Speaking of hints and tips for the RPi dacs, odd there don't seem to be any good recommendations for powering the dacs other than from batteries (whatever happened to the super cap fad?).
Anyway, the dac board itself does need three separate 3.3v regulators to sound as good as it should. DVCC and VCCA could each be powered by LDO regulators (if, say, one wanted to compare linear power supply SQ to that of using batteries). However, ES9038Q2M dacs often don't sound good with LDOs used for AVCC. One could probably use an AD797 opamp more or less as ESS recommends and perhaps use a dedicated LDO to provide the opamp 3.3v input reference voltage. http://www.esstech.com/files/4514/4095/4306/Application_Note_Component_Selection_and_PCB_Layout.pdf
Only precaution about using AD797 (which can sound great used for AVCC power) is that they can get unstable if the power rail impedance is too low (maybe a good way to see if the effective source impedance of batteries is really all that low at the loads
) If stability issues, something like OPA1611 could be tried.
For the output stage, haven't tried to compare batteries myself, but a simple test to see if the effective power rail source impedance is low enough to make the opamps sound their best might be to put some large-ish value film caps from each of the +- output stage power rails to ground very close the power input terminals of the output stage board. If no audible change in sound quality, then the batteries are probably doing quite well. If any improvement in SQ at all from adding film caps, then PS source impedance as seen by the output stage may be a bit higher than optimal.
Be very interested to see if anyone tries any of the above besides me. I don't have a battery setup so unfortunately I can't compare. I tried LiPO4 for AVCC once on one dac and that was enough of that unstable AVCC sound for me. Probably no surprise either. Even if using an opamp supply for AVCC, the ESS recommendations I linked to above suggests putting the AVCC supply as close to the dac chip AVCC pins as physically possible, so as to keep wiring from having any adverse SQ effects (which it can).
Anyway, the dac board itself does need three separate 3.3v regulators to sound as good as it should. DVCC and VCCA could each be powered by LDO regulators (if, say, one wanted to compare linear power supply SQ to that of using batteries). However, ES9038Q2M dacs often don't sound good with LDOs used for AVCC. One could probably use an AD797 opamp more or less as ESS recommends and perhaps use a dedicated LDO to provide the opamp 3.3v input reference voltage. http://www.esstech.com/files/4514/4095/4306/Application_Note_Component_Selection_and_PCB_Layout.pdf
Only precaution about using AD797 (which can sound great used for AVCC power) is that they can get unstable if the power rail impedance is too low (maybe a good way to see if the effective source impedance of batteries is really all that low at the loads
) If stability issues, something like OPA1611 could be tried. For the output stage, haven't tried to compare batteries myself, but a simple test to see if the effective power rail source impedance is low enough to make the opamps sound their best might be to put some large-ish value film caps from each of the +- output stage power rails to ground very close the power input terminals of the output stage board. If no audible change in sound quality, then the batteries are probably doing quite well. If any improvement in SQ at all from adding film caps, then PS source impedance as seen by the output stage may be a bit higher than optimal.
Be very interested to see if anyone tries any of the above besides me. I don't have a battery setup so unfortunately I can't compare. I tried LiPO4 for AVCC once on one dac and that was enough of that unstable AVCC sound for me. Probably no surprise either. Even if using an opamp supply for AVCC, the ESS recommendations I linked to above suggests putting the AVCC supply as close to the dac chip AVCC pins as physically possible, so as to keep wiring from having any adverse SQ effects (which it can).
Last edited:
Taking it to the next "level" (continue)
Just a quick note on powering the 3 rails of the DAC HAT separately. This removes the last bit of digital noise from the sound and I highly recommend it.
Initially I removed the 3 decoupling caps mini-tabs and powered the 3 rails directly using batteries. I later on added the decoupling caps mini-boards back and prefer that as it sounds smoother and with better resolution.
If anyone is planning on doing the same mod I recommend keeping the decoupling cap boards in-place. Just cut the PCB bridge near the "in" label should separate the 3 rails. A simple continuity test using a multimeter will confirm that.
Just a quick note on powering the 3 rails of the DAC HAT separately. This removes the last bit of digital noise from the sound and I highly recommend it.
