@keilau, I read good reports and good measurements of DACMagic+ too. They go for around $350. If the group delay doesn't bother you maybe they are okay. The wah-wah-pedal type phase swish as instruments change pitch was a no-way for me (return label was on it within a couple hours of my first opening the box), but maybe you don't mind or wouldn't notice the same issue. How people hear seems to vary wildly, not in frequency response (for normal hearing), but in terms of other factors some of which may come only with training and practice.
Maybe even a better option, our friend cdsgames and his company, Allo, is in the process of coming out with a new Katana DAC based on ES9038Q2M. The are expected to sell for around $250, I think. They run on Rasberry Pi, again, I think. There is another thread for Allo somewhere in forum, maybe the Vendor's Bazaar section if you may have an interest.
Maybe even a better option, our friend cdsgames and his company, Allo, is in the process of coming out with a new Katana DAC based on ES9038Q2M. The are expected to sell for around $250, I think. They run on Rasberry Pi, again, I think. There is another thread for Allo somewhere in forum, maybe the Vendor's Bazaar section if you may have an interest.
Also, I agree with DPH that you can start improving this Chinese DAC at the output stage, and with high-quality low-noise incoming power.
I would go a step more and say you could improve it again and again as you become more accustomed to the better sound, assuming you have left sufficient room in the case to do so. Pretty sure I am the only one besides cdsgames with personal experience as to exactly how good they can sound at their best or very close to it. If you look at his posts scattered around this thread you can get his opinion. Also, the folks at Allo used AP test equipment to measure their results, if that is important to you.
I would go a step more and say you could improve it again and again as you become more accustomed to the better sound, assuming you have left sufficient room in the case to do so. Pretty sure I am the only one besides cdsgames with personal experience as to exactly how good they can sound at their best or very close to it. If you look at his posts scattered around this thread you can get his opinion. Also, the folks at Allo used AP test equipment to measure their results, if that is important to you.
Maybe I will offer one more bit of commentary. Sometimes people may get the impression a DAC such as this one is weak on bass. Unless there are excessively small coupling caps in the circuit it may not be due to a lack of bass but rather to an excess of low-level distortion. Sometimes some distortion can merely make something sound a bit bright and clear, or as most people seem to hear it, as lacking bass. One's first thought might be that EQ or tone controls sound like they would fix it just fine. In my experience, not so. Tone controls or EQ can make it sound better, but never right since the problem is not being caused by EQ in the first place. High frequency (HF) power supply noise can cause the exact same symptom perceptually speaking. Especially switching power supplies anywhere in the power chain.
That's all I wanted to say. Hard-won lessons your's for nothing. I won't go into all the supporting details, it would take too long.
That's all I wanted to say. Hard-won lessons your's for nothing. I won't go into all the supporting details, it would take too long.
I got my ES9038q2m v1.7 board yesterday and can confirm that unmodified it delivers about -74db THD at 0db. It was a bit better, sort of -80db with digital volume down to -15db. So it requires a decent I/V stage to be added, otherwise there is no point to listen such a DAC.
As a ES9038q2m reference (and as a current headphone amp) I have ProJect Pre Box S2 ( Pro-Ject Pre Box S2 Digital: MQA HW decoding at reasonable cost - Page 30 - DAC - Digital to Analog Conversion - Computer Audiophile ) which I believe represents very much what can be delivered by such a mobile chip. Let see how much can be squeezed out of that SMPCB board intended now for a network streamer with RPI3.
Concerning other e-bay boards I think less mods will require an ES9028PRO board. Just some filtering of AVCC after LT1963 regulators, reduced gain for I/V and increased for filter. Everything else could be left at stock and SQ is pretty good. Perhaps also op amps with higher current output can be considered. With ES9038PRO again total rework of I/V is inevitable.
Last edited:
The Project-Pre S2 looks like an interesting box. Some info here: http://www.box-designs.com/inhalt/en/pdf/preboxs2digital.pdf
Looks like what they did with the clocking is what cdsgames said sounds best and what I am about to try here.
It looks like S2 users may have had some issues with firmware and to some extent EMI/RFI sensitivity. No surprise here on the latter.
Not sure why more don't catch onto the benefits of upsampling. Benchmark and Crane Song got that part right but some others seem to miss it, IMHO, of course. Although I gave a rationale for doing it I am not sure that is the only reason.
Looks like what they did with the clocking is what cdsgames said sounds best and what I am about to try here.
It looks like S2 users may have had some issues with firmware and to some extent EMI/RFI sensitivity. No surprise here on the latter.
Not sure why more don't catch onto the benefits of upsampling. Benchmark and Crane Song got that part right but some others seem to miss it, IMHO, of course. Although I gave a rationale for doing it I am not sure that is the only reason.
Looks like Twisted Pear started testing 'master mode I2S' for Buffalo DACs about ten days ago. Everybody is saying it is an improvement: http://www.diyaudio.com/forums/vendor-s-bazaar/310338-dac-allo-ess9028q2m-rpi-35.html#post5414388
Of course, everybody may have their own way of doing it. Since I am starting with a Chinese DAC and trying to interface with an SRC4392 it results in its own set of issues. Also, since I can't or am not trying to sync a USB-to-I2S converter clock with the DAC clock maybe that would give a different result as well. WIP photo below.
Of course, everybody may have their own way of doing it. Since I am starting with a Chinese DAC and trying to interface with an SRC4392 it results in its own set of issues. Also, since I can't or am not trying to sync a USB-to-I2S converter clock with the DAC clock maybe that would give a different result as well. WIP photo below.
Attachments
I dont agree. ess9038q2m can sound really , really great . Project s2 and topping d50 are text book implementations of the EVM board (sabre). Problem is , EVM design concentrates on good THD+N numbers. All of us DIYers know that those numbers , while important ,don't correlate entirely to SQ.
Markw4, thank you for taking the time to write to me directly. I have much to learn about digital DAC and your input is greatly appreciated. For my purpose, I do not expect the cable box or Blu-Ray player to provide high quality input. May be the I/V output upgrade and the +/- 15 volts supply is good enough for now.
For my main audio system, the DAC function was performed by a Rotel RSP-1570 until recently. I used a Tainyun Zero for the computer sound. This ES9038Q2M is my first DIY DAC project. I looked at all the commercial boxes, including the Topping, Pro-Ject and SMAL using the ES9038Q2M chip, but the draw of DIY is too strong for me. The goal is modest, but should be an improvement over the analogue output of the video sources.
Hi Mark,
is the Chrystek 575 clock the same size as the original clock?
If it is not, how did you connect it to the board?
thank you
Hi terry22, I wrote up the clock replacement but finding old posts can be challenging sometimes. Temporarily removed the volume pot header and then used Chip Quik solder to remove the old clock. The new clock was slightly bigger, but not by much. All I had to do on that was use an X-acto knife to trim away a bit of ground plane near the solder pads to leave a little more clearance. If you look at my photos of the board you can see how the clearance around the pads is bigger as compared to the decoupling caps nearby. Originally they were all the same. If you can't find a photo I could repost that one.
EDIT: Later I went back and put the clock on its own dedicated 3.3v regulator, mostly as an experiment. By then the power issues had been mostly fixed. The clock is HF analog and not really so much part of the digital circuitry, that was the rationale anyway. It was easy to do and I think it may have helped a small amount, but I can't say it was dramatic.
EDIT: Later I went back and put the clock on its own dedicated 3.3v regulator, mostly as an experiment. By then the power issues had been mostly fixed. The clock is HF analog and not really so much part of the digital circuitry, that was the rationale anyway. It was easy to do and I think it may have helped a small amount, but I can't say it was dramatic.
Last edited:
Hi guys,
Does some of you already use this resampler card?
(Or know if it is good stuff or not)
sandy AK4137 DAC SRC Audio 384K 32Bit DSD256 DSD IIS Conversion-in Headphone Amplifier from Consumer Electronics on Aliexpress.com | Alibaba Group
It looks like that it can do some PCM to DSD resampling.
Does some of you already use this resampler card?
(Or know if it is good stuff or not)
sandy AK4137 DAC SRC Audio 384K 32Bit DSD256 DSD IIS Conversion-in Headphone Amplifier from Consumer Electronics on Aliexpress.com | Alibaba Group
It looks like that it can do some PCM to DSD resampling.
Well, I cannot speak about d50 but S2 was designed by John Westlake, whose contribution to the audio engineering you may know. btw he acknowledges that S2 is just a compact DAC and no match to more high-end designs. regarding the numbers I agree, they do not make SQ alone, but there is no SQ without them as well.
That AK4137 SRC board looks interesting. Might be a slightly better than the TI part, at least in some ways. Would have to study the datasheet and sketch out a schematic of a physical board to be sure one way or the other.
It might also be workable with 'master mode I2S' but not sure about that with this SRC or the TI part. If one were to look at that last picture I recently posted it showed some modification of the Chinese DAC board to get access to a DAC signal which required temporarily disabling the DAC board SPDIF input. Could have been the TOSLINK input instead but I decided probably easier to do with the SPDIF input. There is probably a separate pin on the PRO DAC chips for that, but not for mobile. I have mostly scoped out how that would work with SRC4392. Would take some time to consider the same or similar for the AK4137.
It might also be workable with 'master mode I2S' but not sure about that with this SRC or the TI part. If one were to look at that last picture I recently posted it showed some modification of the Chinese DAC board to get access to a DAC signal which required temporarily disabling the DAC board SPDIF input. Could have been the TOSLINK input instead but I decided probably easier to do with the SPDIF input. There is probably a separate pin on the PRO DAC chips for that, but not for mobile. I have mostly scoped out how that would work with SRC4392. Would take some time to consider the same or similar for the AK4137.
Hi,
can you play 16bit PCM? I have es9038q2m v1.04 board connected with Raspberry pi3 via I2S. I use volumio with generic I2S driver. Mp3 files are OK, but flac (16bit/44k)files or output from Spotify plugin are unusable. I hear noise with little bit music in right channel. Both leds LOCK and DSD led are on. When I set resampling to 24bit, noise is gone and I can hear music. (This not working for Spotify plugin). I tried find some info on web, but I found only two relevant links. One is thread in volumio forum ES9038 I2S driver, DoP : DACs - Volumio, second note is about unsupported 16bit on this board ESS ES9038Q2M I2S DAC Module Board 32bit 384khz Balanced DSD 1x LT1963 Regulators - Audiophonics
Have some one info about 16bit support? Thanks
can you play 16bit PCM? I have es9038q2m v1.04 board connected with Raspberry pi3 via I2S. I use volumio with generic I2S driver. Mp3 files are OK, but flac (16bit/44k)files or output from Spotify plugin are unusable. I hear noise with little bit music in right channel. Both leds LOCK and DSD led are on. When I set resampling to 24bit, noise is gone and I can hear music. (This not working for Spotify plugin). I tried find some info on web, but I found only two relevant links. One is thread in volumio forum ES9038 I2S driver, DoP : DACs - Volumio, second note is about unsupported 16bit on this board ESS ES9038Q2M I2S DAC Module Board 32bit 384khz Balanced DSD 1x LT1963 Regulators - Audiophonics
Have some one info about 16bit support? Thanks
Regarding 16-bit operation, no problem here with Windows. How is the DAC board connected to the Rasberry Pi3? Using SPDIF, I2S, TOSLINK?
Most likely there is some problem with the interface formatting. Left or right justified, frame size, etc.
Regarding Spotify in particular, the DAC only supports PCM or DSD. There is no MP3 decoder, the computer OS has to do that. Spotify mp3 bitrates are very low, 128kbps, IIRC. The PCM sample rate might only be 22kHz, 11kHz or something like that, which maybe is lower than the DAC supports.
Windows and most OS'es will resample the default sound device audio stream to some common sample rate that can be user configured. They do that so that multiple audio streams can play at once, such as system sounds like mouse clicks, alarms, notifications, etc., playing at the same time while music is playing. Therefore they may send the DAC a standard format it knows how to play.
Rasberry Pi3 OS may not do that, probably not since it takes some computing power to do the SRC in real time even if not all that high quality.
Most likely there is some problem with the interface formatting. Left or right justified, frame size, etc.
Regarding Spotify in particular, the DAC only supports PCM or DSD. There is no MP3 decoder, the computer OS has to do that. Spotify mp3 bitrates are very low, 128kbps, IIRC. The PCM sample rate might only be 22kHz, 11kHz or something like that, which maybe is lower than the DAC supports.
Windows and most OS'es will resample the default sound device audio stream to some common sample rate that can be user configured. They do that so that multiple audio streams can play at once, such as system sounds like mouse clicks, alarms, notifications, etc., playing at the same time while music is playing. Therefore they may send the DAC a standard format it knows how to play.
Rasberry Pi3 OS may not do that, probably not since it takes some computing power to do the SRC in real time even if not all that high quality.
Last edited:
For ESS recommended circuits please see the "Maximizing ...performance... " document in the link that follows. Please read carefully and completely. ESS Technology :: Downloads
One thing that is not shown in that document is how to derive the Reference voltage from AVCC. That part is shown in the Evaluation Board Schematic available at the same link.
One thing that is not shown in that document is how to derive the Reference voltage from AVCC. That part is shown in the Evaluation Board Schematic available at the same link.
Last edited:
Markw4
one question I have has always been about is whether an OPAMP by itself can deliver the current that AVCC requires. ( Especially if
The LME49720 can deliver a max of about 26ma per channel.http://www.ti.com/lit/ds/symlink/lme49720.pdf) Looking at the table last one on your link, for the ESS 9008 it requires 37ma at 3.3 V for the highest clock frequency. Is this not exceeding what the op amp can deliver?
The 9038SQM board supposedly requires approx 120ma as delivered on the positive rail and only 20ma on the negative suggesting that the 3.3V requires up to approx 100ma. Do we know how much current approx AVCC requires? Especially when you intend to upconvert. Would this not exceed what the LME49720 can deliver? I see the 9008 is an older 8 channel DAC so the current requirements are higher than the SQM variant but then ESS recommends using an op amp still for the Sabre Pro line.