hi, there,
mark is right
green board i2s lrck to pi gpio pin 12,
green board i2s bclk to pi gpio pin 35,
green board i2s data to pi gpio pin 40,
green board i2c sda to pi gpio pin 3,
green board i2c scl to pi gpio pin 5,
disable the on board mcu by j1, j2 on.
pls read post #4293 for more detailed.
cheers
Silabs clock chips are not suitable for use in dacs, although they are getting more sly about hiding the pertinent information designers need. In one distantly linked document they still state:
" The jitter was integrated from 12 kHz to 20 MHz and all of the outputs are LVDS at 2.5 V"
https://www.silabs.com/documents/pu...mizing-jitter-performance-next-generation.pdf
Which is not the frequency range that we need for dacs.
Seems to me they used bury that info less.
Anyway, for dacs it matters what jitter or phase noise is all through the audio band including the critical band between 10Hz and 1Hz. That's why clocks intended for audio use often show graphs and tables of phase noise and jitter down to 1Hz.
On the other hand, 12kHz is the bottom limit of jitter performance that usually matters for RF use. Most standard clocks are not rated for jitter performance below 12kHz. That's why we use very particular clocks for audio and we have to be very careful about ground planes and dedicated voltage regulators for out clocks.
Also regarding Silabs clocks, in the same document linked above you might want to look at the following section:
1.4 Avoid Using CMOS Output Formats in Jitter Critical Applications
Graph showing limits over which they measure is shown in:
Figure 2.1. Spurs and Harmonics Example for CMOS to CMOS Outputs
(...more graphs shown in the appendix.)
In the graphs they show phase noise down to 100Hz, but it worsens a lot below that much like we see with 1/f noise. If it is -76dB at 100Hz, that's already worse than we would like to see it at 10Hz and even 1Hz. There are good reasons these clocks aren't used in high quality commercial dacs.
Last edited:
hi, there,
yes, you are right, my system is,
digital file -> rpi ( volumio ) -> I2S -> es9028q2m.
|_____volumio plugin______i2c ______|
the volumio plugin controls the dac's status including i2r filter, interpolation filter, digital phase locked loop, and harmonic distortion compensation. in my opinion, the plugin is able to control the ak4137 too. a logic analyzer to read a working ak4137 i2c is a good starting point.
i had a ak4137 board for a long time but never use it, i also have a linux driver for it. pm me if you would like to work together for the plugin.
cheers
A real back to basics one here guys but please bare with us 'beginners'!
I am going to improve AVCC as others have done here simply by adding some extra capacitance. Initially I am adding 2200uF Pana FC as a start. Then I am going to try what Democles did. This will be a 0.3F 'super' cap's and feed both AVCCs from seperate 3.3v supplies.
I see he bypassed the 0.3F with some oscon type Polymer hybrid electrolytic.
CDSgames mentioned bypassing the SC with 4 film caps. I assume he uses 4 to get up to a value required that isn't available in one single film cap. Or for the added ESR of multiple caps.
Is there a go to bypass cap you could direct me to...or atleast the value required? ...please.
I also have a THS4032 to try as mentioned by madds.
Thanks
I am going to improve AVCC as others have done here simply by adding some extra capacitance. Initially I am adding 2200uF Pana FC as a start. Then I am going to try what Democles did. This will be a 0.3F 'super' cap's and feed both AVCCs from seperate 3.3v supplies.
I see he bypassed the 0.3F with some oscon type Polymer hybrid electrolytic.
CDSgames mentioned bypassing the SC with 4 film caps. I assume he uses 4 to get up to a value required that isn't available in one single film cap. Or for the added ESR of multiple caps.
Is there a go to bypass cap you could direct me to...or atleast the value required? ...please.
I also have a THS4032 to try as mentioned by madds.
Thanks
if we leave high end DACs alone, then for instance Silabs 532 would be a good choice for a DIY DAC and beyond...imho.
Not so sure about that here. Allo got very low jitter with Katana dac using NDK SDA clocks for ES9038Q2M. Cost of the dac is around $250. Not really high end at that price. Check out the jitter measurements for it at ASR if you like, they look pretty good.
In addition, Allo's next dac, a 5v USB type will use the same clocks and is expected to sell for less than $200. Probably have great jitter performance too.
No way Allo could do as well with any Silabs clock, IMHO, although no disagreement having programmable clocks seems pretty cool from a techie point of view.
Last edited:
Hi Mark, is that 9038 unit you recommended for around $30 on ebay about 4months ago still the best low cost choice for someone with an amanero? I think it says smp cb on the board so you know what I am talking about. can go up to 60 on price if theres something new worth the extra cost. I see many new boards for the 9038 on ebay now and about to buy one. Thanks. Philip
Last edited:
Hi Philip,
The SMP CB dacs with ES9038Q2M (or their clones) are the ones this thread is mostly about. The last version that was popular here was 1.07. People liked it because the MCU I2C bus activity could be stopped by installing J1, J2 at the same time (at least with the most commonly supplied firmware).