Initially I removed the 3 decoupling caps mini-tabs and powered the 3 rails directly using batteries. I later on added the decoupling caps mini-boards back and prefer that as it sounds smoother and with better resolution.
If anyone is planning on doing the same mod I recommend keeping the decoupling cap boards in-place. Just cut the PCB bridge near the "in" label should separate the 3 rails. A simple continuity test using a multimeter will confirm that.
PS mania
Jacklee, I have no doubt your separate PS adds magic.
I wonder if a dedicated PS to the clock on the FIFOPi would also add. It would be easy to remove the + & - pins and supply power separately from the main board. Need to check if the clocks are powered all the time or only when selected.
Jacklee, I have no doubt your separate PS adds magic.
I wonder if a dedicated PS to the clock on the FIFOPi would also add. It would be easy to remove the + & - pins and supply power separately from the main board. Need to check if the clocks are powered all the time or only when selected.
That's very likely will impact the sound. My understanding is that the clock not in use is switched off by the ENABLE pin. Whether the clocks themselves are powered all the time I'm not certain but I don't think that matters.
Please do share your experiences if you plan to do the clock mod.
I have 2 sets of CCHD-957 & NDK @ 45/49’s & both Ian’s old and new clock adapters in their stock form.
There have been many capacitor suggestions. I have used the Cornell Dublier caps in other projects to great results.
I need a little direction please. Folks have thrown around the Black Gate: BG NX HiQ .33u 50v caps, don’t have any.
The included ceramic caps are getting poor results, so I’m looking for definitive capacitor suggestions.
I do have the Nichicon 370-AUDIO-KIT (0.22 uF to 330 uF). Plenty good audio caps in there I can run.
Have an Airwork station, have built for practice a few SMD test boards to pretty good results (not so good with SMD LEDs).
With this open option, what would you do? Suggestions welcome.
Aguaazul
There have been many capacitor suggestions. I have used the Cornell Dublier caps in other projects to great results.
I need a little direction please. Folks have thrown around the Black Gate: BG NX HiQ .33u 50v caps, don’t have any.
The included ceramic caps are getting poor results, so I’m looking for definitive capacitor suggestions.
I do have the Nichicon 370-AUDIO-KIT (0.22 uF to 330 uF). Plenty good audio caps in there I can run.
Have an Airwork station, have built for practice a few SMD test boards to pretty good results (not so good with SMD LEDs).
With this open option, what would you do? Suggestions welcome.
Aguaazul
Just from reading online. Perhaps they are just fine. But I'd like to try something else.
Well, if a film cap added to one clock also affects the other clock, then it would seem likely that tacking a small film cap across one of the clock socket power pins on the back of the fifo board might work just as well. Probably somewhere between .1uf and 1uf. Smaller caps will be more effective at higher frequencies, but which cap(s) works best is a trial and error search problem. Might help if the voltage regulator was closer to the clocks and dedicated to the clocks only. Caps might only be able to help so much if something else is the main underlying issue. Just sayin'
May I ask what source material are you listening to? Also, have you tried turning down the software volume just a bit, maybe 3dB? Most of the volume control should be done using the dac digital volume control, but sometimes reducing the volume level slightly before the digital signals arrive at the dac can help some. Other things can help too, if you think sound quality needs to be a little better.
Last edited:
WOW!
I did it and WOW! Like upgrading from regular NKD SD to SDA. Not a subtle change in my system. All the usual sound improvement that comes from lower phase noise oscillator. Pin point 3d sound stage. Separation, rich natural sound, black background. Ian's FIFOPi has now exceeded my best result with BBB and Ako S03 which was very good.
So here is the set up using mains power.
The clean PS to FIFOPi is a small Hammond 229 transformer, super soft rectifiers, CLCLC filtering using a 13x1800uf caps producing 8v DC. A salas shunt takes it down to 5v. Then Ian's TPS7a4700 reg down to 4.1v. This feeds the clean side of FIFOPi and Ian's I2SPCM bd. So the FIFOPi is using the onboard LT3042 reg. There is a separate feed from the TPS7A4700 reg to a second 4700 based reg soldered directly to the 45mHz clock adapter. It takes the 4.1v down to 3.3v. I removed the power pin from the clock adapter so they share ground but not power.
