The path to good results is not so easy as one might imagine. Very few seem to be able to do it, careful attention needs to be given to every little detail.
OTOH, trying unproven things on the chance they might be better in some ways without use of careful measurements is pretty useless for arriving at a good design. There are far more ways to get it wrong than to get every little detail right by chance.
That being said, of course if some people like to experiment with this or that just to see what happens, it is absolutely, perfectly fine. I guess we seem to agree on that.
I would only ask that posts with suggestions for that type of experimentation be labeled or stated as being such: For experimentation, not for moving directly closer to achieving that elusive ESS -120dB distortion specification claimed possible for ES9038Q2M.
Only one pic of the insides and it isn't too in-focus/detailed. Looks like probably switching power supplies maybe similar to the box I recently acquired and described. Most likely it has some problems like most of these things do, and being in a small case makes it that much harder to fix if the case is to be preserved.
Mostly, these DAC boards still need better power supplies, better clocks, better AVCC supplies, and fixes for output stages. What the more expensive ones come with may be a little better than the cheaper units, but if none of it is usable as-is to get to that good -120dB distortion the DAC chips are capable of it often means we end up throwing away some better parts that we paid more for because they still aren't good enough to provide for the SQ we are looking for.
Therefore, I find myself coming around to the view that all this looking for better boards to start out with is probably a losing prospect. The one Mikett found may be good if a PRO chip is wanted. The basic board I upgraded may be good if a Q2M chip is wanted. Otherwise, we are probably barking up the wrong tree.
Instead of looking for better DAC boards to fix up we should probably be focused on coming up with a PCB design to make a good and proper basic Q2M DAC. That's what I'm thinking anyway. And, to be honest its been a long time since I had to lay out a PCB myself. I'm sure I could do it but I would have to get familiar with the new tools. Yet, working on that might be better use of time rather than looking at a lot of Chinese DAC boards for modding potential.
Last edited:
For those with USB capability, maybe consider getting something like the E-MU 0202 (AK5385 A/D .00007 THD+N) or 0404, to take basic distortion measurements of your modded boards using software like ARTA, etc. They are inexpensive used (mine only cost $60 shipped) and can provide some guidance as you make changes. As well as potentially inform whether the sound is actually better based on just low distortion figures achieved or other factors, etc.
Last edited:
The Stereophile review is probably a little more accurate. The jitter spurs they show at audiosciencereview are real but only if using USB, they don't occur with SPDIF for example. I rarely use USB myself. Also, what they say about the unbalanced outputs being out of level with each other would seem to out of specification for the DAC so either a problem with the test or with the unit needing attention. In any case, I use balanced out to AHB2 power amp.
DAC-3 is also on Stereophile recommended equipment list, as is AHB2 power amp. The combination is extremely accurate.
In addition, DAC-3 does things to protect against intersample overs, and also to address issues with built-in DAC chip interpolation filters. Those things may not show up on the measurements audiosciencereview does.
Guess we will have to wait and see if Stereophile ever reviews something from Topping. That would be interesting.
Speaking of NX-4, did they stop making it? I see where it says no longer available. I would be curious to listen to one.
The Topping NX4 is a portable DAC. The comparison to Benchmark DAC3 should be made with:
Topping NX7
Topping D50
Which are desktop DAC like the DAC3. It is remarkable how well the mobile chip ES9038Q2M based Topping compared with the ES9028PRO based BenchMark DAC3.
Just like I have always said, right? I keep explaining to people that the distortion numbers between Q2M and PRO differ only by 2dB, -120dB vs -122dB. Where they do differ is mostly in noise, but Q2M noise is probably low enough for us, since we are trying to spend wisely, and we don't give up much there.
That is why I keep trying to get people to see that going to a PRO chip is probably not the best way to spend money to get a good low-cost, near-state-of the-art DAC. The extra cost of the bigger DAC chip and the cost of providing the extra power supply current it needs does not contribute that much to improved sound quality. Spend the extra money saved by using Q2M on a good upsampler would be my advice. If you heard it I expect you would agree.
I've heard most of these budget ESS dacs - about 6 of them over the last 2 yrs - at length and because of that I'm hesitant to agree, compared to my 9038Pro Hifime modded or unmodded. That said, they do sound *excellent*.
As is natural for diy'ers, people will eventually get the PRO boards anyway, and we'll have a better sense then of the differences, if any.
As is natural for diy'ers, people will eventually get the PRO boards anyway, and we'll have a better sense then of the differences, if any.
