The funny part in this statement is that exactly (yes, EXACTLY) this method is used by ESS and AKM to design both the ES and AK4499 chips. Therefore, they must be fundamentally flawed
. The idea of designing a chip by the ear is ueber funny, I have to admit.Maybe you should go read post #6049. That's with one chip, I believe.
Because it helps sell dacs. Some people think two is better than one. IME, only slightly better though.
I also have the experience to know.
You're not listening. I did fix one of these dac using extensive modifications. It reached -120dB distortion. With this very board (as I said, in very modified form).
Its already been tried. Nobody found a simple fix like you are looking for. Also, I posted Victor's solution already. That's the easiest one to get measured distortion down. However, you are welcome to find out for yourself
Hi Mark. I understand correctly that it is better to buy ak4493 and get the best sound quality? compared to es9038q2m. I modified the power supply of the dac es9038q2m using lt3042, all separately. made i / v exit. replaced the clock with crystek. converting to dsd using ak4137. the sound is much better. the next step is arduino and working with registers, so I carefully read your posts. can i get rid of es9038q2m hump completely?
Syn08,
AKM is said to have a golden ear on staff. May have been said by John Westlake. I see no reason to disbelieve it. Regarding ESS, I don't doubt you. However, they may do things more the way Bruno Putzeys does. Makes sense if one is going to make a lot of product and has the resources to investigate problems in depth.
AKM is said to have a golden ear on staff. May have been said by John Westlake. I see no reason to disbelieve it. Regarding ESS, I don't doubt you. However, they may do things more the way Bruno Putzeys does. Makes sense if one is going to make a lot of product and has the resources to investigate problems in depth.
Hi,
My opinion would be that its probably easier to make a good sounding dac with an AKM chip than with an ESS chip. That said, Allo has gotten pretty good at making ES9038Q2M dacs. Regarding the hump, there are some posts over at ASR that describe how some people have dealt with it. Its basically a hardware problem involving the output stage. The main easy gain that can be had by accessing registers on the dac you have now is that reducing DPLL bandwidth as ESS recommends, to the lowest stable setting. Its something I recommend doing. Most people who try it report improved subjective sound quality.
Last edited:
I have resources. there is a desire! I solder well. there is a tool besides the oscilloscope. Ohm's law I know) I saw how you can achieve a result in a decent sounding dac with your own hands. es9038q2m gave experience !!!) I want to hear from you - will the modified es9038q2m be worse or better than the modified ak4493? and how big is the difference or how is it manifested? please describe
that is, the only difference is that it is easier with ak than with ess?) Well then it makes no sense for me to buy ak4493, since I almost finished with the es9038q2m modification. or is it more than just "more work"?
"Is said", this sounds very <you know who>. Nevertheless, the scenario of a mask set and chip batches running through the fab, then the Golden Ear on staff steps in and rejects the design based on his subjective evaluations is, as I said, ueber funny, keep going
. One would wonder how AKM is still in business, do you have any idea of the cost of such a cycle? Or how would the chip designers and/or the layout team integrate the subjective results in a new design version?
Last edited:
Getting an AK4499 board to play with might be getting pretty difficult right now. They had a bad foundry fire, and it may take at least 6 months to get it going again. AKM parts are gone from all the distributors and nobody knows what is going to happen yet.
Regarding AK4499 finished dacs, so far some Chinese companies have made what are fairly low cost AK4499 dacs. They are good, but not great. IMO the chip could be used to make a dac rivaling some of the best, say, maybe in the $3k - $10k price range. Its just that nobody has done it yet. Also, I think there is interest in the new Rohm dac chip. Some people are waiting to see how it sounds. Maybe it would be easier to use, don't know.
Regarding AK4499 verses something like ES9038PRO, I agree with John Westlake that AKM dacs sound better, all other things being equal. That said, there are some pretty good ESS designs out there, such as Allo's.