Also for full disclosure, the clock adapter has the 3 ceramic caps removed from the bottom and has one .68uF FCA cap on the bottom and a single .46uF BG HiQ cap on the top power to ground.
Also both the 4700 regs have BG caps per some earlier threads.
I am not advocating that 4700 regs are better than LT3042. I just happen to have the 4700 regs available from my last build so its a 'run what you brung' kinda deal.
I am writing this after listening for about 10 mins. The regs were previously used for a long time but were idle for a few months so I don't expect a lot of run in improvements but again the sound improvement was not subtle. I'll report after a few days if there is any appreciable change.
I expect similar improvement is possible for battery users.
I did it and WOW! Like upgrading from regular NKD SD to SDA. Not a subtle change in my system. All the usual sound improvement that comes from lower phase noise oscillator. Pin point 3d sound stage. Separation, rich natural sound, black background. Ian's FIFOPi has now exceeded my best result with BBB and Ako S03 which was very good.
So here is the set up using mains power.
The clean PS to FIFOPi is a small Hammond 229 transformer, super soft rectifiers, CLCLC filtering using a 13x1800uf caps producing 8v DC. A salas shunt takes it down to 5v. Then Ian's TPS7a4700 reg down to 4.1v. This feeds the clean side of FIFOPi and Ian's I2SPCM bd. So the FIFOPi is using the onboard LT3042 reg. There is a separate feed from the TPS7A4700 reg to a second 4700 based reg soldered directly to the 45mHz clock adapter. It takes the 4.1v down to 3.3v. I removed the power pin from the clock adapter so they share ground but not power.
Also for full disclosure, the clock adapter has the 3 ceramic caps removed from the bottom and has one .68uF FCA cap on the bottom and a single .46uF BG HiQ cap on the top power to ground.
Also both the 4700 regs have BG caps per some earlier threads.
I am not advocating that 4700 regs are better than LT3042. I just happen to have the 4700 regs available from my last build so its a 'run what you brung' kinda deal.
I am writing this after listening for about 10 mins. The regs were previously used for a long time but were idle for a few months so I don't expect a lot of run in improvements but again the sound improvement was not subtle. I'll report after a few days if there is any appreciable change.
I expect similar improvement is possible for battery users.
Well, if a film cap added to one clock also affects the other clock, then it would seem likely that tacking a small film cap across one of the clock socket power pins on the back of the fifo board might work just as well. Probably somewhere between .1uf and 1uf. Smaller caps will be more effective at higher frequencies, but which cap(s) works best is a trial and error search problem. Might help if the voltage regulator was closer to the clocks and dedicated to the clocks only. Caps might only be able to help so much if something else is the main underlying issue. Just sayin'
May I ask what source material are you listening to? Also, have you tried turning down the software volume just a bit, maybe 3dB? Most of the volume control should be done using the dac digital volume control, but sometimes reducing the volume level slightly before the digital signals arrive at the dac can help some. Other things can help too, if you think sound quality needs to be a little better.
As I'm part of the few waiting for the battery power supply board, I'm still in the planning stages for Ian's fifopi parts. Currently I run 2 HiFiBerry's. DAC+ Pro & DAC+ Pro XLR running on MAX2Play images. Music is classic rock, The Tulsa Sound era recordings & live shows, jazz & a nas with live music from the jam band scene. We have over a 1000 live shows on CD / DVD, mostly in lossless format already. Just invested in a 25 CD / DVD Ripper to archive all of our disks. The smattering of wav file disks will be extracted using the ripper using EAC one at a time.
I started out with Ian's Dual XO II Clock Board, I2S FIFO II, I2S to PCM, & more for a Red Baron DAC project that is still in the works. So when this RPi came along, clocks & adapters already in stock make this an good project to slide in to.
Appreciate your input.
Aguaazul
I'd like to introject an experience I've had on ceramic caps here for those trying alternative caps.
First, I am not posting this to say you shouldn't try SMD film or small Black Gates (if you have them) or other caps for local bypassing of clocks. I do this all the time and am a big proponent of it.