Just so long as we compare modded to modded. Otherwise, my concern would be that the Q2M DACs were cutting too many corners compared to the PRO DACs. Implementation is more of getting good results than the mobile or PRO chip, seems to me. With equal quality implementations they should sound the same except mostly differ in noise.
I can say my modded Q2M sounds very good as it is now, but all the usual known things have been modded to ESS recommendations. The clock is a good one for being 100MHz. The AK4137 upsampler is a good one for use with a reasonably low-jitter USB to PCM/I2S board (XMOS XU208 in this case). If all those things were done as needed for a PRO chip I don't believe it could sound much better. There isn't much further possible, I don't think.
On the subject of transformers, this seller has a lot of NOS Nuvotem Talema Toroidal Transformers:
neutronllp | eBay
Are these good units? The price is certainly good.
neutronllp | eBay
Are these good units? The price is certainly good.
Why were Pro-Ject, SMSL and Topping able to get close to theoretical ideal SNR and THD values of the Q2M chip, but the Chinese ESS DAC board cannot?
Shall we just buy one of the Pro-Ject, SMSL or Topping instead?
Last edited:
The q2m might be capable of good sound. We can agree on that. However, to get there from a basic circuit board is another matter. I've gone down that path. Also there are severe weaknesses in basic boards that require a massive upgrade to get better sound.
Let's compare first. Op AMP rolling on a basic board...with no other changes. Pretty much a waste of time and effort. Yes, you get a slightly different sound but it is so miniscule as to be perhaps imperceptible to others.
Changing AVCC to a better source. Yes, it is worth it. No question. However, a stock 9028pro board I got from Ebay beats the 38q2m with a LT3042 for AVCC. This partially confirms my opinion with wushuliu and agree with him on a general sense.
So where is the problem???? The lack of separate power supplies and that IV stage.
When I added a better power supply to the Ebay 9028pro board the jump was like I was adding a good AVCC to the 9038q2m.
That IV stage is a major setback IMO.
My latest addition to the 9028pro board also illustrates why adding that IV stage is tricky to these cheap boards.
The balanced IV stage as it looks like all these Ebat PRO boards carry the same layout, does not have extreme high frequency bypassing implemented. I took Markw4 suggestion and added bypasses on all the op amps. This resulted in a much smoother top end. It was not bad at all but its addition make it a bit more refined. To the point it made it sound like vinyl. I will discuss this in more detail in my 9028pro thread.
Moving from 9038Q2M to a 9028Pro board
So there was probably a bit of RF issues on the balanced IV circuit. Yes, you can build a balanced LME497xx stage but you better get it right.
If markw4 had fabricated or someone else had made up an SMD addition to add to the 9038q2m board then there might be more people that could make a good 38q2m board. But as it is, I followed one path saw the limitations I was running into and jumped over and within a month, on working on it, I am near finished. I simply performed some easy mods, purchased some boards from Ebay and got down to it. I think any diyer with some experience would be able to replicate and actually exceed what I now have.
As we know it is not just the chip but the implementation that counts. One party keeps mentioning the chip and its cheapness but that is not all. IF we were designing from the start that might be the situation but we are not so we're stuck with what we have on hand and they don't look nor sound great the more you get into it. Yeah, I was fooled six months ago but I crossed over to the other side. Take the mobile pill or take the pro pill.
That is a decision to make for someone wanting a decent DAC.
Let's compare first. Op AMP rolling on a basic board...with no other changes. Pretty much a waste of time and effort. Yes, you get a slightly different sound but it is so miniscule as to be perhaps imperceptible to others.
Changing AVCC to a better source. Yes, it is worth it. No question. However, a stock 9028pro board I got from Ebay beats the 38q2m with a LT3042 for AVCC. This partially confirms my opinion with wushuliu and agree with him on a general sense.
So where is the problem???? The lack of separate power supplies and that IV stage.
When I added a better power supply to the Ebay 9028pro board the jump was like I was adding a good AVCC to the 9038q2m.
That IV stage is a major setback IMO.
My latest addition to the 9028pro board also illustrates why adding that IV stage is tricky to these cheap boards.
The balanced IV stage as it looks like all these Ebat PRO boards carry the same layout, does not have extreme high frequency bypassing implemented. I took Markw4 suggestion and added bypasses on all the op amps. This resulted in a much smoother top end. It was not bad at all but its addition make it a bit more refined. To the point it made it sound like vinyl. I will discuss this in more detail in my 9028pro thread.
Moving from 9038Q2M to a 9028Pro board
So there was probably a bit of RF issues on the balanced IV circuit. Yes, you can build a balanced LME497xx stage but you better get it right.
