The power supply looks decently regulated to me. Are you able to measure PS noise at all? Or is it just an A-B comparison with another DAC leading you to think it's noisy?
My only suggestion is perhaps change out all 12 diodes to Schottky types but I doubt if you can hear a difference.
If your OpAmps can't take the current it will distort and give you a harsh sound.
Hence my suggestion to use 1622 on the XLR side and 1611 on the following RCA side.
Burson OpAmps sound good in most other applications but cant take high current and I already killed 2 with my old 1794 based DAC.
My only suggestion is perhaps change out all 12 diodes to Schottky types but I doubt if you can hear a difference.
If your OpAmps can't take the current it will distort and give you a harsh sound.
Hence my suggestion to use 1622 on the XLR side and 1611 on the following RCA side.
Burson OpAmps sound good in most other applications but cant take high current and I already killed 2 with my old 1794 based DAC.
Its not usually power supply noise so much that makes a dac sound harsh. It is power supply impedance over the entire range of audio frequencies up to 100kHz or so for the analog components. There are some different approaches to help with that including different kinds of electrolytic caps, sufficient number and sizes of film caps, or removal of most filter caps except X7R decoupling caps and use of fast shunt regulators, etc. The quickest easiest test to see if you can get its attention would be to solder some film caps together in parallel and add them to each of the +-15v rails for the opamps. I have written elsewhere about using Wima mks4 in the following values from each of the +-15v rails to ground: 10+10+22+33+33 (all values in uf, and, yes, I know its a lot of film caps). So far everyone who has tried the film caps has found immediate sound quality improvement, even if using a total uf value quite a bit less than I did.
AVCC power quality matters a whole lot too for Sabre chips. For that, I would avoid LT304x regulators which most people seem to sound 'flat and lifeless' in that application. ADM7150 or ADM7154 are used for some good dacs. Twisted Pear sells some AVCC dual regulators and single ADM7150 regulators called Trident-SR, but the board you have has the AVCC pins for both channels shorted together, IIRC, so only a single good AVCC supply is needed and can be used. If Trident-SR isn't too expensive you might try that. Or, they still have some of older Trident (pre-SR version) AVCC regulators for about $40. Probably way better sounding that what you have now.
AVCC power quality matters a whole lot too for Sabre chips. For that, I would avoid LT304x regulators which most people seem to sound 'flat and lifeless' in that application. ADM7150 or ADM7154 are used for some good dacs. Twisted Pear sells some AVCC dual regulators and single ADM7150 regulators called Trident-SR, but the board you have has the AVCC pins for both channels shorted together, IIRC, so only a single good AVCC supply is needed and can be used. If Trident-SR isn't too expensive you might try that. Or, they still have some of older Trident (pre-SR version) AVCC regulators for about $40. Probably way better sounding that what you have now.
Last edited:
Index of /9038
@markw4 thanks for the detailed info. Looks like if I were to make PS mods I need to pull the thing out of chassis.
Anyway I am very happy with the sound. Handily beats a MSRP$10,000 DAC based on the 9018.
@idir The caps may not be fake, different outer coat I guess?
http://ec2.snoopy.org/9038/IMG_6214.JPG
http://ec2.snoopy.org/9038/IMG_6215.JPG
@markw4 thanks for the detailed info. Looks like if I were to make PS mods I need to pull the thing out of chassis.
Anyway I am very happy with the sound. Handily beats a MSRP$10,000 DAC based on the 9018.
@idir The caps may not be fake, different outer coat I guess?
http://ec2.snoopy.org/9038/IMG_6214.JPG
http://ec2.snoopy.org/9038/IMG_6215.JPG
I have both the 9028Pro and 9038pro DAC boards. To my ears both boards sound nicer than Topping D50 and SMSL D9. There is some kind of immediacy of the sound of the 9028/38Pro boards that I did not get from the D50 or D9.
Not sure whether the 9038Pro board is a significant improvement over the 9028Pro board. They sound very similar to me. Opamp rolling and even adding different bypass caps to the +-15 rails seem to make more discernable differences than the differences between the two DAC chips.