BUT, unless you give the ceramic caps what I've experienced as being sufficient time to break in, you aren't comparing apples to apples.
My experience was from modifying Juli@ sound cards a number of years back. I realized the caps marked 'BC' were bypass caps to each of the power feeds of the activie chips and clocks on the board. I tried a mod where in addition to feeding it better-regulated power from a stacked pair of modified Dexa regulators (3.3V, 5V) with separate raw DC supplies I also added a 10uF cap across most of the 'BC' caps. Definition, dynamics, bass impact all increased along with treble detailing and delicacy... AFTER it all broke in. Before then, I could hear what I call a 'sandy' texture or fine grain to the sonics. That disappeared after 3 weeks. BUT it still had a bit of harshness even then.
After listening to it for awhile, I tried a different mod on the same board where I replaced those 10uF ceramic with 33uF Oscons. The sound with these in place was VERY sweet, especially in the trebles. BUT bass impact and overall dynamics were diminished. Break-in sonics for these was a slight harshness and wooliness to the bass the disappeared after about a week.
With some extended listening, I decided to go back to the 10uF ceramic caps. BUT I made some adjustments in the overall mod, using Panasonic PPS caps for the power feeds on the chips generating the I2S signal that I fed to the DAC card AND around the regulators along with one of those 33uF Oscons bypassed with a PPS film across the power rails to the error opamp on the Dexas. This overall was the best compromise, sweet in the trebles (though not quite as much as the Oscons alone) with almost all of the definition, dynamics, and bass impact of the ceramics alone, but no harshness. AND again, I heard that 'sandy' texture for another 3 weeks until those new ceramics broke in.
I came away from this experience firmly in the 'parts need break-in' camp.
I say all this because IMHO, if you haven't given those ceramics 3 weeks or so of constant running, you likely aren't hearing them at their best. ALSO in my experience, SMD films and small BG N and NX produce better sonics when used as clock bypasses. BUT the differences you are hearing might not be quite as large if you haven't given the ceramics full break-in time.
YMMV!
Greg in Mississippi
First, I am not posting this to say you shouldn't try SMD film or small Black Gates (if you have them) or other caps for local bypassing of clocks. I do this all the time and am a big proponent of it.
BUT, unless you give the ceramic caps what I've experienced as being sufficient time to break in, you aren't comparing apples to apples.
My experience was from modifying Juli@ sound cards a number of years back. I realized the caps marked 'BC' were bypass caps to each of the power feeds of the activie chips and clocks on the board. I tried a mod where in addition to feeding it better-regulated power from a stacked pair of modified Dexa regulators (3.3V, 5V) with separate raw DC supplies I also added a 10uF cap across most of the 'BC' caps. Definition, dynamics, bass impact all increased along with treble detailing and delicacy... AFTER it all broke in. Before then, I could hear what I call a 'sandy' texture or fine grain to the sonics. That disappeared after 3 weeks. BUT it still had a bit of harshness even then.
After listening to it for awhile, I tried a different mod on the same board where I replaced those 10uF ceramic with 33uF Oscons. The sound with these in place was VERY sweet, especially in the trebles. BUT bass impact and overall dynamics were diminished. Break-in sonics for these was a slight harshness and wooliness to the bass the disappeared after about a week.
With some extended listening, I decided to go back to the 10uF ceramic caps. BUT I made some adjustments in the overall mod, using Panasonic PPS caps for the power feeds on the chips generating the I2S signal that I fed to the DAC card AND around the regulators along with one of those 33uF Oscons bypassed with a PPS film across the power rails to the error opamp on the Dexas. This overall was the best compromise, sweet in the trebles (though not quite as much as the Oscons alone) with almost all of the definition, dynamics, and bass impact of the ceramics alone, but no harshness. AND again, I heard that 'sandy' texture for another 3 weeks until those new ceramics broke in.
I came away from this experience firmly in the 'parts need break-in' camp.
I say all this because IMHO, if you haven't given those ceramics 3 weeks or so of constant running, you likely aren't hearing them at their best. ALSO in my experience, SMD films and small BG N and NX produce better sonics when used as clock bypasses. BUT the differences you are hearing might not be quite as large if you haven't given the ceramics full break-in time.
