The Zishan dsd's corner
Jan 9, 2019 at 3:28 PM Post #1,471 of 3,711
I have x3 opa1622 in my AK4497 DSD, configured as output buffers for both SE and balanced outputs, no output resistors or caps. Actually, no input caps either :L3000:
Sounds great, including BA only IEMs, but requires a bit of work to retrofit it.
Next project will be based around INA1620

Hey @Ivan TT

Have you determined that the upper circuit board sends digital audio data, in the form of an SPDIF stream, to the lower DAC / Analog board?
If that's the case, could the DAC / Analog board be used with any source of SPDIF data (as long as the correct power supply voltages are also supplied)?
 
Jan 9, 2019 at 4:41 PM Post #1,472 of 3,711
Hey @Ivan TT

Have you determined that the upper circuit board sends digital audio data, in the form of an SPDIF stream, to the lower DAC / Analog board?
If that's the case, could the DAC / Analog board be used with any source of SPDIF data (as long as the correct power supply voltages are also supplied)?
It is I2S format, not spdif, so the answer is no.
Also, DAC is initialised using control registers, so is very unlikely to work even when used with external I2S source.
 
Jan 9, 2019 at 5:19 PM Post #1,473 of 3,711
It is I2S format, not spdif, so the answer is no.
Also, DAC is initialised using control registers, so is very unlikely to work even when used with external I2S source.

So, is the DAC initialization done via data over the I2S bus, or is there a second data path from the microcontroller on the 10-12 pin header?
 
Jan 9, 2019 at 5:59 PM Post #1,474 of 3,711
So, is the DAC initialization done via data over the I2S bus, or is there a second data path from the microcontroller on the 10-12 pin header?
The latter, but I never had a need to research it tbh.
Check AK4497 datasheet, it has pin functions in what they call register control mode (as opposed to pin control) so you can track these pins down using multimeter.
 
Jan 10, 2019 at 12:05 PM Post #1,475 of 3,711
My first try was to replace the 2xop275 and 1xLM4562 for genuine parts, the sound is more detailed, but lacks body and sometimes is a bit sibilant, I didn't understood why it sounded like this and didn't like it. Later I went back to this option, just to be sure and got the same sound.

My second try was using 2xopa1652, this is musical and I like it, but it's polite, I want a bit of wow factor, not a polite sound.

My 3rd try was 2xLM4562 with opa275, with 4 nichicon finegold and the 2 stock elna, it's better than before, It's musical, has detail, has body, has bass and I enjoy this sound, but I would like to improve this even more.

Meanwhile with all those 3 options I tried a lot of caps, Panasonic and Nichicon.
I use 2 dip8 sockets to fast switch between a lot of caps combinations, without damaging the PCB pads.
In one of the dip8 sockets I made an optional bypass of the last stage transistors
Depending on the position I place the last 2 caps, I can bypass the transistors.

I'm ready for the next try, but not sure what opamps to try, AD8620A is a bit expensive just to try it, maybe another opamp.
I'm waiting for some TRRS 2.5 jacks to try the balanced output, before I proceed.

dip8.jpg



I have x3 opa1622 in my AK4497 DSD, configured as output buffers for both SE and balanced outputs, no output resistors or caps. Actually, no input caps either :L3000:
Sounds great, including BA only IEMs, but requires a bit of work to retrofit it.
Next project will be based around INA1620

After reading this, I order one opa1622 from AliExpress (Hi IC store), I hope it's genuine, I wanted to try one for some time.
Were did you ordered yours from?
Do you know if it possible to buy only the PCB?

At some time, I've read you had 2xAD8066 in LPF are those worth buying for my next try ?
 
Last edited:
Jan 10, 2019 at 1:54 PM Post #1,476 of 3,711
Tried lme49720: worse than original opamp, much bloated and dark sounding.
4 Nichicon FG did not do much, same as the original ones.
Very much interested in tested in trying the coming Muses02 and Ada4637. Just a few more days...
 
Jan 10, 2019 at 2:02 PM Post #1,477 of 3,711
Were did you ordered yours from?
eBay, the cheapest I can get.
Not sure about the pcb only, sorry.
Great idea re: dip sockets by the way!
At some time, I've read you had 2xAD8066 in LPF are those worth buying for my next try ?
I don’t have LPF in my DSD anymore, but 8066, 823, 8599 and opa1688 are all good, just different sonic profiles that may or may not match one’s earphones and preferences.
 