Kurt
Not sure whether the 9038Pro board is a significant improvement over the 9028Pro board. They sound very similar to me. Opamp rolling and even adding different bypass caps to the +-15 rails seem to make more discernable differences than the differences between the two DAC chips.
Kurt
Last edited:
@idir recapping the whole thing does make a difference.
I am using Nichicon KS for the pre-regulation caps
KW for various 220uF and 10uF
Elna RFS 25V 47uF for all 47uF positions.
So whether the original caps are fake or otherwise is still a mystery.
@Kchang agree with the OpAmp rolling. The trusty old OPA627s need not apply here with the 9038.
I am using Nichicon KS for the pre-regulation caps
KW for various 220uF and 10uF
Elna RFS 25V 47uF for all 47uF positions.
So whether the original caps are fake or otherwise is still a mystery.
@Kchang agree with the OpAmp rolling. The trusty old OPA627s need not apply here with the 9038.
I don't know about replacing the existing caps depending on the their values as they may be helping some at LF, but you might disconnect the outputs of the +-15v supply if possible, and run them a little ways to where you can put a set of parallel film caps, then run the output of that back to the board 15v rails. If you use thick enough wires and the distance isn't too long it would probably work okay. If not, you could always put it back the way it was.
Not having a board to look at myself, not sure what may be possible. Also, if I'm not mistaken, the board you have looks a lot like the one Mikett opened a thread on modding several months ago. I think he may have done some work to improve the power supplies, but don't know specifics.
Not having a board to look at myself, not sure what may be possible. Also, if I'm not mistaken, the board you have looks a lot like the one Mikett opened a thread on modding several months ago. I think he may have done some work to improve the power supplies, but don't know specifics.
@markw4 with my high voltage projects i always find ways to mount my trusty Panasonic 500V 50uF films that look like WIMAs andmount on inner wall. Here the need to use 220uF 50V at the regulator output makes for a challenging size esp given the miniature chassis. Anyway this DACin its current form handily kicks my Silbatone anyday w any type of music.
@jlw1611 make a big improvement in terms of distortion, especially when playing loud. Thank you for that! I will also order 1622 to compare to my current LM49860.
There is still a big area for improvements IMHO. I will Replace PSU with a decent one not sure which one though. I already some LT3042 that I will use as regulators for the digital part.
Then filter OPS with tantal+x7r. as @Markw4 suggests (thanks!) in the thread for ES9038QM These are all my plans for now.
There is still a big area for improvements IMHO. I will Replace PSU with a decent one not sure which one though. I already some LT3042 that I will use as regulators for the digital part.
Then filter OPS with tantal+x7r. as @Markw4 suggests (thanks!) in the thread for ES9038QM These are all my plans for now.
I spruced mine up a bit as i by default dont trust chinese caps from ebay. Changed out most of the electrolytic caps and the bypasses. Swapped the JRC 5532/34 for 4562NA and LME49710 since i had them on hand. Gotta run for now but here are some pics.
Imgur: The magic of the Internet
Imgur: The magic of the Internet
Two comments:
1. The solder points on the topside (the upper half in your picture) of the PCB will be covered by existing components, such as the e-caps and opamps (or their sockets). You will not have access to them so cannot use them for bypassing. It would be much easier to solder small caps for bypassing on the underside of the PCB. For big fat film capacitors that cannot fit in the space underneath the board, you have to find creative solutions to form electric contacts from the top.
2. The upper-left bypass connection in your picture is incorrect. The red line in your picture should end on pin 4 of the opamp, but now it ends on a signal path. The lower-left connection end points are also ambiguous.
1. The solder points on the topside (the upper half in your picture) of the PCB will be covered by existing components, such as the e-caps and opamps (or their sockets). You will not have access to them so cannot use them for bypassing. It would be much easier to solder small caps for bypassing on the underside of the PCB. For big fat film capacitors that cannot fit in the space underneath the board, you have to find creative solutions to form electric contacts from the top.
2. The upper-left bypass connection in your picture is incorrect. The red line in your picture should end on pin 4 of the opamp, but now it ends on a signal path. The lower-left connection end points are also ambiguous.