Now there is a newer version SMP CB dac board without a version number printed on the pcb. Some of those come with a small display that plugs directly into the top of the dac board, no cable. It also has two I2S input connectors that are switched into the dac chip using a logic chip. In addition there is a new MCU. Don't know if there is a way to make that MCU stop I2C bus activity. If not, it means the MCU I2C bus pins could have to be physically disconnected from the dac chip if one wanted to use Arduino or RPi can take control (which can be used to help improve sound quality).
Regarding Amanero USB to I2S boards, they are fine with most any dac if its okay that the USB board be clocked at 22/24MHz rather than 45/48MHz (as is more commonly seen with XMOS USB boards). For ESS dacs running in asynchronous mode (the default) either type of USB board is fine. For AKM dacs or synchronous mode ESS dacs, we would have to talk more specifically about how one might use Amanero or XMOS for best sound quality results.
Hope that answers what you wanted to know. If not, please say so and we can give it another try.
The SMP CB dacs with ES9038Q2M (or their clones) are the ones this thread is mostly about. The last version that was popular here was 1.07. People liked it because the MCU I2C bus activity could be stopped by installing J1, J2 at the same time (at least with the most commonly supplied firmware).
Now there is a newer version SMP CB dac board without a version number printed on the pcb. Some of those come with a small display that plugs directly into the top of the dac board, no cable. It also has two I2S input connectors that are switched into the dac chip using a logic chip. In addition there is a new MCU. Don't know if there is a way to make that MCU stop I2C bus activity. If not, it means the MCU I2C bus pins could have to be physically disconnected from the dac chip if one wanted to use Arduino or RPi can take control (which can be used to help improve sound quality).
Regarding Amanero USB to I2S boards, they are fine with most any dac if its okay that the USB board be clocked at 22/24MHz rather than 45/48MHz (as is more commonly seen with XMOS USB boards). For ESS dacs running in asynchronous mode (the default) either type of USB board is fine. For AKM dacs or synchronous mode ESS dacs, we would have to talk more specifically about how one might use Amanero or XMOS for best sound quality results.
Hope that answers what you wanted to know. If not, please say so and we can give it another try.
Last edited:
Don't know. My ES9038Q2M boards have I2C wires superglued to the top of the MCU. Maybe someone else is able to take a look
Attachments
Last edited:
The should sound the same, the differences are not the kind that affect sound quality. Sorry.
Cool. I'll be hacking the registers before any soldering. I live for hacking..... My own stuff, guess you could call me an aspiring white hat. Never hacked anyone else's property, just don't have it in me to do that. Will be searching this thread and asking questions if I get stuck. Pretty excited at the moment.
Quite an interesting thread, all 500 pages of it. I found interesting info from Markw4's posting.
Looking at the usb to i2s XMOS boards though, i dont want to be dependent on drivers which are shareware, i cant find any good drivers for the XU208 (do all xmos chipsets use the same usb audio class 2.0 drivers?) Does anyone know any good drivers for this that work?
Anyway the fact that it has a spdif input is pretty good but i mobo's realtek drivers dont work whatsoever so no spidf for me.
Isnt the ES9038q2m a 60$ chip though? How can they sell an entire implemented board of it for half the price?
What is the forte of the 9038q2m vs the 9028q2m?
Another interesting thing was the mention of the ES9028Q2M board, apparently it has a higher gain output, so a theoretically lower SNR. That combined with the fact that it has balanced outputs makes it somewhat attractive of an offer, although i think balanced doesnt matter much for headphones. I like the inclusion of DIP socket opamps for easy rolling. I wonder if i could throw out the O2 amp out of my chain if the gain is high enough on this (it has AC 15-18VX2 + AC 7-12V after all). I wonder if it could drive the HD650's without an amp...
Although that means i would need a transformer, while the ES9038Q2M board will take 15V of DC. Actually 2 transformers, one dual output 18V (with the middle post being the ground right?), and one single output 12V. I would probably buy the cheapest ones and at that point there's no point in getting this board because the sound quality might suffer from this alone. Plus i dont really want AC lines lying around, i'd have to figure out a case.
Looking at the usb to i2s XMOS boards though, i dont want to be dependent on drivers which are shareware, i cant find any good drivers for the XU208 (do all xmos chipsets use the same usb audio class 2.0 drivers?) Does anyone know any good drivers for this that work?
Anyway the fact that it has a spdif input is pretty good but i mobo's realtek drivers dont work whatsoever so no spidf for me.
Isnt the ES9038q2m a 60$ chip though? How can they sell an entire implemented board of it for half the price?
What is the forte of the 9038q2m vs the 9028q2m?
