lingDAC - cost effective RBCD multibit DAC design

Okay, thanks. I will leave it then. Regarding the sound my son described, it was something I think few people would probably consciously notice. It is just something my son was trained to listen for, and most likely nothing to be concerned about. And anyway, a tiny bit phasey or not, I still really like being able to hear those reverb tail details. We need to get that ambiance back working in the S-D dacs we mostly use today.
 
Last edited:
Every design is trade off in a sense, doesn't it ?
So, I guess for the price, Abraxalito made a fantastic alternative to the regular ESS based DACs from China.


But, the film caps tip suggested by Mark is very interesting. May be people like him suggest more improvements that can make this humble DAC sound outstanding even ?
 
I already have a suggestion which improves the SQ of the PhiDAC even further - but this one is a little bit tricky to implement. I've created two very low ESR capacitor 'hats' to go on top of the AD8017s. They're so bulky that two cannot be fitted on the same PCB so I've gone dual-mono, meaning I need two PhiDACs, one for left and the other for right. But the effort is well worthwhile :D
 

Attachments

  • 1821239817.jpg
    1821239817.jpg
    141.4 KB · Views: 382
I use ferrite because its easily available (just down the road there's an electronics market) and because its cheap (about 1-2RMB for cores like those). But if the other materials are easy to get hold of and cheap enough I'm open to trying them. Where to source DIY quantities of nano and permalloy cores? I've looked on Taobao and have yet to find...
 
On taobao you also can find 600:600 trafo etc for $10-20. I can find the contact of that winding supplier if you wish. Probably 1KU they will build, maybe half. You know, it is quite inexpensive, about 10RMB I believe, so need to order at least for $500-1000. If my memory serves well our(i.e. factory where I worked) order was 10KU.
 
Last edited:
I already have some of those Taobao permalloy trafos for about 98RMB, they're great but have higher inter-winding capacitance than I prefer (hundreds of pF, mine are more like 40pF) because they're inter-wound to minimize leakage inductance. Right now I'm not interested in $500 orders - I'd want to experiment first to see if different cores bring something worthwhile to the table.
 
I already have a suggestion which improves the SQ of the PhiDAC even further - but this one is a little bit tricky to implement. I've created two very low ESR capacitor 'hats' to go on top of the AD8017s. They're so bulky that two cannot be fitted on the same PCB so I've gone dual-mono, meaning I need two PhiDACs, one for left and the other for right. But the effort is well worthwhile :D
Wow ! Thats a lot of capacitors :p
Can you tell us its purpose ?


Why not a big capacitor instead of many ?
 
Disabled Account
Joined 2019
Thank You M. Abraxalito

@ Markw4 : Great review, very interesting reading. I have a newbie question about your film cap decoupling which is smoothing HF. Did you also try a simple good 100 uF lytic (so whith higher ESR than the // film cap).


@ Abraxalito : impressive design :). I noticed too much X7R caps whatever the quality of the design is not giving good high mid/treble subjective sound quality. There is always from my "poor" experience and taste too much light. For the buffer I prefer smd PPS caps or alike plastics despite a worse inductance due to a bigger package & when the capacitance amount is not big (small choice of capacitance values). Always valuable from my poor experience to use it alone in the buffer stage or // with X7R smd caps when they are used at local decoupling & energy storage for the aops. What are the main lytic caps reference please ? Any subjective difference with the classics ones as Panasonic FR or FM, Nichicons, ... Did you try for the curiosity of it ? Find the Panasonic FR to be one of the most invisible sound signature in audio designs : "fire" & forget caps ! Well I appologize as it's certainly naive inputs and perhaps off topic.

LingDac would be an overkill with added uf-l inputs, even better source could be choosed avaliable in this forum :) for perhap even better results ?

Thanks for sharing a so nice project at super wise cost: a great humility lesson.
 
@ Markw4 : Did you also try a simple good 100 uF lytic (so with higher ESR than the // film cap).

