I2C is a pretty low speed bus, and it is usually run more towards the low end of its possible speed with these DAC boards. Seems unlikely the I2S lines would be particularly noisy. If the lines are truely noisy they could be filtered because I2C doesn't need really high frequency fast rise-time edges to work.
If there is noise from the microcontroller much more likely it is from ground noise coupling, and maybe from radiated RFI/EMI from the microcontroller internal clock and its high speed internal circuitry, all located too near the DAC and it's clock. In that case sometimes taking a piece of PCB material with no holes in it and soldering it vertically on its edge to the ground plane like a little wall between the microcontroller and the DAC and it's clock might help some.
Last edited:
Okay, understood. Would an AA battery holder with short length leads and a LifeP04 battery close to AA size be within present capabilities? Offhand, doesn't seem worse than a module to hook up. How about trying that for AVCC with your favorite caps and see if you hear anything different than whatever else you have to compare with. Should be at least the original circuit that came with the board. If you don't hear any difference, then maybe you are done. Maybe just keep what you have if you don't hear any difference or you don't hear something you like much better. If you do hear a difference and you want something closer to what the battery can do, then lets think more about how to make that happen. In worst case maybe I could modify a module for you. But, lets not worry about that yet. Just find out if there is a difference and if so how big. Does that sound like something you might be willing to try?
Not that ultra-low noise for AVCC is it. Certainly not down at the -120dB level. http://www.ti.com/lit/ds/symlink/lp5907.pdf
Last edited:
Yes you right for this purpose it is not the best. It can be used for the power supply for the different parts clock spdif etc. as a reasonable cost solution. For the -120db something special needed.
Just sayin' (as before), AVCC is common mode to both DAC OP phases so there is cancellation in the I-V. An AVCC that is actually quite noisy can well have a lot of that noise cancelled in the I-V. It will depend how well balanced the I-V is. Hint - this may turn out to be important.
T
So far, we are assuming that AVCC noise is additive and not multiplicative with audio signal. That is, noise going in comes out as simple noise, and that the noise does not become intermodulated with signal. That latter would be worse, IMHO. And it may turn out to be the case, or partially the case.
Agreed, it probably helps to use tight tolerance resistors for IV and differential output stages, but I noticed ESS never went a step further and trimmed them to match even more exactly. Depending on what exactly distortion compensation does, perhaps it may be expected to compensate for some mismatches there.
ESS was also quite clear about AVCC needing very high quality power. If noise cancels out, poor AVCC regulation could be viewed as just another noise and also cancel. IF so, then why did they strongly recommend a particular AVCC circuit and particular opamps that were qualified by them to work with it and provide necessary performance?
I suspect it turns out that differential outputs can only compensate for so much and poor quality AVCC is not one of them, including poor noise quality. Or, at least differential outputs can only help so much with AVCC power quality, and the quality needs are extremely high requiring both cancellation and high quality power input.
For the record and for whatever its worth, I used .1% resistors for IV stages and .01% for differential stage.
Agreed, it probably helps to use tight tolerance resistors for IV and differential output stages, but I noticed ESS never went a step further and trimmed them to match even more exactly. Depending on what exactly distortion compensation does, perhaps it may be expected to compensate for some mismatches there.
ESS was also quite clear about AVCC needing very high quality power. If noise cancels out, poor AVCC regulation could be viewed as just another noise and also cancel. IF so, then why did they strongly recommend a particular AVCC circuit and particular opamps that were qualified by them to work with it and provide necessary performance?
I suspect it turns out that differential outputs can only compensate for so much and poor quality AVCC is not one of them, including poor noise quality. Or, at least differential outputs can only help so much with AVCC power quality, and the quality needs are extremely high requiring both cancellation and high quality power input.
For the record and for whatever its worth, I used .1% resistors for IV stages and .01% for differential stage.
Last edited:
Step by step fo find the nirvana
What are your OPA ?
Serge
Currently opa2134
Regarding opamps, please see pages 4 and 5 for information on component selection and layout, including for opamps: http://www.esstech.com/files/4514/4095/4306/Application_Note_Component_Selection_and_PCB_Layout.pdf
1. VCCA 3.3v (at two different pins)
2. DVDD (internal regulator, only needs decoupling)
3. AVCC_L (needs ultra-high quality power per ESS application note)
4. AVCC_R (needs ultra-high quality power per ESS application note)
Attachments
Last edited:
Yes. Unfortunately it is not easy. Effectively, you have to rewrite some of the microcontroller code.
Last edited:
Right. I missed DVCC, thank you. It can be +1.8v to +3.3v. The difference between DVCC and AVCC is that DVCC powers digital circuitry, whereas AVCC powers analog for osc.
Regarding DVDD, 1.2v nominal, although there is a an external regulator on some of the boards, Q2M DS says there is an internal regulator. Don't know why they put a regulator on some boards.
The old ES9018 may have needed it.
Last edited:
To bad. Just as I thought.
An externally hosted image should be here but it was not working when we last tested it.
I use ChromeCast Audio for streaming. Then DSD converter board. Then es9038q2m board. Board is powered by Salases. Separate 5V for digital and 5V analog. Output stage is twistedpearaudio Legato 3.1, powerd by Salases -/+15V.
For DVDD I use TPS7A4700 3.3V, for AVDD R/L Trident v1.1 2x.3.4V shunt from TwistedPearAudio. There is no 100Mhz onboard. MLCK is taken directly from SRC board. In between the boards there is Fifo reclocker from DiyInHk. Then Modulus-86 amp from neurochorme audio.
It is actually one of the best sounding boards I had among many that I build e.g.: pcm1704, ak4497, soekris, ...)
The only thing is missing is synchronouse clocking. DPLL could be turned off in such setup if I only knew how ...
Turning off DPLL is doable, I could tell you exactly how if you are up for doing some H/W mods. And I could give you some Arduino code to do it. But, we would need to be clear on some more things first. Do you want the DAC to run as master or slave? Can you supply I2S audio 8x oversampled?
Also, maybe worth mentioning: Whenever people want to turn of DPLL because they are looking for some improvement in sound quality, I have to believe they are probably looking in the wrong place. Don't know if you are familiar with Benchmark DAC-3. If not, there is a good review at Stereophile. It is also on their recommended equipment list as an A+ DAC. In addition it is used in mastering rooms and recording studios to decide how to make records sound. It is generally regarded as a SOA DAC. Here is the thing, it uses an ES9028 with DPLL and ASRC turned on. However, they do some other interesting things to make the DAC much better than most. There are some semi-technical application note articles at their website explaining a lot of how they do it. If you want to look just search for Benchmark DAC-3. Anyway, if a DAC like that can perform as well as it does with ASRC and DPLL still turned on, then I have to question why people keep wanting to turn it off. Makes me think there must be some other problem.
Also, maybe worth mentioning: Whenever people want to turn of DPLL because they are looking for some improvement in sound quality, I have to believe they are probably looking in the wrong place. Don't know if you are familiar with Benchmark DAC-3. If not, there is a good review at Stereophile. It is also on their recommended equipment list as an A+ DAC. In addition it is used in mastering rooms and recording studios to decide how to make records sound. It is generally regarded as a SOA DAC. Here is the thing, it uses an ES9028 with DPLL and ASRC turned on. However, they do some other interesting things to make the DAC much better than most. There are some semi-technical application note articles at their website explaining a lot of how they do it. If you want to look just search for Benchmark DAC-3. Anyway, if a DAC like that can perform as well as it does with ASRC and DPLL still turned on, then I have to question why people keep wanting to turn it off. Makes me think there must be some other problem.
Thanks Mark!
The oversampling SRC board I have can do DSD 11.2. I suppose it will be slave mode with DPLL turned off and MLCK delivered from SRC board clocks. The reason I want to turn off DPLL is very simple: experiment
This is DIY. If I just wanted to sit down and listen I would propably bought Benchmark DAC-3 and some decent amp , but this is not the case in here ...
Last edited: