Oh Boy, flames to come...
Short answer: not as good as mine, which is 20y old but back then SOA.
But it has potential. Get the feeling though that despite all the hype and figures DAC chip makes far less to the sound than its environment.
Long answer...
I am lucky to come from a family where both my parents and grand parents were top opera singers and I am used to real music and also recording, at international level. I am not a musician, I am not a golden ear, but I am rumoured to be picky and I know what I like. Let's say I am an average Joe with trained ears and brains.
I was one of the main contributors of tha DI/O hype 20y ago and I still use that DAC, or what is left of it after many DIT mods. I wanted though to find out how modern DACs stand, 6 generations after my AKM Codec...
I helped a friend building a non expensive HIFI system, built around Klipsch RP-8000F. Hi demand is far less demanding than mine but I wanted a nice system for him and that was a nice exercice. We tweaked a Class D amp (in fact 2) quite intensively and it is reported elsewhere on that forum.
The source is stil a problem. He wanted the lastest options and so we tried the current Chinese offering in the 400$ bracket as that seem to make financialy sense. I started as an engineer and later black belt 6 Sigma, so know figures, but coming from Formula 1 I know also their limits... Anyway, we trusted the ASR forum and purchased several DACs at the same time to return the ones that wouldn't be good, thanks to A...N. In fact we kept them for more than a month and some are still around.
I tried these SMSL and Topping etc. DAC and expected them to be far better than mine. My old DAC played them to the wall. 3D stage, details, micro dynamics (huge gap), dynamics overall (huge gap), bass depth (huge gap), slam, distorsions / harshness... shocking. OK, admittely not a world as for amps or LS, but quite suprising and defo no need to swap. Admittely these played in my system, which is not in the same class, but once these DAC played in his system we found quite the same, perhaps a tad less in amplitude due to a lesser resolving nature.
We played a lot with external PS and Mark Johnson's excellent filter, and that moved things indeed to the right direction. Say from 100 points missing down to still 65 missing, call it a third of the way. It also conforted us that we were right and all things moved... towards my good old dac. That was only having to notch PS. Some are still OK re pricing given the result (they made these dacs acceptable), but I cam to the conclusion that no DAC working with a 5V could bring us what we needed.
Regardless what people say re measurements (say current usual measurements), they IMHO defo don't say it all and you have to be deaf not to hear differences between so called inaudible differences and lowest distorsion etc. levels. The ASR approved dacs sounded quite different, and mods we did to his Class D amps engenced also the sound albeit not very measurable by normal std. Anyway, not the debate here.
The DAC that retains currently our attention is not one with 4499 in the 900$ bracket or the latest toys. We were very surprised by the SMSL SU-9, using ES9038 PRO. It is quite a move in the right direction and could do the job. It is a bit typed re character, which has its positive... and limits. We want to keep its drive and fundament but see what wan be done in the other registers and 3D soundstage and transparency.
Currently I figure out that this still affordable DAC could be a nice starting point if we stick to it. I had a look inside and there might be some small bits to do re various decoupling, not too difficult it appears. Not really DIY, more tweaks and trial and error until we get what we want. Note it sounds acceptable in my friend's system, and probably enough, so we don't want to waste it for the sake of science (somthing we were ready to do with Class D). On the other hand, if a few clever SMS parts can enhence the existing...
Looking into it
Claude
Short answer: not as good as mine, which is 20y old but back then SOA.
But it has potential. Get the feeling though that despite all the hype and figures DAC chip makes far less to the sound than its environment.
Long answer...
I am lucky to come from a family where both my parents and grand parents were top opera singers and I am used to real music and also recording, at international level. I am not a musician, I am not a golden ear, but I am rumoured to be picky and I know what I like. Let's say I am an average Joe with trained ears and brains.
I was one of the main contributors of tha DI/O hype 20y ago and I still use that DAC, or what is left of it after many DIT mods. I wanted though to find out how modern DACs stand, 6 generations after my AKM Codec...
I helped a friend building a non expensive HIFI system, built around Klipsch RP-8000F. Hi demand is far less demanding than mine but I wanted a nice system for him and that was a nice exercice. We tweaked a Class D amp (in fact 2) quite intensively and it is reported elsewhere on that forum.
The source is stil a problem. He wanted the lastest options and so we tried the current Chinese offering in the 400$ bracket as that seem to make financialy sense. I started as an engineer and later black belt 6 Sigma, so know figures, but coming from Formula 1 I know also their limits... Anyway, we trusted the ASR forum and purchased several DACs at the same time to return the ones that wouldn't be good, thanks to A...N. In fact we kept them for more than a month and some are still around.