YMMV!
Greg in Mississippi
Last edited:
Personally, I kind of doubt the validity of most audio break-in judgments. I know from study of cognitive psychology that human brains adapt, most often without the owner being aware of what is actually happening. That can make it hard to be sure exactly what changed. Interesting that 3-weeks is often quoted for break-in time for stuff. Say, what is the time for audio brain adaptation? For vision its about two or three weeks. That's how long it takes to get used to wearing glasses that flip everything upside down. Suddenly the brain mentally flips the inverted images from the glasses and everything looks right-side up with the glassed on. Then when one stops using the glasses, everything looks upside down without them. Fortunately, visual perception flips back to normal after another two or three weeks. What should one think, that that glasses have a two or three week break-in period? Probably, we would say the glasses don't change, the brain does. Maybe we should give some allowance that something similar could happen with listening.
Wow, that is a long break in Greg. I find BG one of the worst for long break in, but even after 1 day they show their potential, and the longest I believe I have been able to detect improvement is at about 100 hrs. I am not doubting your observation, just surprised.
Checking back it looks like I ran the ceramic caps for around 10 days. I never power my digital section down, so that is 240 hrs. Still a pretty noticeable improvement following diyiggi recommended FCA cap replacement at that stage.
Checking back it looks like I ran the ceramic caps for around 10 days. I never power my digital section down, so that is 240 hrs. Still a pretty noticeable improvement following diyiggi recommended FCA cap replacement at that stage.
@Markw4,
I am quite comfortable in what I heard and reported and that it was not a 'audio brain adaptation', especially as I went through the process 2 times with very similar results.
That is all I will say on this. You are welcome to your opinions and experiences in audio, just as I am. I know from some past exchanges we have some significant differences in our views of audio and what is truth and not. I don't find it productive to discuss that or even try to point out that I have a different view.
I will respect your views and experience and expect you to do the same. Please don't go behind me nay-saying what I said... and I won't do that to you.
@wlowes, I agree that was longer than I expected. I was really dismayed with the sound I got at first, but decided to give it time. It was in a setup I listened to daily and really depended on for musical enjoyment. That it finally settled was a serious relief... and made going through the same process a 2nd time more bearable. I am glad I did give it time the first go around as I learned some valuable lessons in the process.
BG break-in, in my experience, tends to have ups and downs. I don't have as definitive of a read on the stages, though in my experience it has been longer than yours. I've rarely used BGs just by themselves, so it is hard to separate their breakin from that of other components in a build.
AND I don't doubt the improvement with the FCA caps at all. Again, I use components like these a lot and as I mentioned that was one of the things I did to get a better result on that Juli@ modification. Once that worked, I have been using them ever since.
One place I prefer alternatives to PPS and similar SMD films is post DAC filtering. Sometimes I have gotten a 'tizzy' sound there. I really like good polystyrene caps in that application and the LCR polystyrenes from Newark / Element14 have become one of my favorite for this application. BUT one of the values on my last order looks very different from previous buys and I am cautious that the manufacturing process had changed and they may not sound as good... I need to give these new ones a try.
But compared to other options there, the PPS and other SMD film have always sounded better to me than other SMD alternatives such as C0G.
For all, with due repect,
Greg in Mississippi
I am quite comfortable in what I heard and reported and that it was not a 'audio brain adaptation', especially as I went through the process 2 times with very similar results.
That is all I will say on this. You are welcome to your opinions and experiences in audio, just as I am. I know from some past exchanges we have some significant differences in our views of audio and what is truth and not. I don't find it productive to discuss that or even try to point out that I have a different view.
I will respect your views and experience and expect you to do the same. Please don't go behind me nay-saying what I said... and I won't do that to you.
@wlowes, I agree that was longer than I expected. I was really dismayed with the sound I got at first, but decided to give it time. It was in a setup I listened to daily and really depended on for musical enjoyment. That it finally settled was a serious relief... and made going through the same process a 2nd time more bearable. I am glad I did give it time the first go around as I learned some valuable lessons in the process.