Another interesting thing was the mention of the ES9028Q2M board, apparently it has a higher gain output, so a theoretically lower SNR. That combined with the fact that it has balanced outputs makes it somewhat attractive of an offer, although i think balanced doesnt matter much for headphones. I like the inclusion of DIP socket opamps for easy rolling. I wonder if i could throw out the O2 amp out of my chain if the gain is high enough on this (it has AC 15-18VX2 + AC 7-12V after all). I wonder if it could drive the HD650's without an amp...
Although that means i would need a transformer, while the ES9038Q2M board will take 15V of DC. Actually 2 transformers, one dual output 18V (with the middle post being the ground right?), and one single output 12V. I would probably buy the cheapest ones and at that point there's no point in getting this board because the sound quality might suffer from this alone. Plus i dont really want AC lines lying around, i'd have to figure out a case.
Windows drivers for XMOS boards are either limited functionality freeware or more full featured paid versions. Which driver works with which XMOS board also depends on the firmware programmed into the XMOS. For the foregoing reasons it is pretty much a waste to buy cheap ebay XMOS boards unless PCM and possibly SPDIF with default drivers is okay for one's purposes.
If native DSD support is wanted and or if more freedom from sound quality damage caused by Windows Sound Engine is wanted, it is required to use ASIO drivers which are only supported in paid version drivers.
For the above reasons, because of very low jitter high quality clocking, and because of built-in galvanic isolation it is recommended to use JLsounds I2SoverUSB cards: I2SoverUSB - I2S over USB Audio
They are about $15 in single digit quantities.
The '28 has higher output current, but IIUC, it has less other features than '38.
Looks like no volume control or ability to select interpolation filters. Don't know about other things that might be missing since I don't have a data sheet for '28. The Sabre volume control is very good to have, it doesn't damage the audio like virtually all software volume controls do since the Sabre volume function is dithered to avoid loss of bit resolution. http://www.esstech.com/files/3014/4095/4308/digital-vs-analog-volume-control.pdf
Opamp of choice for modern switched resistor dacs is OPA1612. If it doesn't sound right, there is a problem somewhere else, maybe the surrounding circuity and or component quality, and or the +-15 power supply quality/performance.
Problem with the balanced output from a standard three opamp output stage is that there is still a lot of RF noise from the dac that has not been filtered all the way. The last opamp does more filtering as well and common mode distortion cancellation and any needed dc offset removal. For those reasons, and for one example, Benchmark DAC-3 rebalances the SE output so that the balanced outputs are fully cleaned up. For an alternate example, in newer versions of ESS evaluation boards there are additional OPA1612 opamps after the I/V opamps to buffer and filter the balanced outputs they use.
Not really, unless one is willing to accept a lot of added distortion. Some people try to use OPA1622 as the differential summing opamp with the idea that it could drive headphones. One problem is that the DIP-8 adapters don't ground the OPA1612 ground pin which causes increased distortion.
A better idea would probably be to stick with OPA1612 for all opamps and if on a limited budget, mod an ebay LME49600 headphone amp board to improve the sound quality some. If a bigger budget, maybe a Massdrop THX 789: Drop: Passions lead here (formerly Massdrop)
What works pretty well is a linear +-15v supply for AVCC and the output stage, and a separate 5v power supply for the 3.3v regulators for the various digital functions. For the 5v supply, and ebay LT3042 board works well:Low Noise LT3042 Linear Regulator Power Supply Module For Amanero XMOS DAC Power | eBay
There are lots of options for the +-15v supply. NJM7815/NJM7915 are good enough although there will be a little line ripple if only a single stage regulator. What cleans up the opamp sound quality if dac is otherwise cleaned up enough to warrant it would be some large value film caps from each of the +-15v rails to ground. Everyone hears an improvement with that, but depending on the quality of film caps they can be kind of costly. However, low cost ones can help some too. One ebay 18v/18v/9v/9v R-Core transformer is a good choice to power everything since they are particularly good at not coupling into the dac HF/RF noise from the AC line (which is always present in today's world of electric motors, computers, cell phones, etc). Separate transformer windings for +-15v supplies are better than center tapped since one may choose where to tie the grounds together, ideally only at the dac board if the transformer quality allows for it.
Last edited:
That's a lot of information to take in, thank you for the detailed and honest explanation, it helps me make a more informed decision.
Thank you for this info. Yeah i realised the limitations of the freeware ones, i was just wondering if there was some unofficial driver that everyone was using that supported ASIO or WASAPI i wasnt aware of, but i guess if there's no then, i think i just might give up on the I2S idea as the 70$ is a bit steep for me for an I2S-usb interface.
So the chinese units wont work with those drivers because of the wrong firmware PID not matching with the licensed software from the jlsounds site?
