ES9038Q2M Board

Member
Joined 2009
Paid Member
ps. We are using discreete opamps to feed the current to..I disagree with the fact that " they don't make a difference" . Please check the whitepapers on Spakoslabs to understand the difference between monolithic and discreet opamp .
When I wrote "But I don't expect it to make any significant difference anyway." I meant when used in voltage mode.
 
I don't have an op-amp that runs in class A. But I don't expect it to make any significant difference anyway.
Even with an ideal op-amp (not in stock anywhere the last time I looked), I wouldn't expect the performance to change compared to using the NJM5532.

As I see it, the distortion is created by the DAC, when operating in voltage mode, not the op-amp.
And the noise is almost entirely defined by the high resistor values in the original circuit. My measurements correlated well with simulations done in TINA TI.
Changing the op-amp from the NJM5532 to LME49720 made almost no difference. Actually I measured an increased distortion of 0.5 dB, but that could easily be measurement uncertainty, temperature differences etc.

It would probably be possible to reduce the noise of the filter/buffer by scaling the component values (lower resistor values/higher capacitor values), but I have a feeling that it might result in increased distortion due to the higher load on the DAC.

If you want low noise and low distortion,the current mode operation is the way to go.

Ahh. I see. Thank you for the explanation. If you are interested I can ship you a spare THS4032, just PM me.

The mystery to me so far is that THS4032 with this board is so distinctive and a cut above the rest of DACs I have heard in a long time. Even LME49720 sounds better than the stock (even with increased distortion).

There must be something else at play here. The second and third harmonics affect tonality and presence. So I don’t mind them as much. But this board with the THS has a clarity and LP like feel that amazes me especially with the minimum phase slow roll off. Feels like the absence of pre-ringing and the one other factor related to the opamp choice is at play here

Cheers.
 
Last edited:
Not sure if this is the right place, but one last (hopefully benign) question for you Jens: acknowledging the differences in layout/integration/gain, how much difference in performance are you seeing between your AK4490 + buffer implementation and your 9038 + 49990 I/V implementation? I'd be using the inexpensive I2S input 4490 boards available so probably not as good as what you've established on your analyzer box, but I'm still curious.

Rough answer off the top of your head, nothing more. :) I expect the differences are tiny and inaudible, provided the implementations are good. Thanks!
 
Last edited:
Member
Joined 2009
Paid Member
At the moment the distortion of my AK4490 implementation is much lower than the distortion of the ES9038Q2M based design, but I have not tried to optimize the ES9038Q2M design, so perhaps it can be improved.
The noise of the ES9038Q2M DAC (with current mode output and LME49990) is lower than the noise of the AK4490 based design.
 
Are you sure the THS4032 is stable?
It is a rather fast device.
If you have an oscilloscope, you should probably check.
Thanks for your offer by the way. I do expect to get a few THS4032's soon, but they are for a different purpose.

Fortunately I have one and I checked. It is completely stable. I was surprised too. I had a heck of a time getting it not to oscillate in my preamp

Let me know if you get a chance to measure and listen to THS. Would love to hear your opinion

Regards
 
Thanks so much, Jens! I doubt these Chinese boards are terribly well optimized probably missing some key details from the EVM/DS layout, and, frankly, they're low enough to not sweat (after redoing the analog side at least).

Looks like the volume control is a little easier on the 9038, so that'll be my leaning.
 
Member
Joined 2017
Paid Member
It seems like your analysis SW does not show the THD. Some of the numbers are a bit confusing, e.g. the PCM1792 meaurement shows the RMS as -111.47 where the third harmonic is around -102.

What is the distortion of the ADC used for testing? At these levels of distortion a passive notch filter may be needed to see the real DAC distortion without uncertainty whether the distortion is from the DAC or from the ADC.[/QUOTE

Sorry, my previous post was a little bit confusing. The attached are FFT by audio Tester. I intended to show S/N of audio Tester, but some calculation was needed. I measured data again today. They were almost same as the previous one, which is what I want. Absolute accuracy below -110dBFS is beyond a DIYer. It's very difficult for a commercial product to achieve such precision. My destination is relative comparison and repeatability.