I tried these SMSL and Topping etc. DAC and expected them to be far better than mine. My old DAC played them to the wall. 3D stage, details, micro dynamics (huge gap), dynamics overall (huge gap), bass depth (huge gap), slam, distorsions / harshness... shocking. OK, admittely not a world as for amps or LS, but quite suprising and defo no need to swap. Admittely these played in my system, which is not in the same class, but once these DAC played in his system we found quite the same, perhaps a tad less in amplitude due to a lesser resolving nature.
We played a lot with external PS and Mark Johnson's excellent filter, and that moved things indeed to the right direction. Say from 100 points missing down to still 65 missing, call it a third of the way. It also conforted us that we were right and all things moved... towards my good old dac. That was only having to notch PS. Some are still OK re pricing given the result (they made these dacs acceptable), but I cam to the conclusion that no DAC working with a 5V could bring us what we needed.
Regardless what people say re measurements (say current usual measurements), they IMHO defo don't say it all and you have to be deaf not to hear differences between so called inaudible differences and lowest distorsion etc. levels. The ASR approved dacs sounded quite different, and mods we did to his Class D amps engenced also the sound albeit not very measurable by normal std. Anyway, not the debate here.
The DAC that retains currently our attention is not one with 4499 in the 900$ bracket or the latest toys. We were very surprised by the SMSL SU-9, using ES9038 PRO. It is quite a move in the right direction and could do the job. It is a bit typed re character, which has its positive... and limits. We want to keep its drive and fundament but see what wan be done in the other registers and 3D soundstage and transparency.
Currently I figure out that this still affordable DAC could be a nice starting point if we stick to it. I had a look inside and there might be some small bits to do re various decoupling, not too difficult it appears. Not really DIY, more tweaks and trial and error until we get what we want. Note it sounds acceptable in my friend's system, and probably enough, so we don't want to waste it for the sake of science (somthing we were ready to do with Class D). On the other hand, if a few clever SMS parts can enhence the existing...
Looking into it
Claude
Layout for IV-stage
Hi, if anyone interested, I can provide gerber data / schematics / BOM for an IV stage based on the experience gathered in this group. It can be used as an IV addon for the usual ES9038Q2M boards, and it contains also a few low noise LDOs for every auxiliary functions. As I designed it, I realized, that the only thing I need from the original board is the ES9038Q2M chip itself... So finally I also placed the footprint of this chip as well on this PCB...
Only drawback of this design, that it's not DIY friendly: only for experienced guys with good soldering experience. (SMD only).
It's a small PCB, with 4 layers, 63x92mm, so, quite cheap to produce at JLCPCB (I paid 7$ + postage for 5 pieces).
I have produced one piece for myself, transferred the ES9038Q2M from the original board to this one, sounds quite good, however I cannot measure it without proper equipment. I just run it on DSD signals (AK4137), cannot evaluate PCM.
Hi, if anyone interested, I can provide gerber data / schematics / BOM for an IV stage based on the experience gathered in this group. It can be used as an IV addon for the usual ES9038Q2M boards, and it contains also a few low noise LDOs for every auxiliary functions. As I designed it, I realized, that the only thing I need from the original board is the ES9038Q2M chip itself... So finally I also placed the footprint of this chip as well on this PCB...
Only drawback of this design, that it's not DIY friendly: only for experienced guys with good soldering experience. (SMD only).
It's a small PCB, with 4 layers, 63x92mm, so, quite cheap to produce at JLCPCB (I paid 7$ + postage for 5 pieces).
I have produced one piece for myself, transferred the ES9038Q2M from the original board to this one, sounds quite good, however I cannot measure it without proper equipment. I just run it on DSD signals (AK4137), cannot evaluate PCM.
Maybe. Most of my layout experience is with Cadence Virtuoso. So a little different from PCB.
I bought pcbCAD720 but it is a bit too different from what I would like in a CAD program. I was considering trying KiCAD a second time. Then there is also Eagle.
Any suggestions or recommendations? Three op amps with nice bypassing, ground plane and good capacitor and resistor choices should be a reasonable starting project.
Hereby I try to attach schematics and layer structure of his PCB. Check, if it suits your needs. If needed, gerber data / excel BOM can be added as well. Let me know.
View attachment ivstage.pdf
View attachment ivstage_layers.pdf
The TOSLINK receiver has active electronics in it, so it needs +3.3v power. Its output is already at CMOS voltage levels, so it can go right into the dac chip.
Also, IIRC the inverter used for SPDIF buffering has a feedback resistor around it so as to make it behave a little more like a linear amplifier.
Also, IIRC the inverter used for SPDIF buffering has a feedback resistor around it so as to make it behave a little more like a linear amplifier.