I did not do that with this dac, but with another one I tried that without much luck. It could very well be that there are other ways to get good results besides using film caps, such as Abraxalito has just reported with his stacked electrolytics. It was just that I have had pretty reliable results with film caps so far and I had some on hand.
 
Last edited:
Member
Joined 2010
Paid Member
I already have a suggestion which improves the SQ of the PhiDAC even further - but this one is a little bit tricky to implement. I've created two very low ESR capacitor 'hats' to go on top of the AD8017s. They're so bulky that two cannot be fitted on the same PCB so I've gone dual-mono, meaning I need two PhiDACs, one for left and the other for right. But the effort is well worthwhile :D


Hi abraxalito, looks great – ESR is in the range around 0.5 miliohm? Only organic polymer caps have better (lower) ESR.

Any current production (aluminium electrolytic) caps with such low values as the Nichicon HZ?
 
Disabled Account
Joined 2019
Thank you Markw4, I have no technical background and my question was naive, my too much simple though was a too low ESR at the inputt power supply rails of the dac board may have cause "oscillation" . But I really don't know, maybe lytic caps smooth highs by lack of resolution (higher ESR than film caps) but just on the flawed design which is not the case with ABraxalito's for sure. I just saw for my taste than too low ESR gives often too fast and light sound, but again having no technical background it is surely a bad inputt.
 
Last edited:
... too low ESR at the inputt power supply rails of the dac board may have cause "oscillation" .

It could happen depending on the particular circuit or opamps. For instance, AD797 might not like it. But, it is known to be sensitive to that so not surprising if there is some oscillation problem. Most other opamps should not have a problem.
 
Last edited:
Thanks for the explanation Chermann.

Chris gave the short answer, but there's a bit more to it than just ESR. ESL (equivalent series inductance) is also a concern when impedances are low, such as in power supply applications. A single capacitor of 18,000uF/16V will have a pin spacing much wider and hence more ESL than a 10mm diameter cap such as these 1800uF/16V. The ESL divides down by the number of caps too, just like the ESR. So it turns out, if you really want a low impedance supply, the only way to go is multiple paralleled caps.

I've used raw double-sided PCB material to keep the ESR and ESL low and the caps are soldered with very short leads to the two power planes.

@enregistree - my LCR meter showed an impedance about 1mohm of X and 1mohm of R @10kHz but down at such low levels its tricky to get a reliable reading as getting a good contact isn't easy. I don't know which current production 'lytics achieve the same ESRs - Nichicon HZ seem to come and go in production, various Taobao sellers appear to have them at the moment.
 
Chris gave the short answer, but there's a bit more to it than just ESR. ESL (equivalent series inductance) is also a concern when impedances are low, such as in power supply applications. A single capacitor of 18,000uF/16V will have a pin spacing much wider and hence more ESL than a 10mm diameter cap such as these 1800uF/16V. The ESL divides down by the number of caps too, just like the ESR. So it turns out, if you really want a low impedance supply, the only way to go is multiple paralleled caps.

I've used raw double-sided PCB material to keep the ESR and ESL low and the caps are soldered with very short leads to the two power planes.
.

:up:
 
@chribzzz

Here are the revised active filter values to make the droop suit a 96kHz sample rate :

C2,12 change from 820pF to 680pF
C14, C26 change from NF to 2n2
C24,C27 change from 22n to NF
C25,C28 change from 22n to 4n7

(NF means no fit)

Thanks!
What kind of caps do they need to be? x7r? Need to buy some if i want to try this mod.

Also is it an option since i might use this with my dsp to do the droop freq correction in my dsp with PEQ?
 
I can put them in a small bag along with your 2 kits if you like, they're all values I have to hand. They do need to be NP0.


Your last question I'm not sure I've grasped - if you want to do droop correction in DSP you'll lose a bit of digital headroom but you'll then be wanting a flat filter - the one I've just worked out has a small boost at 20kHz.