I added 12 caps (220/16 Panasonic FR) to VREF. Big difference. Mandatory IMO. This cured a lot of the harshness in the high/midrange and lack of power in bass. The board should come with these caps inserted at the factory.
Is there a concensus WRT to diminishing returns, in other words how much capacitance is enough with the build in regulators?
Is there a concensus WRT to diminishing returns, in other words how much capacitance is enough with the build in regulators?
Last edited:
VREF supplies
I too found that increasing the capacitance to VREF did indeed significantly improve the SQ as you describe. I found that powering each the + and - VREF lines directly with batteries was better than that, and with LPS 1.2s was better still.[/SIZE][/SIZE]
I too found that increasing the capacitance to VREF did indeed significantly improve the SQ as you describe. I found that powering each the + and - VREF lines directly with batteries was better than that, and with LPS 1.2s was better still.[/SIZE][/SIZE]
I too noticed a significant SQ improvement in removing the output filtering caps from the single ended output board. However, my old guy eyes are having trouble finding the locations of the output caps on the 1541 balanced output board.
Attached are photos of the 1541 board. If someone will identify the locations of the output filtering caps, I'll remove them and report my findings.
Attachments
Soekris DAC: Modding VRef | H i F i D U I N O
This probably contains all the technical discussion there is on this topic.
Which rev are you using? I feel that I'm not getting everything out of my rev4 with no mods implemented, but Soren said many times that caps mods are no longer necessary. He did make paralleling electrolytic caps extremely easy with the vias though in the new versions though.
See link above. Maybe 3300uF per rail (4 in total)? Soren added an additional 300uF per rail in rev.4 and rev.5 so 3000 more?... But to be fair the lower noise may not be audible.
Sorry the last post was incorrect. 3300 recommendation is for rev1 without resistor mod only. But the measurements do show a 7-10db decrease in noise(?, the measurements did not specify the level at which the white noise is played) in some frequencies going from factory mod + 1000uf to factory mod + 6800uf. Note rev4 should be factory mod + 300uf if I understood it correctly.
If ripple rejection is all there is to the cap mods, I wonder if a good enough power supply or batteries would Reduce the importance of the mods.
If ripple rejection is all there is to the cap mods, I wonder if a good enough power supply or batteries would Reduce the importance of the mods.
Last edited:
Thanks. I currently use a pretty good switchmode PSU +/-15V that feeds a linear PSU +/- 7V using standard regulators with lot's of capacitance. Before the power lines go into the DAM they travel through chokes to get rid of any RF noise. There is zero hum detectable above the noisefloor on the DACs output.
So the power supplied is pretty good, nevertheless, the added VREF capacitance made a huge and very audible difference. I have recorded before / after samples that show this. It's very obvious in the difference signal as well, a lot more than anything I usually find with doing changes in analog gear.
Interestingly enough, it doesn't seem to matter much for measured harmonic distortion.
Because the capacitance is added at a spot where it would make a substantial difference. The same increase upstream is not likely to have a comparable effect.
It's an NPO ceramic cap, so not really crappy (and this really depends on application anyway).
It seems most people who removed the cap are not using the balanced outs. I have to use the buffers, so they need to work properly.
No so easy when you have two boards stacked with extremely tight pin/pin headers and basically wired together
You have me really tempted to do the mod.... if the difference is measurable but doesn’t show up in THD+n numbers, what do you think changed...? I’m going to hunt for some good caps now, is FPCAP best? Thanks a lot!
Styroflex of course
//
That sounds fancy. How about regular polymer caps? I can get 6.3v 1500uf FPCAP(nichicon) or chemicon caps pretty easily and cheaply to plug directly on the board through the vias. Low ESR at 8-10 mR I think. This would amount to 1500*3+100*3+47=4847uF per rail. Should be enough?...
Edit: Nichicon (NU) has better ESR at 7mR 10*12.5 footprint 2000hrs life. Chemicon (PSC) is 10mR 8/10*11.5 15000hrs (not typo). Maybe I should go with latter?
Last edited:
Hifiduino recommenced the FPCAP 10mR for low ESR. But the switches are 0.3R so no point in super low ESR caps. I chose Chemi-con PSC (APSC6R3ELL152MJB5S) mostly for its size (direct plug-in) capacitance (1500uf 6.3v) and durability (tested at 105C 15000hrs). It’s very cheap on Taobao (15 for less than $5 total). I would prefer mouser for peace of mind but don’t have access this month... the Taobao vendor seems legitimate and was rather surprised that I was so worried about fakes...
Last edited: