@redjr,
As I said above....
You can find the K&K audio 12watt low voltage supply here:
Other kits – K & K Audio
I modify it with different/larger pre and post regulator filter caps, a polyprop regulator filter cap instead of mylar, different rectifier diodes (although the ones he includes are very good too), an AC CLC filter, AND adding a 3.3V regulated output for things like the input of the isolators on my Soekris DAC setups.
In post #169 of this thread:
Getting the best out of Allo.com's new Katana DAC...
I relate some experiences of a friend's trials with some alternatives in powering the RPi for a Katana setup. My experiences mirror his, in that a good linear supply beats both generic SMPSs and the iFi iPower for this application.
In post #173 of the same thread, @jrocker goes into some details about why it makes sense to power the RPi via the GPIO pins to bypass the protection circuitry in the micro-USB, though not without some dangers.
AND in post #196, I relate my friend's move to the K&K Audio supply and his comments on that.
The K&K Audio low voltage supply IS NOT an end-all-be-all power supply for the RPi. My personal non-modified RPi reference supply is to use 2 Uptone Audio LPS-1.2 Ultracap supplies set to 7V with each feeding 1/2 of an MPAudio dual 3||LT3045 board with fixed setting resistors for 5V and paralleling the outputs to feed the RPi.
AND my all-time reference for an RPi is similar to the one above, but using an RPi 2B that has been modified with an LDOVR.com RPi Mezzanine board to replace the RPi's onboard DC-DC converters with linear-regulated 3.3V & 1.8V. I feed this with a similar dual LPS-1.2 -> dual MPAudio 3||LT3045 board, but with the output of the MPAudio board set to 6V.
Both are significantly better than the K&K supply, but for a LOT more $$$ and in the case of the modified RPi, a good bit of delicate DIY work.
The K&K Audio low voltage supply is just a good-sounding, well-engineered unit that can be purchased at a very reasonable price for a kit. There are others out there that will perform well too.
I'll post details of my mods to the K&K supply here or in that other thread at some point.
Greg in Mississippi
As I said above....
You can find the K&K audio 12watt low voltage supply here:
Other kits – K & K Audio
I modify it with different/larger pre and post regulator filter caps, a polyprop regulator filter cap instead of mylar, different rectifier diodes (although the ones he includes are very good too), an AC CLC filter, AND adding a 3.3V regulated output for things like the input of the isolators on my Soekris DAC setups.
In post #169 of this thread:
Getting the best out of Allo.com's new Katana DAC...
I relate some experiences of a friend's trials with some alternatives in powering the RPi for a Katana setup. My experiences mirror his, in that a good linear supply beats both generic SMPSs and the iFi iPower for this application.
In post #173 of the same thread, @jrocker goes into some details about why it makes sense to power the RPi via the GPIO pins to bypass the protection circuitry in the micro-USB, though not without some dangers.
AND in post #196, I relate my friend's move to the K&K Audio supply and his comments on that.
The K&K Audio low voltage supply IS NOT an end-all-be-all power supply for the RPi. My personal non-modified RPi reference supply is to use 2 Uptone Audio LPS-1.2 Ultracap supplies set to 7V with each feeding 1/2 of an MPAudio dual 3||LT3045 board with fixed setting resistors for 5V and paralleling the outputs to feed the RPi.
AND my all-time reference for an RPi is similar to the one above, but using an RPi 2B that has been modified with an LDOVR.com RPi Mezzanine board to replace the RPi's onboard DC-DC converters with linear-regulated 3.3V & 1.8V. I feed this with a similar dual LPS-1.2 -> dual MPAudio 3||LT3045 board, but with the output of the MPAudio board set to 6V.
Both are significantly better than the K&K supply, but for a LOT more $$$ and in the case of the modified RPi, a good bit of delicate DIY work.
The K&K Audio low voltage supply is just a good-sounding, well-engineered unit that can be purchased at a very reasonable price for a kit. There are others out there that will perform well too.
I'll post details of my mods to the K&K supply here or in that other thread at some point.
Greg in Mississippi
Last edited:
Last week I recieved the sparkos opamps and one broke down during the first day of playing, it got pretty hot! Could this be oscillation? Will it make sence buying another sparkos or will they have the same problem with this design?
What about the burson opamps?
I am now playing with AD797 opamps, using the balanced output (not using the ground) into my tvc volume transformers of my amps.
First impressions are good, more transparant and detailed than the standard opamps.
I also have some OP627 waiting for testing.
On the picture you can also see Elna silmic capacitors on the power supply line.
What about the burson opamps?
I am now playing with AD797 opamps, using the balanced output (not using the ground) into my tvc volume transformers of my amps.
First impressions are good, more transparant and detailed than the standard opamps.
I also have some OP627 waiting for testing.
On the picture you can also see Elna silmic capacitors on the power supply line.
Attachments
Last edited:
@sq225917 It's been about two weeks now, in various configurations for testing.
I Will follow.
Can you also share the modifications you did on the board ?
