ES9038Q2M Board

Hi, im new with brzhifi bt-30 dac 9038. When i measure it, freq response is drop too much at 18-20khz, i tried with 5532, lm4562, opa1612 opamp at i/v stage but problem still not gone. So could i modify to fix it
478CE1C6-1AFB-4828-A139-9F58F866C27F.png

A8C87AF0-292D-42E5-B988-EAFB6231C99B.jpeg
 
Last edited:
Hello trungdtmc,

Pearhaps do you change the FIR in your settings (if available) ->

FIR.jpg


Otherwise, it is a too low cutoff frequency of the low pass filter.

In this case, it is necessary to determine these constituent elements and recalculate new values for the components (resistors and capacitors).

These components will be in SMD format and it is not easy to replace them without having good soldering skills.

It often looks like this ->

LP.jpg


Regards.
 
  • Like
Reactions: 1 user
Hello trungdtmc,

Pearhaps do you change the FIR in your settings (if available) ->

View attachment 1131533

Otherwise, it is a too low cutoff frequency of the low pass filter.

In this case, it is necessary to determine these constituent elements and recalculate new values for the components (resistors and capacitors).

These components will be in SMD format and it is not easy to replace them without having good soldering skills.

It often looks like this ->

View attachment 1131536

Regards.
Thanks for the informative reply, this DAC can only change the volume through the encoder and the ST ic on the left of my picture. I don't think I can interfere with the code of this ic, but according to ess's datasheet, the filters all cut at frequencies above 20khz, is there anything in the circuit that affects the cutoff frequency? As for soldering skills, I think I can do it well.
 
Last edited:
  • Like
Reactions: 1 user
Hello.

I think I know where your problem comes from...

First of all, here is, in my opinion, the diagram that is closest to that of your DAC ->

diagram.png


It calls for the use of a 'Vref' voltage for each non-inverting input (according to the example diagram) of your double OPAMP of the I/V stage ('input stage' according to Markw4).

As you say, the problem seems to come from this part of the analog circuit of your DAC and knowing very well the habits of certain Chinese assemblers with their principle of economy of scale: check the value of your two WIMA capacitors (red polypropylene) whose value must be too low so that the cutoff of your lowpass filter starts too soon.

Indeed, the higher the value of the WIMA capacitors, the more expensive they are and the notion of precision of the constituent elements of a diagram for its correct functioning seems somewhat 'unimportant' for the assemblers of such affordable products provided that the value of the component is quite close to what was initially expected by the circuit designer.

If the case is verified, you can use other capacitors with the just higher value proposed by WIMA and this will solve your problem.

You will of course check by calculation the new frequency at -3db cutoff of the filter thus modified.

Regards.
 
  • Like
Reactions: 1 user
Hi trungdtmc,

Can you put back your diagram/statement of values because it has disappeared?

This will verify.

No: it is necessary to put capacitors with a lower value if it is indeed a low-pass filter (it cannot be otherwise).

In other words, if your resistors are 1.5KΩ then your capacitor (WIMA FKS2) will have to be 3300pF/100V/5% ( MOUSER: 505-FKS23300/100/5) for FC @ -3 dB ~ 32,168 Hz because a value of 4700pF/100V would give FC @ -3dB too limited ~ 22,586Hz which would limit the high frequency response if we did a response curve simulation with SPICE, for example, which would visually confirm what I'm saying.

2200 pF is too high with an FC @ -3dB ~ 48000 Hz which is too far from the Nyquist frequency ->

Ny.jpg


if your resistors are in series i.e. 1890 Ohms with the new 3300pF capacitor, then FC @ -3 dB will be ~25500 Hz which should be correct ->

A.jpg


PS: It is likely that he used these original capacitors in your DAC to limit a certain 'insistence' in the treble with certain assemblies with the ES9038 Q2M chip...
 
Last edited:
It is still surprising to use a low-pass filter, since even the 2 FIR settings (SLOW ROLL OFF MINIMUM or LINEAR PHASE) have a maximum response curve at 32kHz to minimum - 80 dB (!!!) knowing that they are necessarily upstream of the I/V conversion OPAMP since they are integrated into the DAC ->

ES9038 Q2M.jpg
What do you think about Markw4 ?
 
The digital filters inside the dac are for interpolation of upsampled data points. Part of how an oversampling dac works. Some of the internal filters can start to roll off a bit below 20kHz as can be roughly seen from the graphs. IME the default filter usually sounds best, if it doesn't then there is usually something else wrong in the hardware that might be worth fixing.

Also, even with digital filtering there is remaining noise from the modulator, and or from other possible sources (AKM might point to substrate coupled noise as one possible source) that still needs to be filtered in the analog domain.
 
Last edited:
  • Like
Reactions: 1 user
C15,C19,C16 should be C0G/NPO, if they are going to be ceramics. C19,C15 should be the same value. If you want, C17 could be changed to something more like 100uf Nichicon UKZ bypassed with .01uf WIMA FKP2 (then give it a few days of running to settle in).

The biasing arrangement shown in the schematic with R26,R28 was to provide a virtual ground in case the dac was operated from a single +15v power supply. Then the -15v rail would be jumpered to ground. Probably better if that node is connected directly to ground if +-15v rails are used.

5532 opamps show in the schematic were cheap, but not so good for this circuit. OPA1612 are generally recommended for I/V. However, the output stage circuit shown in the schematic isn't really I/V, rather it is operating the dac chip in voltage output mode (which makes HD considerably worse than if proper I/V had been used).
 
Last edited:
  • Like
Reactions: 1 users
The schematic has a lot of errors but the most important thing is that Markw4 is right.

To complete his intervention I think that the analog output diagram should be reviewed, the ideal would be to use the DAC as a current output friends this would require adding two double OPAMPS in the box with the difficulty of being able to integrate them in order to end up with a traditional diagram and much more efficient than the one that exists in your DAC.
Buying a BRZHIFI DA10 DAC for an additional €40 would have been better (at least in terms of final cost price) than trying to modify it in my opinion.
 
C15,C19,C16 should be C0G/NPO, if they are going to be ceramics. C19,C15 should be the same value. If you want, C17 could be changed to something more like 100uf Nichicon UKZ bypassed with .01uf WIMA FKP2 (then give it a few days of running to settle in).

The biasing arrangement shown in the schematic with R26,R28 was to provide a virtual ground in case the dac was operated from a single +15v power supply. Then the -15v rail would be jumpered to ground. Probably better if that node is connected directly to ground if +-15v rails are used.

5532 opamps show in the schematic were cheap, but not so good for this circuit. OPA1612 are generally recommended for I/V. However, the output stage circuit shown in the schematic isn't really I/V, rather it is operating the dac chip in voltage output mode (which makes HD considerably worse than if proper I/V had been used).
Thanks. So circuit of my DAC seem no problem right?