If markw4 had fabricated or someone else had made up an SMD addition to add to the 9038q2m board then there might be more people that could make a good 38q2m board. But as it is, I followed one path saw the limitations I was running into and jumped over and within a month, on working on it, I am near finished. I simply performed some easy mods, purchased some boards from Ebay and got down to it. I think any diyer with some experience would be able to replicate and actually exceed what I now have.
As we know it is not just the chip but the implementation that counts. One party keeps mentioning the chip and its cheapness but that is not all. IF we were designing from the start that might be the situation but we are not so we're stuck with what we have on hand and they don't look nor sound great the more you get into it. Yeah, I was fooled six months ago but I crossed over to the other side. Take the mobile pill or take the pro pill.
That is a decision to make for someone wanting a decent DAC.
There is one thing that is mentioned all the time. That these cheap boards are made to entice potential buyers with no knowledge of implementation just that they they are getting all the "features" but sound dreadful.. Markw4 keeps mentioning that aspect all the time. However, he seems to like the 9038q2m but the only place you find them are on these cheapened boards made for the uninformed. So are we leading potential diyers down a really rough path? Remember most of these buyers are starting out and not experienced. More likely if they knew better, they would know not to get it.
I'll admit I started this whole thing in December and out of the gate I did not know what a 9038Q2M was versus a 9038PRO. I actually thought they might be the same thing. I was getting some kind of 9038 and those are supposed to be good. I thought that I could hang a decent power supply on, I had some in the box, roll some op amps and nirvana. Yeah, stupid and ignorant at the time. But I am sure not only me.
I'll admit I started this whole thing in December and out of the gate I did not know what a 9038Q2M was versus a 9038PRO. I actually thought they might be the same thing. I was getting some kind of 9038 and those are supposed to be good. I thought that I could hang a decent power supply on, I had some in the box, roll some op amps and nirvana. Yeah, stupid and ignorant at the time. But I am sure not only me.
Last edited:
Actually that was not such a bad suggestion at all
....Subjectively speaking, IME ADA4898 is a good opamp.
Distortion wise it's a little more complicated, WRT data sheet you have to read between the lines so to speak.
The good - it is fast and it has very good current OP capability, much better than most other opamps.
The bad - it has high(ish) H2 distortion.
Ref data sheet, page 9, figure 22. H3 distortion is better than -125dB at *100kHz. H2 is around -115dB (at typ I-V voltage OP)
3 things to note:
a/ if there is large H2 with non inverted operation, it could be common mode distortion from IP voltage swing and with inverted operation at least some of this will disappear
b/since this is a balanced circuit, as discussed elsewhere H2 will be largely cancelled after I-V
c/ these distortion figures are for 100kHz. At 20kHz they will be lower. Exactly how much lower is difficult to know.
Given all the above it gets my recommendation for I-V to at least try and see where it stacks up.
NOTE - If you use ADA4898 be careful not to short out metal pad underneath which is internally connected to - supply
It's not that we like the PRO better but in real terms it appears that one can come up with better sound easier than if one takes the mobile route. That is the real issue. I think readers should know this up front.
That said, if I had acquired s 9028PRO or even a 9038PRO in December, chances are I would not have ended up with the sound I have today. I would probably have left it stock except for some op amp rolling and maybe just maybe the IV stage power supply upgrade. Which is still two to three rungs below what I now have.
So I still like the 9038Q2M because it was a very valuable learning exercise and I wish to thank you for the suggestions. It was $40 very well spent in the end. However if one really wants an easier path to better sound I would say think about the PRO seriously. However if you really like to tinker and have the skills and requisite abilities then why not the 9038Q2M? Nothing suits everyone perfectly.
That said, if I had acquired s 9028PRO or even a 9038PRO in December, chances are I would not have ended up with the sound I have today. I would probably have left it stock except for some op amp rolling and maybe just maybe the IV stage power supply upgrade. Which is still two to three rungs below what I now have.
So I still like the 9038Q2M because it was a very valuable learning exercise and I wish to thank you for the suggestions. It was $40 very well spent in the end. However if one really wants an easier path to better sound I would say think about the PRO seriously. However if you really like to tinker and have the skills and requisite abilities then why not the 9038Q2M? Nothing suits everyone perfectly.
Hi all, if someone find a cheaper board with pro please let us know. The ideal would be with much less functionalities - it would be the best. (We already built several external parts like I/V stage, power supplies avcc circuits. And we would replace it even a complicated but not perfect pro board. I hope my logic was understandable.