Last edited:
@kchang
1. I will most probably use soic to dip so will for sure be able to solder SMD caps inder the opamps
2. Sorry for my drwaing skills ;-) all the lines are supposed to be between +-15v and GND.
Now I have heard many options on what caps to use the 2 that speak to me most are:
- 10 uF tantal + 0.1 uF Ceramic X7R
- 0.1 uF MKS or MKP
@idir
The 47uf electrolytics are usually for lower frequency reserve power, whereas immediate bypassing or decoupling from supply pins to ground for opamps is mostly for high frequency stability to avoid oscillation or ringing of the opamp. For digital circuits and maybe for faster opamps the decoupling caps would provide reserve current for fast pulses and fast edges of transitions, which in digital circuits can help reduce chances of logic errors.
The idea of bypassing at device power pins is based on using a tantalum and ceramic in parallel with closer to the pin than the tantalum if they have to be at different distances. The expectation is that the tantalum can provide lower impedance at lower frequencies than the ceramic can, and at higher frequencies the tantalum will become lossy and the ceramic will have low impedance. In that scenario one would not expect to have resonance problems at the power pins due to too many capacitors of different sizes and with different parasitic inductance forming multiple high-Q tuned circuits, which is why this method is recommended. Therefore the tantalum should be one that is not good at very high frequencies, and the ceramic should be X7R (not C0G or NPO which are too high-Q).
On the other hand, film caps aren't very good for bypassing since they are not as low impedance as ceramic at higher frequencies and they have much more parasitic inductance. My recommendations for using film caps to best effect has been written up in other threads.
One other thing, I think when Mikett went through making some power supply changes to the board he has that looks a lot like this one, IIRC, he said that adding local bypassing helped the sound. You might check his thread on the subject to see what he reported.
The 47uf electrolytics are usually for lower frequency reserve power, whereas immediate bypassing or decoupling from supply pins to ground for opamps is mostly for high frequency stability to avoid oscillation or ringing of the opamp. For digital circuits and maybe for faster opamps the decoupling caps would provide reserve current for fast pulses and fast edges of transitions, which in digital circuits can help reduce chances of logic errors.
The idea of bypassing at device power pins is based on using a tantalum and ceramic in parallel with closer to the pin than the tantalum if they have to be at different distances. The expectation is that the tantalum can provide lower impedance at lower frequencies than the ceramic can, and at higher frequencies the tantalum will become lossy and the ceramic will have low impedance. In that scenario one would not expect to have resonance problems at the power pins due to too many capacitors of different sizes and with different parasitic inductance forming multiple high-Q tuned circuits, which is why this method is recommended. Therefore the tantalum should be one that is not good at very high frequencies, and the ceramic should be X7R (not C0G or NPO which are too high-Q).
On the other hand, film caps aren't very good for bypassing since they are not as low impedance as ceramic at higher frequencies and they have much more parasitic inductance. My recommendations for using film caps to best effect has been written up in other threads.
One other thing, I think when Mikett went through making some power supply changes to the board he has that looks a lot like this one, IIRC, he said that adding local bypassing helped the sound. You might check his thread on the subject to see what he reported.
Last edited:
@idir arent the caps labeled 104 0.1uF which are already mighty close to the +15V -15V pins of each OpAmp?
I am curious if you are planning on replacing all the 47uF Silmic IIs with 10uF then pulling the existing film caps (I think you might be using Silver mica on your board) and doing a direct to pin 0.1uF for each rail of each OpAmp? To me this sounds a bit weird.
I am curious if you are planning on replacing all the 47uF Silmic IIs with 10uF then pulling the existing film caps (I think you might be using Silver mica on your board) and doing a direct to pin 0.1uF for each rail of each OpAmp? To me this sounds a bit weird.
In terms of power supply bypassing, I've been wondering whether it'd better to leave good enough alone. Even in stock form, the 9028/38 pro boards do not seem to have any issue of the opamps going into oscillation. Perhaps I really should hoop up my oscilloscope to look for any sign of oscillation, but I have not heard anything particularly wrong/unnatural that would cause me to suspect the existence of oscillation. If there is no oscillation, why try to chase the ghost by adding a fairly random combination of caps in an attempt to "solve" a non-existing problem?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.