That says it all. Your power supply impedance may be too high for best sound quality with more suitable opamps. AD797 becomes unstable if impedance at HF is too low. Other opamps such as the recommended OPA1612 love low power supply impedance (at HF, not just DC). If you would like to try a passive solution, enough uf in good quality film caps such as Wima mks4 from +15 to ground and from -15v to ground should fix it. You may be able to start hearing some improvement with as little as 15uf per rail. More uf might help more.
Your answer is puzzling: you mention that high ps impedance is not good but also that AD797 is unstable at low ps impedance??
I use the battery ps with A123 batteries so the ps impedance should be fairly low, or not?
The problem I had was with the sparkos opamps not the AD797. I also think that I have discovered the problem; the smps ps was making a lot of static noise, this has probably led to oscillations of the opamp. This noise was apparant through the complete system so I guess ps quality feeding the battery supply is also important!
I am now using a lineair unregulated pc to feed the battery supply.
I have changed the ps capacitors on the iv board for silmic 470uf with 33uf parallel. In my experience elna silmic is one of the best capacitors for analog stages.
Regards,
Attachments
At HF, its true. If you look at the AD797 data sheet there is advice about adding resistance to large decoupling caps near the dac chip.
In my own experience, I was using two of AD797 as opamp-based AVCC supplies for ES9038Q2M. Adding some film caps in parallel where the +-15v rails entered the board to power the output stage and AVCC opamps made AD797 go unstable. I could have added a small few ohms of resistance in line with +-15v power for AD797 only, but I just switched the AVCC opamps to OPA1612. Also used OPA1612 for the output stage. Film caps fix some graininess in the sound for all audio opamps other than AD797, and in that case OPA1612 sounds better than AD797 did before adding film caps.
In same vein, people have the exact same issue trying to stabilize AD797 as the error amp in a Jung-Didden regulator. Impedance at HF is too low for it.
Other opamps don't have that same particular sensitivity and sound great with some good quality parallel 10uf film caps filtering the +-15v rails to ground. Silmacs might be good, but if they are not as good as film caps at HF (such as Wima mks4 10uf). A combination of mks4 caps in parallel such as 10+10+22+33 could be one possibility (all values in uf, per rail).
Easiest thing for you to do might be to try it. Put OPA1612's in your output stage and maybe ~50uf of parallel film caps from +15v to ground and -15v to ground (or as much parallel film cap uf you can muster for a test). If the film caps change the sound for the better, and they likely will, then you will have a better sounding dac. If you like the sound and want a lower cost solution, that may be possible too. Please let us know if you decide to give it a try. Some people have reported better sound quality with as little as 15uf of film caps per rail, so try what you can reasonably manage would be my suggestion. You might be happy you did.
Last edited:
If true (which I doubt), this seems to indicate a major flaw in Ian's battery supply.
In essence what you are saying is that noise from the PSU charging the batteries affects the function of the opamps in the output stage, even if the batteries are never connected to the load and the charger contemporarily
How did this 'static noise' arrive to the sparkos and how did it kill it?
Please show us this static noise of the PSU on the opamp rails.
How do you know something would be on the rails? Sparkos opamps stand up on the output stage board like an antenna. Perhaps there is radiated RFI/EMI from the SMPS? Or, perhaps ground conducted leakage currents passing from SMPS through the dac boards to some grounded power amp that couple into analog signals?
I've done a fair bit of listening now with the full pi based stack, fifopi, dual dac board and battery supply. Powering the battery board in use from a 16.8v 18650 supply made no difference to using a smps out front. The lower registers are good, soundstage is excellent, but there's something wrong at the top end, voices take on a papery tone, singers lose body and sound oddly multitracked. My Brooklyn fed from volumio doesn't have this quality at all.
Ill let it burn in a bit longer, it has twenty hours on it now. A friends unit has exactly the same quality. Both stacks are set up the same, with same settings on the dac, ldos bypassed, true sync mode enabled. Ill swap to opa1622 and try NDK clocks and see what that does.
I have a feeling avcc may need a good reg on it.
Ill let it burn in a bit longer, it has twenty hours on it now. A friends unit has exactly the same quality. Both stacks are set up the same, with same settings on the dac, ldos bypassed, true sync mode enabled. Ill swap to opa1622 and try NDK clocks and see what that does.
I have a feeling avcc may need a good reg on it.
I made an ES9038Q2M that some listeners with respectable backgrounds in high end audio design say they prefer to Benchmark DAC-3. And a review I just looked at places DAC-3 as preferred over Brooklyn. Just trying to make the point that I have some experience with making ES9038Q2M sound as good as I could figure out after working on it for about a year, and I am a retired EE with a successful career behind me, not a newbie.
To get to the point, if you want this RPi dac to compete with $2,000 class commercial dacs, you have more problems than just AVCC (although a proper AVCC regulator would be a good start). As I keep trying to tell people, some film caps on the +-rails for output stage would likely help too. As would fixing the output stage opamp heating problem (probably due to some RF ingress). Everything you fix should help. Separating VCCA from DVCC power helps too. There are reasons those two things come out on different dac chip pins. In high performance designs they need clean power independent from each other. In addition, the output stage could use more filtering than this dac comes with in order to sound its best (this looks more like an ESS evaluation board filter configuration). One reason for using OPA1612 for all the opamps (if not using directly attached headphones) is because it makes better sounding filters, including for the filtering done by the output opamp. OPA1622 is more of a specialty headphone driver part, that is also an opamp. Also, 8-pin adapters for OPA1622 do not properly ground the internal compensation cap, whether some people hear it or not.
Even doing all the above probably will not get performance better or maybe slightly better than a properly powered Allo Katana (which is very good, but not quite $2,000 class). This dac is a fun project where one can play around with clocks, etc., but its not a fully optimized product finished product, IMHO.
You might be able to take it to the next level if you fix everything you can first, and still want to do something special with it. Easiest things would be to get it off of RPi and on a more powerful host computer. Then with a good USB card and high sample rate native DSD from HQplayer you could probably have a very nice sounding dac. There are other ways too, but that is probably the easiest.
That's not a fair comparison, is it...? Remind us the price of a Mytec Brooklyn DAC...
The Brooklyn is a commercial dac with three tiers of sales channel mark up included in the MRRP, it's also a single chip dac run directly from a smps and with a fair bit of the BOM spent on the phonostage, casework and marketing. It's a perfectly fair comparisson based on the budget allocated and parts used around the dac chip.
The point I'm making is that there's a glaring fault in the sound of Ian's stack that no one has mentioned. I've heard plenty of cheap dacs, Khadas tone board, Allo dacs etc that don't have this flaw. It's not a subtle piece of intonation that i'm talking about here- it's broken/wrong. Not cheap dac sounding wrong.
@Greg, yes the IV is pretty much the ap note from TI. I need to get into it over the weekend and see what's what before moving further down the stack.
The point I'm making is that there's a glaring fault in the sound of Ian's stack that no one has mentioned. I've heard plenty of cheap dacs, Khadas tone board, Allo dacs etc that don't have this flaw. It's not a subtle piece of intonation that i'm talking about here- it's broken/wrong. Not cheap dac sounding wrong.
@Greg, yes the IV is pretty much the ap note from TI. I need to get into it over the weekend and see what's what before moving further down the stack.
Last edited:
It needs ~100 hrs break-in. Before that I didn't care for the sound either. To make it competitive to high-end DAC one will need a few mods:
- Separate (3) battery rails for the DAC HAT
- Bypass the LDO in the fifopi
- Good clocks
For I/V the transformer one is really good. I like that better than the opamp output, even with Sparkos.
- Separate (3) battery rails for the DAC HAT
- Bypass the LDO in the fifopi
- Good clocks
For I/V the transformer one is really good. I like that better than the opamp output, even with Sparkos.
No they really wouldn't make that level of difference.
A single vocalist sounds like a badly multitracked recording sung through a sheet of newspaper. The jitter would have to be gross- utterly huge to have this effect. Even a jDunn sample file with -80db at 1kh doesn't sound this bad through my other dac.
I'll wait for the hundred hours.
OK, understood. If it is that bad, then it does seem like something else is wrong. In my case it was a pretty controlled test of the SQ of the PI with FIFOPi since these were the only parts replacing a BBB with an acko S03 reclock isolator into an existing well sorted DAC. With the test clocks it was surprisingly good, but nothing to compare with a top system. The add of NDK SDA clocks brought it close, but it required the sort of tuning that jacklee mentions to become great. But the break in and tuning was really only the kind of improvement that is subtle enough that could create the kind of debate where some people claim they hear it and it is significant and others would argue that we are imaging what we want to hear.
@sq225917
I don't have this kind of problem. I've compared my ES9038Q2M dual mono DAC with many other high end DACs in my system recently. I like my gears very much. I'll post my setup after I get back from vacation.
I think there could be something wrong. Just hope it can be figured out.
BTW, what are the best sound quality DACs do you think? I'd be very interested in challenging them.
Regards,
Ian
I don't have this kind of problem. I've compared my ES9038Q2M dual mono DAC with many other high end DACs in my system recently. I like my gears very much. I'll post my setup after I get back from vacation.
I think there could be something wrong. Just hope it can be figured out.
BTW, what are the best sound quality DACs do you think? I'd be very interested in challenging them.
Regards,
Ian
The best I've owned was the Weiss 202 (firewire), that's the only piece of kit I regret selling. The Young DAC that came before it had an issue with intersample-overs that was never fixed in new firmware and the Audiolab MDAC that followed it wasn't as good, but -7x the cost.
I'll let it (and me) settle for a week before I start looking at it on the bench.
For the record my setup is Bruno's BPBP with a Khozmo stepped pot, and Tom's Mod86P (bridged) into a set of NS1000's (using RAAL ribbons).
I'll let it (and me) settle for a week before I start looking at it on the bench.
For the record my setup is Bruno's BPBP with a Khozmo stepped pot, and Tom's Mod86P (bridged) into a set of NS1000's (using RAAL ribbons).