Okay. You guys posted links to two different boards. Neither is a new one at all.
Regarding the one with the row of LEDs, they didn't have displays and rotary controls before. These are not recommended for modding, as the ground plane is mostly used up, or broken up, on the back side due to the layout they use.
Last edited:
Hi Mark & other members,
Am still working on the DAC. At the moment, got the dac fried again and waiting for the new board.
My problem is am very forgetful(famous for that am afraid) so accidentally turned off before disconnecting the i2c lines.
Since two of the sparkfun translators I tried didn't work(goes to show the bad quality we have here) I had to connect without any protection.
While it was working, the DPLL setting to 1(best) did enhance the sq to my hearing. Regarding HD compensation, was thinking of making esp's twin tee notch filter and Asus Xonar soundcard for measurement if I can.
While waiting for the board, went back to rpi allo piano in dual mono with Kali reclocker and moode player. Measurements say 112db dnr/snr for this rig. Compared to our modded dac, this sounded ordinary and less clean.
Also been working on my amp which is a ti tpa3255evm board with a meanwell smps. Put a Murata line filter($7.19) on the smps output and implemented Post Filter FeedBack(Ti's own).This PFFB on paper takes the amp from 112 dnr/snr to 116.5 and about halved the output noise @ 46uV. Very welcomed enhancement in SQ.
On another matter that Mark talked about in this forum a while ago. Windows will upsample all audio(when not using ASIO/WASAPI) and make it worse. Upgraded my TV Set-top box to Android box which can play youtube, netflix, primevideo etc. Playing youtube with it direct to my amp with rca connection was surprised to hear how much better it sounds. Am guessing must be because the box has no upsampling capability. Want to look at Windows Server OS which some say can bypass windows audio processing entirely. This ofcourse only applies to browser players like youtube. For Audio only, HqPlayer has the best quality to my ears.
Cheers,
Kay.
Hi Mark,
Thanks for your commiserations and concern. If you send me, am sure would be no problems, but no worries. I have another Arduino nano which I will dedicate to this job only. We only learn by mistakes, so I should have enough experience now to take care what to do and not to do. I enjoy the learning process immensely.
Your help to me and other members has been outstanding.
Cheers,
Kay.
If not a 3.3v Arduino, perhaps I could send a little level translator board from the US? It could fit in a letter envelope. Just let me know by PM if you want/need anything along those lines and where to send them, since there is obviously a problem getting some parts there.
The thing is you need to find a way to protect both the dac chip and the Arduino from inadvertent damage. Nobody is 100% attentive 100% of the time, so it makes sense to design for reliability.
Also, I very much appreciate you kind words for what I have been trying to do here.
It remains a problem that useful information is spread out in so many places around the thread which makes things hard to find find for any newcomers. Don't know what to do about that.
Ok, my replacement 1.07 board has arrived.
Time to start doing the mods again, I think this time I might do a few things differently with the placement of the regs and use connectors to make the assembly and disassembly far easier!
I had a brief listen to the board as it was the first time in months since I heard the original board, oh dear, it's not until you do some of the mods do you realise just what you can improve!
Time to start doing the mods again, I think this time I might do a few things differently with the placement of the regs and use connectors to make the assembly and disassembly far easier!
I had a brief listen to the board as it was the first time in months since I heard the original board, oh dear, it's not until you do some of the mods do you realise just what you can improve!
Yes, indeed. I always start with the clock first before I have too much other work invested. Just in case, you know. If you have trouble soldering the new clock, don't try to force finish it. Put some Chip Quik on it and take out back out, clean up the solder pads, and re-plan the operation, would be my suggestion. Also, try hard not to scratch up the solder mask when removing/replacing clocks. If it does get scratched, please let me know and post focused close up pics. It may be possible to fix it before proceeding. There is some high temp epoxy stuff people use for that, but it may not be the only solution. Intact solder mask makes soldering a lot easier if there is no risk of solder sticking where its not wanted.
It is amazing how it keeps getting better and better until every single recommended thing described in the thread is done. It can end up sounding a little bit better than a $2,000 dac if you get AK4137 sounding its best, along with all the other mods done.
Last edited:
I have just pulled a trigger on DAC8 PRO from Okto, very excited to get it in some future.