Last edited:
Claude,
I think I mentioned in the AK4499 thread that its CD audio is not perfect. I know of one forum member who has been a very careful listener for a long time who bought one and decided to return it because of one particular imperfection he really didn't like. He decided to stick with his modded ESS dac instead. What it was that he didn't like about D90 was the sound of vocals (IIUC it was 16/44 playback, don't know if DSD was tried). Anyway, there is some graininess that can be heard at vocal frequencies and above, although some people may never notice it.
I think I mentioned in the AK4499 thread that its CD audio is not perfect. I know of one forum member who has been a very careful listener for a long time who bought one and decided to return it because of one particular imperfection he really didn't like. He decided to stick with his modded ESS dac instead. What it was that he didn't like about D90 was the sound of vocals (IIUC it was 16/44 playback, don't know if DSD was tried). Anyway, there is some graininess that can be heard at vocal frequencies and above, although some people may never notice it.
Last edited:
Mark i want to ask your opinion, i want to use 24,576mhz clock due to low phase compared with 80 or 100mhz clocks. For sure if i do that i am able to support 48khz for spdif. Which is enough for cd player 44.1khz and tv connection 48khz( saw few models support 192khz but not a big deal).
Please comment if i miss any other usage of coax or toslink connections.
Please comment if i miss any other usage of coax or toslink connections.
On page 52 of the datasheet there is a table titled "Analog Performance." That is where they show what MCLK frequencies can work with which audio sample rates. IIUC, using a 24.576MHz clock would allow a for maximum sample rate of 192kHz, but in synchronous mode only. Looks like 48kHz and 96kHz could be in either async or sync mode.
Regarding use of SPDIF and or TOSLINK, its possible to route the signals into the dac chip in different ways. In the case of the dac board we usually use in this thread, SPDIF goes in to one of the two GPIO pins. TOSLINK goes into the other GPIO pin. That can be fine if the GPIO pins are not needed for anything else, but its not the only possible way to do it. For example, something like SN74LVC2G157 could be used to route either SPDIF or TOSLINK into a shared GPIO pin, leaving the other pin free for other uses.
For the experimentally inclined, there is something of possible interest that AKM did with the AK4499 evaluation board, and that should be possible try with ES9038Q2M. The way that would look would be to add a small, say, 10R resistor (or less) between the output of the AVCC regulator and AVCC input pin on the dac, one resistor per channel. On the dac chip side of the resistor they put an electrolytic cap more or less to ground (it would have to go to ground in the case of ES9038Q2M). The cap value was adjusted to affect the sound in a subjectively pleasing way. The downside of doing that is that measured HD is increased.
So, why would anyone make a really good dac, do everything possible to make distortion and noise as low as possible, then muck it up a bit by adding more distortion? Only way to find out is to try it and see what you think. Please bear in mind that some electrolytics can take a week or so break-in period. Some may sound smeared, others not. It may take some time to find one you like. Fellow forum member Abraxalito recently described using heat from resistors placed next to electrolytics to change their properties a bit. Haven't tried that one myself. The caps AKM used for AK4499 were United Chemi-Con (also known as UCC) KMG series.
So, why would anyone make a really good dac, do everything possible to make distortion and noise as low as possible, then muck it up a bit by adding more distortion? Only way to find out is to try it and see what you think. Please bear in mind that some electrolytics can take a week or so break-in period. Some may sound smeared, others not. It may take some time to find one you like. Fellow forum member Abraxalito recently described using heat from resistors placed next to electrolytics to change their properties a bit. Haven't tried that one myself. The caps AKM used for AK4499 were United Chemi-Con (also known as UCC) KMG series.
Last edited:
OMG. Set aside any questions regarding the validity of the above (since I already know your answer, it's related to the "low level noise" and "non-stationary distortions"), what makes you thing any ES9038 DAC/board would behave the same as the AK4499 DAC/board? Have you tried this modification on one or more ES9038 boards, or you just powered up your fan?
For some reason AKM still use the relatively noisy voltage regulator design, which needs filtering to achieve a low noise level from the DAC. And the passive filter requires a large capacitor to avoid excessive distortion at low frequencies.
IMHO they should rather use a very low noise voltage regulator with a low output impedance, e.g. the LT3042, and eliminate the RC filter. That's what I did with the AK4490 in the RTX6001.
But if added distortion is seen as an advantage, you may of course consider using the RC filter.
IMHO they should rather use a very low noise voltage regulator with a low output impedance, e.g. the LT3042, and eliminate the RC filter. That's what I did with the AK4490 in the RTX6001.
But if added distortion is seen as an advantage, you may of course consider using the RC filter.
"For some reason AKM still use the relatively noisy voltage regulator design, which needs filtering to achieve a low noise level from the DAC. And the passive filter requires a large capacitor to avoid excessive distortion at low frequencies."
Ah, but what is the reason? What if the filter is not primarily for noise?
Ah, but what is the reason? What if the filter is not primarily for noise?