My ADC is AD7960. FFT by evaluation board of AD7960 outputs almost same data as mine. But of course, more than -120dBFS THD is not guaranteed by the datasheet. What I can say is relatively low THD exists, but absolute value isn't reliable.
 

Attachments

  • ess_TD.jpg
    ess_TD.jpg
    415.9 KB · Views: 954
  • ESS_EVboard.jpg
    ESS_EVboard.jpg
    318.4 KB · Views: 946
  • pcm_TD.jpg
    pcm_TD.jpg
    425 KB · Views: 932
I just got one of the new Chinese ES9038PRO ver 1.06 boards today. I also have an ES9038 based Benchmark DAC-3. The new board sounds really awful in comparison, IMHO. I see what people mean when they say "dry." It's harmonic distortion, probably associated with running in voltage mode. Looks like the output circuit hasn't changed in these new boards compared to a schematic I saw posted somewhere. There is an RC RF filter before the op-amp, then a little EQ shaping with the op-amp, out through an electrolytic, and out to the RCA jacks, near as I can tell with with a quick look anyway.

Since these DACs are really only known to sound good in current mode, and they also apparently have about 100 control registers, I don't know if there are control bits to set them for current or voltage mode. Maybe if someone else knows they could chime in with a yes or no. (I would expect probably its a yes, but maybe not.)

If there is a control bit, I don't think these boards could be recommended. It just doesn't seem like there is much that could be done with them on a DIY basis unless someone wants to get the documentation and write micro-controller code similar to what comes with the DAC board, but to set it for current mode operation.
 



I have this board it can sound rushed and harsh, lively at times, very close soundstage with vocals up close in my setup, I'm powering with I2s Board from DIYinHK, just a bit confused with the pins there are 5 on this board so when you see the pictures of pin configuration they are 6 pin usually for the earlier boards. Not sure what to do with this for mods, in the past low noise power supplies make a huge difference to sound quality. my is green V1.06
 
in the past low noise power supplies make a huge difference to sound quality. my is green V1.06

Yes, low noise power is needed too. Tried it on an LM317 grade type of power supply and the hum and noise were very noticeable. Then tried it on a SilentSwitcher and that problem was fixed. It's still quite distorted compared to a well designed ES3890 implementation. With a good IV output stage (in current mode of course) it would be a lot better. Remaining problems might include clock quality, but at this point it isn't even worth thinking about.

The other thing Benchmark does with the DAC-3 is always run the ES3890 at 210kHz and with the smoothest passband reconstruction filter. Keeping the sample rate up there keeps reconstruction filter transition band out of audio frequencies even for HD sample rate inputs. 210kHz may be a sweet spot for other reasons too, I don't know. To do that they they use a TI SRC chip before the ES3890.
 
Last edited:
Regarding power supply sensitivity, it occurs to me that for other Sabre schematics I have seen there is a signal called AVCC+ or something like, a reference voltage which is supposed to be heavily filtered and buffered with a low noise op-amp. If they are using a voltage divider and cap, probably no wonder if it is sensitive to PS quality.

Could be another problem for distortion even with very clean 15 volts. Don't know.
 
Last edited:
Definitely NOT wanting to stifle conversation, but would it make sense to move the 9038pro discussion away from the 9038q2m or are they close enough that implementations that help one help the other? Honest question before we run into mass confusion.

* I do wish we could fork the technical part from the listening impressions. Very different respective goals.
 
* I do wish we could fork the technical part from the listening impressions. Very different respective goals.

Don't know about a new thread for ES9038PRO, what do others think?

Um, regarding listening impressions, if a goal is about making technical changes to affect SQ then seems to me that directly related listening impressions should be categorized as part of the technical discussion.
Other listening comments, reviews, etc. might well fit better into a separate thread.