For now I will keep the idea of integrating the minisharc "in my pocket" and will try to get it to work with software based crossover.
Regarding the balanced to unbalanced buffers, I got a reply from Pavel that he would actually also recommend the approach I proposed above and to use the OPA1612 as an op amp. So I will wain until the unit arrives to see what resistor values they have there and will then probably ask for more support here to integrate everything correctly.
Meanwhile I want to ask if OPA1622 can be used instead with somewhat similar results or should I look for OPA1612 all in all?
Thanks.
For now I will keep the idea of integrating the minisharc "in my pocket" and will try to get it to work with software based crossover.
Regarding the balanced to unbalanced buffers, I got a reply from Pavel that he would actually also recommend the approach I proposed above and to use the OPA1612 as an op amp. So I will wain until the unit arrives to see what resistor values they have there and will then probably ask for more support here to integrate everything correctly.
Meanwhile I want to ask if OPA1622 can be used instead with somewhat similar results or should I look for OPA1612 all in all?
Thanks.
OPA1622 could be considered if you want to drive headphones directly, otherwise OPA1612 is better for line level outputs. There are other ways to make a great headphone amp, but OPA1622 offers very good performance in a simple package, particularly if higher efficiency headphones are to be used. If using OPA1622 at all, it is important not to use the 8-pin dip adapters since that will degrade performance. The ground pin needs to be connected to the real ground plane, not a simulated ground (nor left floating). Also, there is a solder pad on the bottom of the chip that needs to be cooked onto the PCB, not soldered with an iron.
Last edited:
Okay, then OPA1622 would not be the best option since no plan to use it with headphones. What about this one then with the adapter: ?
OPA1611 Single OPA DIP8 (Unit) - Audiophonics
Have you looked at the AK4137 board with the view of doing mods? All I've done is give it dedicated Dexa NewClassD reg, although if I connect a USB XMOS/Amareno I will have to give it a little more power.
I did use AK4137 with my ES9038Q2M board project, but it was very complicated to get working at the very best sound quality due to some very quirky timing issue. Modification of the AK4137 board and further modification of the ES9038Q2M board were both required, and I also had to add a fancy clock divider board. I need to come back to it at some point and see if I can figure out an easier way to do it.
Hi Mark, there is a calculation in ASR for i/V and output filter resistor values to avoid "ESS hump" issue. Did you (or someone else) check if it matches with your latest recommendation for HW output stage design?
Hi Freezebox,
The schematic we have remains unchanged, but easy to make the adjustment if someone wants to. Just replace the Vref divider with a pot (there is one Vref divider for each channel). When playback is stopped the dac outputs should be sitting at AVCC/2, which might be good to verify. Measure the I/V DC output voltage with a meter and adjust the pot (for that channel) to make the I/V DC output = ground = 0v. Once the new Vref voltage is found that way, you might want to replace the pot with two fixed resistors equivalent to the pot setting you found.
The schematic we have remains unchanged, but easy to make the adjustment if someone wants to. Just replace the Vref divider with a pot (there is one Vref divider for each channel). When playback is stopped the dac outputs should be sitting at AVCC/2, which might be good to verify. Measure the I/V DC output voltage with a meter and adjust the pot (for that channel) to make the I/V DC output = ground = 0v. Once the new Vref voltage is found that way, you might want to replace the pot with two fixed resistors equivalent to the pot setting you found.
Have you checked if offsetting the Vref from AVCC/2, to e.g. a lower voltage, has any influence on the distortion or other performance aspects?
By the way, has anyone made performance measurements lately? It would be interesting to see some objective measurements of the performance achieved after the modifications. And of course, a before/after comparison.
The whole purpose of that change is lower measured distortion in 'ESS Hump' region, which is down at much lower output levels than full scale. Lots of discussion about it over at ASR, as I said before.
I am aware than Allo has done that with the above change during development of their forthcoming USB dac. They say it worked to eliminate the 'hump.' Using a 10-layer board and careful design they say they have been able to get measured ES9038Q2M distortion down as low as -126dB to -127dB range according to their AP analyzer. They may sacrifice a bit of that distortion performance to try to improve sound quality, but not more than a couple of dB, IIUC.
