diyAudio (
-   Parts (
-   -   Sound of caps: 22uF vs. 10 uF (

sangram 17th September 2007 09:50 AM

Sound of caps: 22uF vs. 10 uF

As part of my modding projects, I've been recapping my soundcards and I've already finished with an SB Live! and a Delta 66.

In order to not be a wise-***, I maintained the original cap values, and ordered them a hundred at a time from Digikey.

Now it seems to me that the 22uF FC is slightly inferior to the 10uF FC. I cannot understand why a cap with lower ESR can sound worse, or am I looking at the wrong end of the horse?

The SB Live! has a UDA 1334, a very old Philips Bitstream DAC chip, and the Delta has an AKM 4524 'Codec'. The output opamps are LM4562 for both the cards, and output levels are similar for both configurations. The Live! originally powered the output off a +5V line, yours truly did some track hacking and now it runs off a RC filtered 12V supply. The other difference is 10 uF FCs on the Live!, and 22uF FCs on the Delta.

The difference in sound quality is not subtle. The Delta lacks the pace and speed of the Creative, sounding sluggish and bloated specially in the midrange. I must add that before the mods, the Live didn't sound anywhere near this good and the Delta beat it hands down.

The Delta pays excellent attention to layout and supply bypassing, far superior to the Creative card or the E-mu 1212M, (which is my next target). It runs the analog section off a +/- 12V supply, off what looks like a very decent RC filtered supply.

The other thing I did was to totally recap the Delta. All except 5 caps are changed out for FCs.

Or will it improve with time? I'm not much of a believer in burn-in for anything except solid-wood acoustic guitars, and I'm not experiencing any improvement that I can distinguish. The Creative has about 100 hours on it and the Delta maybe 40, but the Delta gets used a lot more now.

My next card is the 1212M, and that uses 47uF coupling caps. Should I switch them down to lower cap value (after checking for LF rolloff, I don't think that's going to be a problem as they couple to a buffer stage and that goes naked to the next stage) or stay with 47uF? The reason I ask is that the boards are pretty much one shot only, the second time I try to work on the board I'll damage it for sure.

TIA for any clues.

AndrewT 17th September 2007 10:22 AM

for audio coupling (=DC blocking) we usually aspire to using foil or film caps.
The only reason to use electrolytics is to widen the low bass response cheaply.

Try adding a film or foil bypass cap across the pins of the electros that are coupling audio signals. If this helps, try removing the electro and use as big a film cap that will fit the space available. You should be aiming for 1 to 2Hz high pass filter action with the coupling cap. You need to know the input impedance of the circuit to calculate the necessary cap value.

jameshillj 17th September 2007 10:42 AM

...... and if you replaced the 10uF/15v (?) FC
with a 22uF/50v FC, it may never replicate the original sound, even "breaking-in" forever.
On the other hand, with a bit of patience and with trying different bipass caps (thanks Andrew), you may be very pleasantly surprised.
And, as has been mentioned before many times, some experimenting with the "power snubbers" on your altered power supply can also give rather good results with this type of Cct/device, but Bipass first.

... james

AndrewT 17th September 2007 10:47 AM


Originally posted by jameshillj
...... On the other hand, with a bit of patience and with trying different bipass may be very pleasantly surprised. ....... but Bipass first.
what circuit and what values and what didn't work?
The "didn't work" list is probably the most important.

sangram 17th September 2007 11:26 AM

Thanks guys, I had used bypass caps on the earlier electros in the same circuit position. Maybe it's time to try those as film caps will just not fit, period, even the 5mm spacing of a .1 film cap is too big for the pads. The card is very tightly packed and the PT holes are miniscule, and with very small spacing - even the FCs had to be wiggled in. I calculated a C value of around 2uF for this exact application, but wanted to leave the original values in (M-Audio used a 22uF cap in this position - and a really cheap one) place so if I sell the card again, the next owner gets better than original performance (which it is, I was hoping for much better).

Thanks for comfirming that break-in won't help things. I guess it's time for my .47 MKP 16s to be pressed into active duty - except they're 15mm spacing, and the pads are 3mm apart. Let's see.