Last edited:
Jan 10, 2019 at 2:26 PM Post #1,478 of 3,711
Tried lme49720: worse than original opamp, much bloated and dark sounding.
4 Nichicon FG did not do much, same as the original ones.
Very much interested in tested in trying the coming Muses02 and Ada4637. Just a few more days...

lm4962 is identical to lme49720. Marketing decided to change the name, but then changed their mind. Today these names are interchangeable and have the same specs.

I mistakenly brought a lme49860 to test against the stock lm4962 just because it was cheap on eBay ($2-3) . It turned out that these two are also identical except 49860 has widely voltage range. But.. 49860 definitely has a clearer and slightly stronger bass and mid bass compared to lm4962. Vocals are more positioned in front of me, unlike the middle between the ears sound by the stock lm4962.
Zisan should have selected 49860 instead. May be it is cost saving.

Anyway, I am in the middle of testing lme49860, ad8066, and Burson V5i on Zishan DSD with stock everything. May be 1 more op amp if I have time. Will write it up next week.
 
Jan 11, 2019 at 11:37 AM Post #1,483 of 3,711
Try 4:
Replaced the LDOs and oscilators, as recommended by @Ivan TT:
https://www.head-fi.org/threads/the-zishan-dsds-corner.826185/page-57#post-14300445
https://www.head-fi.org/threads/the-zishan-dsds-corner.826185/page-61#post-14371597

I'll resume his mod here:
(some termination letters in replacement chips can be different)

3,3V LDO (1 unit in DAC board, 5 units in CPU board):
Stock markings: DE=A1D
Replacement: LP5907MFX-3.3/NOPB
https://www.arrow.com/en/products/lp5907mfx-3.3nopb/texas-instruments

5V LDO (1 unit in DAC board):
Stock markings: 4BMD
Replacement: TPS73250DBVR
https://www.arrow.com/en/products/tps73250dbvr/texas-instruments

Oscilator replacements (1 unit each):
https://www.arrow.com/en/products/dsc1001di2-049.1520/microchip-technology
https://www.arrow.com/en/products/dsc1001di2-045.1584/microchip-technology

My impressions:
To me this was a "big" improvement, all the tiny problems I could identify in sound disappeared, attack and decay is now well defined with more impact, to sum it, it has a more definition and sounds right.
I can now change to other opamps that have more resolution, but before reveled problems in sound, now they reveal their character.
Same thing with caps, now it's ok to reveal definition, there's no need to hide some previous (tiny) problems in sound :)

This mod is a winner, thanks Ivan!
 
Last edited:
Jan 11, 2019 at 3:05 PM Post #1,484 of 3,711
Try 4:
Replaced the LDOs and oscilators, as recommended by @Ivan TT:

My impressions:
To me this was a "big" improvement, all the tiny problems I could identify in sound disappeared, attack and decay is now well defined with more impact, to sum it, it has a more definition and sounds right.
I can now change to other opamps that have more resolution, but before reveled problems in sound, now they reveal their character.
Same thing with caps, now it's ok to reveal definition, there's no need to hide some previous (tiny) problems in sound :)

This mod is a winner, thanks Ivan!

Thank you for your impressions!

I followed the same very helpful suggestions from @Ivan TT
I also ordered 4x caps ECH-U1C102JX5 around dip socked
8x 10ohm thin film resistors for output ERA6AHD100V
1x original lm4562

Then Analog Devices kindly sent me as free samples
2x ad8620 for LPF
And the following opamps for gain to try instead of the 4562:
Ad8066
Lt1358
Lt1364

Can’t wait to receive everything and start soldering!

I have some problem using the zishan as usb dac:
With Windows and MacOS it “works” but it is limited to 16/48. Which is very strange because for ex usb class 1 specification (and with this you don’t need drivers) is limited to 24/96. With android it does not work at all because the usb receiver does not tell the o.s. during the handshake its bit depth/channels/max freq. that it supports. And while win or macOS just use a standard 16/48 Android needs those infos to send the audio.
So, I think the main problem is the usb receiver in our Zishan. Does anyone know what usb receiver is installed? (For sure it is neither xmos nor amanero). And can it be replaced/make it work properly?
 

Users who are viewing this thread

Back
Top