uf Panasonic SEPC on each of the positive and negative rails of the output filter/buffer op amp circuit for my AK4490 dual mono dac. Subjectively this was equivalent to a couple of dB improvement in the SNR. The improvement in SQ was more marked than every time I have played around with different capacitor makes in power supplies. In this particular application they seem to be miles better than Panasonic FMs etc. All subjective off course as the only test instruments I have are a multimeter and my ears.


Sent from my iPad using Tapatalk
 
I'm using low impedance, low ESR, low leakage lytic cap which I replaced the X7R's.
Improvements: More musical, liquidity smooth, very natural, not dry clinical as was before, improved bass extension, tonal balance, excellent decay. This DAC is audiophile grade now and beats up on $2-4K DACS ive heard. Also, OSCONS in this position are not as good either as they have a higher leakage rate.
 

Attachments

  • sabre.jpg
    sabre.jpg
    665.3 KB · Views: 698
You mean electrically?

They're leaky in the electrical sense yes. Which isn't normally much of an issue since they're used across power supplies which are low impedance by design. But occasionally you'll see them specified in higher impedance circuits (for example on the ADJ pin of a 3pin regulator) where they only bring a downside vs a conventional 'lytic.
 
I'm using low impedance, low ESR, low leakage lytic cap which I replaced the X7R's.
Improvements: More musical, liquidity smooth, very natural, not dry clinical as was before, improved bass extension, tonal balance, excellent decay. This DAC is audiophile grade now and beats up on $2-4K DACS ive heard. Also, OSCONS in this position are not as good either as they have a higher leakage rate.

I cannot say this is at all recommended.
 
I cannot say this is at all recommended.
Apparently your recommendation is based on 1970s college text books? . Because not only are the improvements there. Many other diy's on other dac threads have done similar, in removing the ceramics on the op amp power rails to be replaced with low esr caps and heard nice improvements. My own experience with a class AB amp, I also deleted the X7RS on the op amp power rails, with low esr lytics, and the Sq was amazing especially in the lf. Abraxilito I'm sure can explain the scientific theory behind it if you ask him.
 
Last edited:
No, I'm reading plenty of datasheets, from Murata, AVX, etc.; I understand quite well what's going on. Have a quick read of what Marce wrote earlier in this thread.

Polymer electrolytic capacitors are great in specific locations, getting a good x7r ceramic capacitor local to your low-power, high-bandwidth chips is good engineering practice. Your higher value electrolytics can be a little further away, but you want that low inductance as local to the chip as reasonably possible.

What are impedance/ ESR frequency characteristics in capacitors? | Murata Manufacturing Co., Ltd.

http://www.murata.com/~/media/webrenewal/support/library/catalog/products/emc/emifil/c39e.ashx
(Another good one from Murata, relevant to DACs)

Likewise, I simply don't trust your subjective evaluations. It's not personal, just follows from an understanding of how easy we are to inadvertently bias ourselves. (And hope you don't trust mine either, as I don't!)

I'm glad you enjoyed your changes, I just cannot recommend that as a smart idea.
 
Last edited:
They're leaky in the electrical sense yes. Which isn't normally much of an issue since they're used across power supplies which are low impedance by design. But occasionally you'll see them specified in higher impedance circuits (for example on the ADJ pin of a 3pin regulator) where they only bring a downside vs a conventional 'lytic.

I've recently seen a power supply with polymer caps across the voltage reference. One of the caps leaked a few mA, which made the output voltage wrong.

erik777:

You report better sound with the X7R ceramics removed. One possible explanation is that the MLCCs were paralleled using traces (ie, inductors) instead of the proper way (ie, GND/power planes). ESR of MLCC is too low. Using traces to connect several MLCCs creates a nice big resonance. A solution is to add a small resistor (like 0R1 in 0603 package) in series with your MLCCs. It increases the inductance a bit, but kills the resonances.

Note that power delivery inductance must be compared to the chip power pin inductance. There is no need to go much lower, unless the chip has lots of power/GND pairs. For an opamp, the package and lead frame already contribute several nH... much more than a MLCC. If several chips sit on the same rail, power delivery network impedance serves to prevent interactions, though.

As a final note, the inductance of a 6mm diameter aluminium cap with 2.5mm lead spacing is the same as a through-hole ceramic cap or film cap of similar physical dimensions. Which means "decoupling"' this elcap with a through-hole 100nF ceramic or film is useless (sometimes, worse than useless).
 
Last edited:
Member
Joined 2013
Paid Member
I put a pair of 330uF 25V rated Oscons (Panasonic SVPF) into a power supply circuit where they would be exposed to no more than 22V, and once I powered it up, they started a catastrophic breakdown / leakage mode that could not be stopped without removing the power supply.

Essentially, these things need to be de-rated for voltage, sort of like a Tantalum cap, but nobody says "how much is safe". What's more annoying is that these parts aren't available in higher voltages than what I used - I really need a part that can handle 22-24V, and these seem to not be able to do that.

The promise of solid Al caps was that they could eliminate the wear-out effect of traditional wet Al electrolytics, but if they're gonna self-destruct and burn the room along with them, I need to avoid the whole thing.

Now, I'm happily using Panasonic FR caps instead, and find little to fault.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.