BG break-in, in my experience, tends to have ups and downs. I don't have as definitive of a read on the stages, though in my experience it has been longer than yours. I've rarely used BGs just by themselves, so it is hard to separate their breakin from that of other components in a build.
AND I don't doubt the improvement with the FCA caps at all. Again, I use components like these a lot and as I mentioned that was one of the things I did to get a better result on that Juli@ modification. Once that worked, I have been using them ever since.
One place I prefer alternatives to PPS and similar SMD films is post DAC filtering. Sometimes I have gotten a 'tizzy' sound there. I really like good polystyrene caps in that application and the LCR polystyrenes from Newark / Element14 have become one of my favorite for this application. BUT one of the values on my last order looks very different from previous buys and I am cautious that the manufacturing process had changed and they may not sound as good... I need to give these new ones a try.
But compared to other options there, the PPS and other SMD film have always sounded better to me than other SMD alternatives such as C0G.
For all, with due repect,
Greg in Mississippi
Greg,
I didn't mean break-in can't be real. That's why I say things like 'often' and other words allowing for exceptions. Also, my primary concern is I would like for readers of these threads to be exposed to opinions on both sides of issues like the one at hand. In addition, I think it is accepted that a few types of components do break-in, and other things less so. In any case, I don't hear the opamps in DAC-3 anymore, but I doubt it is because they are broken in. I think a more likely explanation is that I don't have a better dac around to compare against, so it is hard to prevent myself from getting used to the sound of the one I use as a reference.
I didn't mean break-in can't be real. That's why I say things like 'often' and other words allowing for exceptions. Also, my primary concern is I would like for readers of these threads to be exposed to opinions on both sides of issues like the one at hand. In addition, I think it is accepted that a few types of components do break-in, and other things less so. In any case, I don't hear the opamps in DAC-3 anymore, but I doubt it is because they are broken in. I think a more likely explanation is that I don't have a better dac around to compare against, so it is hard to prevent myself from getting used to the sound of the one I use as a reference.
Last edited:
Also, a some tips... first, if one looks carefully at the bypass cap side of Ian's new Crystek adapter you'll see that the masked copper on one side of where you mount the caps is all a continuous power run and the other side is ground. SO you can carefully scrap off the solder mask to provide solder pad area for larger caps or even bridge them across multiple pads. Just be sure not to place them where they interfere with the socket.
AND of course this doesn't help with NDK SDA's. I did send a couple pair to Ian to try... perhaps he'll draw up a new adapter for them.
AND as long as you keep them clear of the socket you can stack multiple mid-sized SMD film caps in the bypass area of the adapters easily. I have done this in other applications with the 1uF Panasonic's and the 20uF Rubycon PML caps.
Then I REALLY love @wlowes' and @jacklee's experiments with separate supplies for the different DAC rails and the FiFoPi clocks. I think a really trick setup might be to use 2 of the LiFePO4 boards ganged together... one setup in the typical manner with 2 3.3V outputs (FiFoPi main and one of the DAC rails) and the 2 13.2V for the output stage. But then use the 2 3.3V outputs from the other LiFePO4 board for the other 2 3.3V DAC board rails and setup the other outputs with a single cell each so you could power the 2 clocks with their own cell each. Could be a singing and dancing setup, though at a cost.
AND if you are doing it with an AC supply like @wlowes, consider using the LT3042 regulator boards Ian generously gave out in this group buy. The only issue with them is they are a challenging SMD build. BUT IF you can build them, they are suffiently small you would have more leeway placing them right at the power input pin of each clock adapter.
I've powered the various combos with both the LiFePO4 board and some of my AC-connected power option so far along with some use of Ian's Ultracap experimenters boards. I've gotten good sound with each so far, with the exotic supplies better, but not trashing the AC-connected supplies.
Then sorry I haven't yet done any experiments with various clock bypass cap schemes. I've been focusing on the different I/V stages and transformer outputs Ian included in his GB AND 2 prototype boards I have here from him, one based on the OPA1632 and the other on the OPA861. ALL of the active I/V stages powered by the battery supply sound good, with some tradeoffs that might have one preferring one over another. AND then there's opamp rolling in the IVStd. So far I've focused on either Sparkos or LKS discrete opamps, both of which are sounding good, though different.
