And I thank you for that. I will listen. As much as I possibly can. I promise.
Please let me clarify something though: I do not take position on the rather sterile debate of objective measuring vs. subjective listening. I deeply believe that both are absolutely necessary, for a lot of very good reasons that we do not need to cover on this thread (other people can do a much better job at this than myself). In fact, on most reasonably contested issues, I usually do not take any position, for I believe that different viewpoints can bring complementary enlightenments on a particular topic, and one would be wise to carefully listen to as many contradictory opinions as possible before passing any judgement.
Therefore my prudent approach to subjective listening should not be interpreted as some kind of quantitative bias in disguise. Instead, it is nothing more than a humble acknowledgement of my limited skills. But I am here to learn — about this, there can be no doubt...
Last edited:
Regarding 470uf caps near dac chip: I would suggest allowing for eight of them, two for each channel to start with. Depending on how it sounds, you might want to add another 1,000uf cap for each channel. Doesn't necessarily have to be Oscon though, Panasonic FM type might be okay (for example).
No worries. I know your heart is in the right place. And online communication has such limited bandwidth...
That I do not understand. Based on the chart of page 28, two 470µF capacitors will give you close to 1,000µF. At 20Hz, that will give you 6dB more THD+N than 2,200µF. Therefore, why take a chance and not go for two 1,200µF capacitors? The size difference is not that significant, and I'm pretty sure they would fit. The chart is rather unambiguous: the bigger the capacitance, the less distortion at lower frequency.
In that case, the EEE-FT0J222GP could be an option: 10mm diameter, 2,200µF. Going OS-CON does not give us this option and would require the use of two separate capacitors, taking more space on the board. But the voltage rating is limited to 6.3V, while the DAC chip is limited to 6V, so that's cutting it pretty close. Is that an issue?
Another option is the Nichicon PCG0J272MCL1GS, which offers 2,700µF within the same form factor.
And if we find a way to squeeze four capacitors with a 12.5mm diameter, we can go all the way up to 3,500µF with the Panasonic EEE-FK0J352SQ. But the height would go from 10.2mm to 13.5mm, so we'd have to figure out what the implications might be.
Bottomline: it looks like we have many options that we can play with.
Last edited:
You can if you want. Two 1,200uf Oscons might be enough space allowance for whatever you eventually decide.
Its just that there obviously has to be some reason AKM didn't go for the lowest distortion cap values. I wanted to know why they chose the values they did, so I listened, grabbed some test jumpers, added some caps, listened, shorted out the resistors using some test jumpers, listened...
You get the idea. That's why I suggested you do it, so you could find out for yourself. No special ear training required, its easy to hear how it changes. That's how you learn, not by me telling the results.
Okay, I get it. I'll do some experiments as soon as my evaluation board is up and running.
OSCONs and other true solid polymer Al electrolytic caps have higher leakage current than normal and hybrid types. Not sure it makes a difference here, honestly, but I would use a hybrid cap like Panasonic ZK. They top out around 470 or 560 uF I think, though.
I would test it myself, but the safe bet is the largest amount of capacitance. I wouldn't think too hard on this, from what I've seen the eval board is weird in some places and you never know who actually designed it and if they are any good, or if they forgot to update the BOM, or it was completed before the graphs were added, etc.
I might also test other VREF supply circuits anyway, it is unclear to me the reason a very high performance and quiet supply should require a huge cap or an RC filter at ~5 Hz with 3300uF cap.
I would test it myself, but the safe bet is the largest amount of capacitance. I wouldn't think too hard on this, from what I've seen the eval board is weird in some places and you never know who actually designed it and if they are any good, or if they forgot to update the BOM, or it was completed before the graphs were added, etc.
I might also test other VREF supply circuits anyway, it is unclear to me the reason a very high performance and quiet supply should require a huge cap or an RC filter at ~5 Hz with 3300uF cap.
Last edited:
Thanks a lot for this Chris.
I will try to get in touch with the designer(s) next time I'm in Japan (September).
Might also try some of the exact same type caps AKM used. Possibly instructive depending on the outcome.
Thing is, wondering if they really want something close to an ideal supply, or something that somehow makes the dac sound 'right' to them. It will take some more work to find out.
Thing is, wondering if they really want something close to an ideal supply, or something that somehow makes the dac sound 'right' to them. It will take some more work to find out.
Last edited:
Indeed, because looking at the datasheet for the EKMQ500EMC471MJ20S, I really can't find anything remarkable about it. If anything, it's a relatively low-cost domestically-produced SKU.
I would really like to talk to the designers of this board...
I know you already wrote about it, but could you remind me what you would replace the 7805 regulators with?
470µF.
Can you develop why you picked the ZK series? What is it about them that could make them particularly suitable? I know it's a guess on your part, but I'm trying to better understand your line of thinking on that one.
One may recall that Jung regulators can be made unstable by using modern electrolytic caps that are too 'good.' The diyaudio store sell Jung-Didden regulator boards and there is a support thread. Electrolytic caps affecting stability are talked about there, IIRC.
In this case I doubt putting good caps between the 10-ohm resistors will cause stability problems, but not too surprising to me if they affect sound by some means we have not yet considered. (Not that I have not been thinking about a possible plausible physical explanation
)
In this case I doubt putting good caps between the 10-ohm resistors will cause stability problems, but not too surprising to me if they affect sound by some means we have not yet considered. (Not that I have not been thinking about a possible plausible physical explanation
)The AD site seems to be having a fit right now, but I'd probably use ADM7150 or TPS7A47 or similar. Probably overkill, though. The fact that AKM used an input voltage of 15V limits your choices somewhat. I'm also not sure how much current they use on those supplies, so I picked >= 800 mA capable parts.
I picked it just because it had good overall specs. The hybrid caps combine a lot of the advantages of the OSCONs with higher voltages, lower leakage, etc.
https://www.mouser.com/pdfdocs/Panasonic_Capacitors_WP_final.PDF
I picked it just because it had good overall specs. The hybrid caps combine a lot of the advantages of the OSCONs with higher voltages, lower leakage, etc.
https://www.mouser.com/pdfdocs/Panasonic_Capacitors_WP_final.PDF
Last edited:
One may recall that Jung regulators can be made unstable by using modern electrolytic caps that are too 'good.' The diyaudio store sell Jung-Didden regulator boards and there is a support thread. Electrolytic caps affecting stability are talked about there, IIRC.
In this case I doubt putting good caps between the 10-ohm resistors will cause stability problems, but not too surprising to me if they affect sound by some means we have not yet considered. (Not that I have not been thinking about a possible plausible physical explanation
I will check the site out. Thanks!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.