Cap recommendation for DAC Vref, Vcom pin

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I'm using a Vout DAC chip, PCM1793 & about to get delivery of another Vout DAC, the new Sabre ES9022. I've noted that these Vout DAC have a Vcom, Vreg pin (named differently in the different datasheets) which seems to decouple the bias/Vref.

I have heard that this pin is sensitive to the decoupling cap used here & I noted some of this variability myself on the PCM1793 DAC. A thread on Diyhifi suggests that firstly the datasheet value for this cap is very undervalued at 1uF & suggests a 100uF, low esr type - an oscon or panny FC/FM. Then Jocko rows in with high leakage in these caps being very bad for noise issues. Then somebody else says that film types will have too high an inductance.

My own brief experiment with limited cap supplies lean me towards film type. It seems to be like signal passing cap; film type needed for high frequency clarity but higher capacitance needed for better bass response. I only had low value film polypropylene to try.

Before I go off wasting money on expensive high capacitance film caps - has anyone got any experience in this area they could share? I was recommended to look at Audyn Plus caps & reading about them they are built to minimise inductance, so a good start.
 
I'd be very interested in an answer to this also... Is it possible that this is a place where using both in parallel might be a good idea? I mean, a larger electrolytic (say 47 - 100 uF) and a higher-quality low-value film cap across it as a snubber...

Cheers

Nigel
 
Jocko is right - leakage current usually does matter here, because impedance is fairly high and therefore small currents make for (relatively) large voltage noise.

1LSB on 16bit at 2Vout rms is only 32uV remember...

I like using film caps, 5mm pin spacing is mosty OK here IME. A better answer is usually to build an active buffer to drive the Vref pin with the expected voltage, because this can be a very low impedance source over a wider bandwidth when done right. Think 'filtered low noise reference plus suitable buffer'; detail left to your imagination.
 
Well, both low leakage and low ESR are important for Vref. From my experience, low-ESR has a slight edge over low-leakage. There 2 kinds of OSCONs. The later conductive polymer type is not recommended. Look for classic TCNQ type. SC-series is a good candidate for Vref filtering. Perhaps the best is the PMLCAP from Rubycon. However, it's quite expensive right now.

Poting
 
Last edited:
It very much depends on what's inside the DAC chip. If the capacitor is just there to filter off some noise from the common mode voltage, almost anything that doesn't leak excessively should work fine.

It's quite different when it is a decoupling capacitor for the DAC reference voltage and the DAC is some switched-capacitor circuit that draws nasty current spikes from it and needs to settle to extreme accuracy in each clock cycle. The parasitic inductance in the loop from the reference decoupling pin to the capacitor to the return ground then has to be kept as small as possible, and the mutual inductance of this loop with any loop that could induce interfering voltages in it also has to be kept as small as possible. That means using a capacitor with a small physical size placed as close as possible to the DAC chip. Mind you, a sigma-delta DAC is sensitive to interference at its reference at all frequencies except exact multiples of its clock frequency.

In this case I agree with weissi, except that I would recommend X5R instead of X7R. X5R capacitors usually provide more capacitance at a given case size. The disadvantage is that they may stop working above 85 degrees C while X7R works up to 125 degrees C, but that is unlikely to make any difference in domestic audio equipment. The rated working voltage has to be several times greater than the actual voltage because of the non-linearity of small-size X5R and X7R capacitors; it is not unusual for the capacitance to drop to 20 % of the nominal value at the rated voltage.
 
Last edited:
Thanks for all the input guys , all new to me and I would like to learn and not just copy ones idea.


So im leaning towards keeping the 220µF cap along with a bypass cap. Suggested was a 0.01µF cap like https://be.farnell.com/wima/fkp2d021001i00hssd/cap-0-01-f-100v-3-pp/dp/1890179 this Wima film, any reason not to go 0.1µF?
Also wondered about position, put the film right on the Vref pin or bypass right on the 220µF cap? Thanks again.
 
Film right on the Vref pin then, the film capacitor is meant for high-frequency bypassing so the loop from the Vref pin to the film capacitor and back via the ground pin has to be kept as small as possible.

By the way, due to the size and the lead wires it will never have as low an inductance as an SMD, but for a through-hole capacitor, this is indeed a good choice. The internal construction drawing in the datasheet shows that many capacitor plates are straight in parallel, reducing the inductance.

About going for a higher value: there is nothing wrong with that as long as the lead spacing doesn't increase and as long as the capacitor has an internal construction that ensures low inductance.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.