Last edited:
Sure, its funny. But its just a story you made up. Nobody said anything specific about how the golden ear works with other personnel.
buy ak4499 no problem - Ali / ebey. and then finalize. crashing into the tracks is not a problem, datasheet and tester to help. and it's cheaper than the eval fee! I don’t understand - so is akm better than ess? Or can you explain this moment to me in some other way?
in general I understand that ak4493 will sound better than es9038q2m. I am writing all this because I am impressed with the sounding results of es9038q2m after the work done. and thought about what the purchase of ak4499 could change and its revision in terms of improving the sound quality. just the price, taking into account the value of the dollar in relation to the ruble, is already something ... or to stop there, knowing that this is a fairly high level. help me please with this question
I don't know if AK4493 will sound better than a well implemented ES9038Q2M. Never compared the two myself. AK4499 is the only AKM dac I have spent a fair amount of time with.
Regarding AK4499, I have heard it sound a lot of different ways. IME and IMO, it can sound worse than the Chinese AK4499 dacs, about the same, or much better. It all depends on implementation.
Regarding AK4499, I have heard it sound a lot of different ways. IME and IMO, it can sound worse than the Chinese AK4499 dacs, about the same, or much better. It all depends on implementation.
Last edited:
Sure, I would appreciate any ideas (even theoretical) on how a Golden Ear could impact a chip design. Meantime, obviously, you know that the AKM chips are designed with the help of a Golden Ears, isn't it?
Now waiting for you to take it back "I didn't say this" , following the usual habit of denying what is written black on blueish by your very own fingers on a keyboard.
The guy that developed it is on this forum, here is the post with some internal renders, I think it's a single chip because he's trying to keep the power consumption as low as possible: the TRUE digital one
I guess you can ask him
The spectrum in
https://www.diyaudio.com/forums/vendor-s-bazaar/326697-true-digital-8.html#post5766285
is similar to what I have myself, so it’s not a magic for a single chip, but it’s far away from 120-125 db
as in
#9038S | index
What am I missing?
Anyway, thanks a lot, I will try to contact IVX
now I'm on vacation and not at home. as soon as i get back i will put dvdd and ak4137 on clean power and be ready to work with the register. I will definitely ask you questions. it is very pleasant that you spend time and effort to help. as soon as everything works out, I'll sign off on the sound results. thank you for the answers. I was considering the option to purchase es9038q2m dac on two circuits and modify it, thinking that two circuits are better. thank you for the post where you explained that this is a marketing ploy. also thanks for the clarification regarding ak4493 and ak4499. I will wait for ak4498, I think it will be a little easier there. I would also like to try ak4191.
>es9038q2m dac on two circuits and modify it, thinking that two circuits are better. thank you for the post where you explained that this is a marketing ploy.
I don’t think it’s a marketing ploy. Mu humble theory is below.
Sabre made significant step in DACs improvement, but got a drawback – high C2/C3 harmonics they had to deal with. So invented a compensation strategy with registers 22-25 to set C2/C3 in the reverse phase to make the output clean. Obviously the register values are unique to DAC instance, otherwise it will be hidden from the users availability.
The curious thing is the default C2/C3 values are not zero. The C2/C3 presence was promoted though as a way to compensate the following “cascades”. But looks like it’s for the DAC itself.
Method to inject the harmonics is not perfect, so DAC injects C2/C3 equally into DAC+ and DAC- (symmetrical) outputs (DAC+/- pins don’t come from separate “ladders” obviously) . So the following symmetrical (balanced) stage easily eliminate C2 but doubles C3 (that is exactly what I saw during my experiments). C3 can’t be compensated completely at all as the level of C3 presented in the symmetrical DAC+/DAC- outputs is equal (in the reverse phases), so the next balanced (subtracting) stage just doubles the C3 level
The method to move forward is using 2 DACs and get DAC- and DAC+ for balancing from different DACs to make the C2/C3 approach really universal and exhaustive.