The reason I stayed with the values is that the 22uF is designed with an output impedance of 600ohms in mind, I guess. For the 1212 I have no idea what the impedance of the next stage is, as it is on the card and no easy way to measure or approximate - I can't even read any values on the little SMD resistors :(

And james, thanks for the tip on the PS, we will try that if there is space (the entire card is smaller than my palm.

But back to the Q, will a small cap be 'better' for audio than a 'bigger' cap, all other things being equal?

AndrewT 17th September 2007 11:43 AM

MKS (=PES) is a much smaller package than MKP.
100nF is only 2.4mm thick and if glued flat against the back of the PCB takes up very little space. 1uF is about 4.93mm thick. These are both 100Vdc, it may be possible to source smaller packages if you look for 50V or 25Vdc.

The legs/pins can easily be bent in towards the 3mm pad pitch to match the pins of the electros.
Fitting them instead of the electros might be very difficult.
I have no idea how badly xr7 ceramic would perform as a bypass, but SMD ceramics are tiny.
Are MKS available in SMD?

sangram 17th September 2007 11:59 AM

Aren't MKS polyester, and aren't polyester generally considered poor for audio? My understanding was that all smaller caps are non-poloyprop, and across the entire digikey line the non-ceramic SMD caps are all PPS, not polyprop/ester. Which I'm not sure about.

I'm not keen on trying a ceramic for signal path applications anytime soon though.

I think the MKP will have to be given a lead-bending time, and even then I'm not so sure about 4 caps in a width of about 15 mm (for all 4 caps:xeye: ).

I swear when I finish one card I need two days to recover from partial blindness. I just did a P2P with two LM chips yesterday and my eyes are still recovering, using 1/8 w resistors and working in 3 dimensions!

Conrad Hoffman 17th September 2007 01:04 PM

The one cap I would not use, unless it were an NPO/C0G, is a ceramic. The NPO caps are very limited in value, so not of much use. Modding SMT stuff is tough because you're so limited on the parts you can use, and making any physical changes is risky if there's nearby digital circuitry. Going from 10 to 20 uF with the same type of cap, and hearing a difference, almost certainly results from simple amplitude or phase differences. IMO, a -3dB point down around 3 hz is desirable, and I think you can hear differences if it's higher than that. What do you calculate for the rolloff difference between the two caps?

sangram 18th September 2007 06:31 AM

Just to clarify again - the original value was 22uF, and I swapped it with exactly the same value.

The comparison is between two different cards, so oranges and apples in a sense. And it was not in low end, but on the top end (which affected my perception of the low end too), where the (much inferior) Creative Live! was beating a card that cost 10 times as much.

For both cards the impedance they look 'into' is something like 20K, so a value of even 4.7uF is enough for the rollof of 3Hz, but I'm worried if I sell the card the next owner should be able to operate it within spec - this is studio equipment after all.

Having said all that I bypassed the card yesterday with some film caps extracted from the output stage of my NAD CDP. I am sure they are polyester and not polyprop because of their very small size. And my complaints are almost gone, the sound has cleaned up and is much improved, and I can now tell the difference between a Q6 Vorbis file and the source wave file, which leads me to believe that it is much better than the Live! again.

For the next card I will replace the caps with the same value as they look into an output stage whose impedance I cannot determine. So I will use the original 47uF caps and add .47 uF polypropelene film caps. The trick will be to solder the caps into the circuit, as they are 15mm lead pitch, and the are for 4 film caps is really small. Or I may use my axial stock - those caps are larger, but with some tricks I can squeeze them into unlikely places.

1212M, here comes the soldering iron! Lay down your arms!

AndrewT 18th September 2007 07:12 AM

the jump in sound quality when changing from electro to PES is significant.
From PES to PP is arguably "much" smaller.
If you can't fit 4 * PP @ 0.4inch pitch into the space, then stay with 4* PES 1uF @ 0.2inch pitch.
You will get much of the benefit.
An alternative would be 2.2uF PES, but these are thicker than 4.9mm, but still available in 0.2inch pitch.

All times are GMT. The time now is 07:27 AM.

Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 18.75%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Copyright ©1999-2018 diyAudio