AND I've been playing with the filtering cutoff frequencies on the different output stages based on what I'm hearing and on the experience of hearing different cutoff frequencies during the tweaking and refinements of the Allo Katana. I know I'm liking lower cutoff frequencies than stock (and closer to that of the Katana). THIS may be a personal preference thing.
No serious reports yet, I've got a lot of listening and tweaking to do. BUT I have gotten good sound out of each I/V stage and Ian's LL1544A transformer board and would not strike any of them off the list.
Greg in Mississippi
AND of course this doesn't help with NDK SDA's. I did send a couple pair to Ian to try... perhaps he'll draw up a new adapter for them.
AND as long as you keep them clear of the socket you can stack multiple mid-sized SMD film caps in the bypass area of the adapters easily. I have done this in other applications with the 1uF Panasonic's and the 20uF Rubycon PML caps.
Then I REALLY love @wlowes' and @jacklee's experiments with separate supplies for the different DAC rails and the FiFoPi clocks. I think a really trick setup might be to use 2 of the LiFePO4 boards ganged together... one setup in the typical manner with 2 3.3V outputs (FiFoPi main and one of the DAC rails) and the 2 13.2V for the output stage. But then use the 2 3.3V outputs from the other LiFePO4 board for the other 2 3.3V DAC board rails and setup the other outputs with a single cell each so you could power the 2 clocks with their own cell each. Could be a singing and dancing setup, though at a cost.
AND if you are doing it with an AC supply like @wlowes, consider using the LT3042 regulator boards Ian generously gave out in this group buy. The only issue with them is they are a challenging SMD build. BUT IF you can build them, they are suffiently small you would have more leeway placing them right at the power input pin of each clock adapter.
I've powered the various combos with both the LiFePO4 board and some of my AC-connected power option so far along with some use of Ian's Ultracap experimenters boards. I've gotten good sound with each so far, with the exotic supplies better, but not trashing the AC-connected supplies.
Then sorry I haven't yet done any experiments with various clock bypass cap schemes. I've been focusing on the different I/V stages and transformer outputs Ian included in his GB AND 2 prototype boards I have here from him, one based on the OPA1632 and the other on the OPA861. ALL of the active I/V stages powered by the battery supply sound good, with some tradeoffs that might have one preferring one over another. AND then there's opamp rolling in the IVStd. So far I've focused on either Sparkos or LKS discrete opamps, both of which are sounding good, though different.
AND I've been playing with the filtering cutoff frequencies on the different output stages based on what I'm hearing and on the experience of hearing different cutoff frequencies during the tweaking and refinements of the Allo Katana. I know I'm liking lower cutoff frequencies than stock (and closer to that of the Katana). THIS may be a personal preference thing.
No serious reports yet, I've got a lot of listening and tweaking to do. BUT I have gotten good sound out of each I/V stage and Ian's LL1544A transformer board and would not strike any of them off the list.
Greg in Mississippi
@wlowes
Thanks for the encouraging results. That's one more mod to try.
I also observed the fifopi to take around 100 hrs to run-in. Before run-in there is a slight hardness to the low treble and midrange. I originally thought that's the sound of the opamp output so I changed to transformer output but it remained. And then after around 100 hrs the hardness suddenly disappear and stay that way.
This break-in is the biggest hurdle to trying different mods as every change needs time to run-in before it can be properly evaluated.
Thanks for the encouraging results. That's one more mod to try.
I also observed the fifopi to take around 100 hrs to run-in. Before run-in there is a slight hardness to the low treble and midrange. I originally thought that's the sound of the opamp output so I changed to transformer output but it remained. And then after around 100 hrs the hardness suddenly disappear and stay that way.
This break-in is the biggest hurdle to trying different mods as every change needs time to run-in before it can be properly evaluated.
Happily, we can agree on something. I was a little dismayed not to find dedicated regulators solely for the clocks, and for each of the dac rails. Always find some improvement from that in my own modded dacs.
We share the same thought, i.e., powering the clock with yet another battery rail. Well, a newclassD NeutronStar 2 clock will be here soon and that will be powered by a separate 13.2V battery rail so we'll see how well that works.