I think windows 1703 actually has USB audio class 2.0 built in, but unsure of the functionality, it seems its generic so would it work with any chinese XMOS chip? The prospect of not using the default windows mixer but system wide WASAPI is enticing.
That was a pretty informational pdf, i tried looking for more infor on the ess site but cant find where they are located, do you know where there are more?
I dont get this part, it takes AC 5-13V or DC 7-18V and outputs 0-15V (if input is AC then output is AC?). To feed it 5-13V AC i'd need a transformer anyway right? so why not plug that transformer directly into the DAC board?
Does the voltage have to be DC for the filter cap mod? Would the caps mess with AC signal?
Thank you for this info. Yeah i realised the limitations of the freeware ones, i was just wondering if there was some unofficial driver that everyone was using that supported ASIO or WASAPI i wasnt aware of, but i guess if there's no then, i think i just might give up on the I2S idea as the 70$ is a bit steep for me for an I2S-usb interface.
So the chinese units wont work with those drivers because of the wrong firmware PID not matching with the licensed software from the jlsounds site?
I think windows 1703 actually has USB audio class 2.0 built in, but unsure of the functionality, it seems its generic so would it work with any chinese XMOS chip? The prospect of not using the default windows mixer but system wide WASAPI is enticing.
That was a pretty informational pdf, i tried looking for more infor on the ess site but cant find where they are located, do you know where there are more?
Someone in the reviews part of the aliexpress link rolled the opamps for OPA1612's, from what i can see.
I thought the output stage had 4 opamps? Maybe im just misunderstanding the board, i will trust you.
Couldnt i jumper the DIP leg to any ground trace to ground it? Sorry if it is a bad question
Does that mean you recommend i roll all the NE5532DD's for OPA1612 or did i get something wrong? I guess im going to stick with the O2 if i cant run the hd650 straight out the balanced output. Actually if i cant use the balanced outs (since i have to run it into a non balanced amp), then i might as well get the 9038q2m board discussed in this thread, or this ES9028Q2M+ES9028 one and just be done with all the I2S and similar issues. Do you think its a bad board?
I dont get this part, it takes AC 5-13V or DC 7-18V and outputs 0-15V (if input is AC then output is AC?). To feed it 5-13V AC i'd need a transformer anyway right? so why not plug that transformer directly into the DAC board?
This is a good tip to know, thanks. I didnt know i could use one R-core trafo, that's good.
Does the voltage have to be DC for the filter cap mod? Would the caps mess with AC signal?
Not full-featured WASAPI unfortunately. The drivers work for PCM and SPDIF if the board has a hardware output for it. No Exclusive Mode support. Problem with Windows Sound Engine explained in post #410: https://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-410.html#post5731319
ASIO is highly recommended is all I can say.
Very little ESS information is public domain. They require one to sign an NDA obtained from a distributor for access to data sheets and possible answers to questions. Maybe sometimes google can be a friend, don't know.
If you mean 5532 was replaced with OPA1612, that would be one step in the right direction.
If switched away from OPA1612, it just means they have other problems they don't understand. Seriously. They could always try asking here if they would like some help.
ESS does have a little old but at least public advice: http://www.esstech.com/files/4514/4095/4306/Application_Note_Component_Selection_and_PCB_Layout.pdf
You could ground the OPA1622 ground pin, which is usually not connected to a DIP-8 adapter pin. If you keep the wire as short as possible it should help. Still not quite as good an output stage device as OPA1612. OPA1622 is more optimized for use as a simple, low cost, low distortion headphone driver.
Link is broken for me. Happy to take a look if you can repost it.
Actually, I didn't look closely at the power supply. It can probably work as they say, but with the default output stage and AVCC supply, it would be awful sounding junk to me, so I wouldn't consider using any of that. Some people try the default dac sound and find it okay if not great, and a few others are like me and say it is so bad as to be unlistenable. Don't know how you would feel about it.
Thing about dacs is that what the dac chip does is set some limit on the very best a dac board can sound depending on the quality of the surrounding design implementation. In that regard, one can make a very good dac indeed using ES9038Q2M, but it is never cheap or easy. Its the board circuitry implementation that adds cost and complexity if the dac chip is to sound its best. Implementation is huge in terms of the effect on sound quality so much so that some people exaggerate a little and say the dac chip doesn't matter at all, only the implementation does.
Yes, it only applies to the DC rails powering the opamps. One could try it with the DC rails after the voltage regulators and just before the opamps wherever things are located. Could be the power supply is on the dac board or on a separate power supply board. Mostly, we want the caps as physically close to the opamps as is reasonably possible.
EDIT: I should also probably mention that eziitis has recommended Nazar regulators as an alternative to expensive film caps. I have been wanting to try that out for myself, but too many competing projects to get things done as quickly as I would like. If you are interested in giving it a try, there is a schematic eziitis posted we could find for you.
Last edited: