China ES9018PRO ES9028PRO 9038PRO mods&upgrades

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
No it's always with the lt3042 for one avcc, I've changed the i/v opa from ne5532 to op1612 it's much better ( I've also added again the 100pf caps), I've not yet done the mesure with khadas tone but the es9038pro board seems to be far better now.
 

Attachments

  • es9038pro_lt3042_op1612.png
    es9038pro_lt3042_op1612.png
    28.8 KB · Views: 462
No it's always with the lt3042 for one avcc, I've changed the i/v opa from ne5532 to op1612 it's much better ( I've also added again the 100pf caps), I've not yet done the mesure with khadas tone but the es9038pro board seems to be far better now.

I added lme49720 49710.
But the advantage can be heared if the power is good. I read it somewhere and also experienced.
I dont know.what is the relation between the 1612 and 49720.
 
Hi,
I've started to use the lt3042/49720/ lm49600 as avcc source. For now, it's not good (maybe the wires are a bit long), I will try again with shorter wire, and more integrated circuits. I've also removed all the 47uf caps around opamps and replaced with 1uf metalized polyester film, it seems to help.
 
Is the 3042 the reference voltage only - as you mentioned?

I haven't removed all the 47uF - this can be a good idea - but I feel somehow much risky to do modifications on this board than the q2m was. (definitely the price is much higher - just to destroy) And anyway the details now are on a level which are acceptable - at least for a while :)

I think the next step should be to change the I/V, but currently no idea yet which would be the best idea to handle the 60mA on the best way.

Were you abel to sort out the issue meanwhile?

If you are ready wiht the AVCC, I would suggest to check the power supply for the OPAs. For me as I have written the Salas reg brought a big diff. As I remember Mikett tried the super reg with good result too.
 
routing >5cm long wires over HF circuit perhaps is not the best solution regardless of silver coating... put the reg PCB underneath inverted and route 30-28AWG isolated wire through the elcap holes. 3.3V/50mA Nazars shunts for each channel with 22uF X7R on pins. take care of cooling though.
if you consider use a scope (>=100MHz) for validation of your power solutions, please remember that without correct measurement methodology this might be useless. a correct probe, something like this A wideband 1:21 1 kΩ DIY oscilloscope probe is pretty useful

I just realized that with the q2m board I did what you offered. The AVCC ref was on the back side of the board with direct connection- i will consider how to realize here :)

For me the x7r didn't work as I wrote in the q2m thread, I prefer the voice without.
 
Hi is there anyone, who replaced the crystal to a low phase one?

I mean i'm interested in which low phase one can be built in easiest (which is not very costly). How to solder out the current one?

Have you applied special power supply (eg lt3042? or simillar) what is the experience with sound improvement?
Thanks,

Sz.
 
Best cheap clock for a Sabre dac might be an 80MHz NDK SD series, possibly powered from a local super regulator. Diyinhk in Hong Kong sells the clocks if I recall correctly.

For expensive clocks, a 100MHz Crystek is usually the preferred choice for the default asynchronous mode operation.
 
MArk, thanks for the suggestion, I will check them out. Crystek produces a lot of crystal which version is the good? (Outpout level,etc.

Any suggestion how should I solder the current one out? Is there any special to know? To be honest I never did it before.

Thanks again,

Sz.
 
A good clock for (asynchronous mode) ESS dacs is: https://www.crystek.com/crystal/spec-sheets/clock/CCHD-575.pdf
The 100Mhz one is the one to get.

It may be hard to remove the old one if it is one of the big ones in a big metal can. If you had a Hakko desoldering tool it might help. Otherwise, one might try Getting some Chip Quik desoldering material: https://www.amazon.com/ChipQuik-SMD1-Leaded-Temperature-Removal/dp/B07XD9VFSZ ...that stuff works amazingly well.

Or, maybe get two people with two solder irons to heat up all the pins at once so it can be gotten out or at least lifted off the board enough to cut the pins under the can. Then the pins could be unsoldered one at a time.

Last method offhand would be to destroy the clock case from the top, until it is in four pieces, one piece still attached to each pin. Then remove one piece at a time. A Dremel tool would be good for that approach. In the old days I probably would have used some diagonal cutters to cut apart the clock case.
 
Last edited:
It can be risky cutting up some components using surface cutters, with the chance of stressing or lifting a trace.

Radio Shack solder, awful stuff for audio, seems to work great to flood the pins, allowing the chip to be removed in the same manner as the chip removal variety.
I would look for some low temp, (cheap) solder, now, since there’s no more radio shack.
 
I've apply few mods of this thread on my ES9038Pro board ( Thank You ;) )

- Removed 47uF Fake Silmic II caps, and replaced with Wima MKS 10uF
- Remplaced the two LT1963 3.3 regulators with Dual LT3042 board (33K resistor for 3.3V & 22uF Tantalum cap for CSet)
- Aluminium & copper shield beetween transformers & DAC Board
- IEC filter and safty earth connection

Huge improvement in term of resolution and spacialisation




I've removed the two 47uF caps on the DAC Board, just after the new regulation board.
Make a real difference ! More relax sound ... Less is More

 
Have you tested this board

I tested one pretty similar in layout, only it had a switch mode regulator between the filter caps. The one you linked too looks like it uses linear regulators, but it matters a lot for such dacs what the bottom side of the board looks like. Usually, the boards need a lot of modding to sound as good as the dac chip is capable of. If the bottom side of the board has some unbroken ground plane where a bit more circuitry can be added, then modding becomes more practical. Otherwise, as is out of of the box its questionable if is sound quality is worth the price of the board, it depends what the user will find to be satisfactory. A little more for Khadas Tone Board should give considerably better sound, but Khadas is not so good for modding.
 
The dac chips are generally real. What's fake are the capacitor brand names and things like that. Every component on the board is as cheap as possible, which means the SMD caps and resistors are junk, the brand name capacitors are cheap clones, the clock is cheap, etc. Usually, an improvement in sound quality can obtained by replacing every part on the board except the dac chip with better quality parts. That's only part of the problem though, the circuitry around the dac chip is also the cheapest possible topology. That means modding the board to use improved circuit typologies also tends to improve sound quality. In short, you don't get something for nothing when buying cheap Chinese electronics. If you like the sound quality as is, and some people do, then fine. If you don't like it then you can always start a diy project to improve it.
 
I used LT1763 with ES9038Q2M, worked fine.

Also, AD797 sounds better for AVCC use than LDOs I have tried. However, it needs sufficient local electrolytic energy storage on the rails not too far away from the opamp in order to produce best bass. Have to be careful using AD797 with too low ESR caps and or too close to the IC (there is a mention in the app section of the data sheet about bypass caps). It needs a little ESR for stability, but bass reproduction doesn't need low ESR at high frequencies anyway so standard aluminum electrolytics worked okay for me. Another thing to keep in mind is that AD797 can't output enough current to power ES9038PRO AVCC loads. Maybe work okay for ES9028PRO, fine for ES9038Q2M.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.