That's weird if CM6635 is already stopped production. I was contacted to Cmedia about CM6642 problem(the IC has no CRC validation and a bad connection of USB creates loud white-noise) and also asked about CM6635, and always had recommendations do not use CM6635 so far because it has some problems. They said - try to back half of the year later about CM6635. From 2018 I asked a couple of times and gave up Cmedia's chips completely. And now I see CM6635 is obsolete.
No, the CMEDIA web site doesn’t list the 6635 as obsolete, and there’s no newer equivalent anyway.
CMEDIA acts as an Asus division, I am not aware of any other major customer they may have. I hope the Asus business will sink and CMEDIA will then rot in hell. Maybe they’ll understand the price of putting all the eggs in the same basket, and of their total lack of support and cooperation with the ROW.
CMEDIA acts as an Asus division, I am not aware of any other major customer they may have. I hope the Asus business will sink and CMEDIA will then rot in hell. Maybe they’ll understand the price of putting all the eggs in the same basket, and of their total lack of support and cooperation with the ROW.
syn08 are you really so nervous guy or you kidding me? A lot of morons in such tiny Asian companies(realtek an excellent example, actually, the whole planet is stupid), even I, being in that center of chaos, not so pessimistic and evil. Take it easy and watch yourself, you are a lucky guy and living in a comfortable city 😉
I can’t stand this type of companies that are advertising products impossible to get to, unless you buy 1M, and hiding any trace of real information from whom? Their competitors can always get whatever information they want, it’s only the small guys that get hurt.
I’ll be more tolerant if and when I’ll understand their sales and support policies. I’m not holding my breath waiting.
I’ll be more tolerant if and when I’ll understand their sales and support policies. I’m not holding my breath waiting.
CMedia has a large business in low end USB audio interfaces. I have met with them and they have been nice to me in the past. We were talking about the premium USB interface and when I mentioned 10,000 units they almost fell off the chair. Most high end audio companies buy 500 or 1000 ever.
1) they are too small
2) I guess, they often buying a third party design to resale as cmedia, comtrue etc(some programing tools for cm6642 used realtek DLL)
4) no time to processing all requests of help from everyone. Pretty much no support department at all. For example, when I asked cmedia about cm6642 white-noise issue, their project manager replied to me. I'm also a tiny customer, for instance, I bough just 6000pcs of CT7601, far not 1M, right?
2) I guess, they often buying a third party design to resale as cmedia, comtrue etc(some programing tools for cm6642 used realtek DLL)
4) no time to processing all requests of help from everyone. Pretty much no support department at all. For example, when I asked cmedia about cm6642 white-noise issue, their project manager replied to me. I'm also a tiny customer, for instance, I bough just 6000pcs of CT7601, far not 1M, right?
1audio, cm6642 is exactly what you mentioned USB UAC2 + DAC+HPA and ADC but..
nobody will seriously use that chip if it makes white-noise explosions if you shake the USB cable a little. The same time, extremely cheap mainland Chinese USB headset solutions came to the market, so, cmedia did lose there or will lose soon.
nobody will seriously use that chip if it makes white-noise explosions if you shake the USB cable a little. The same time, extremely cheap mainland Chinese USB headset solutions came to the market, so, cmedia did lose there or will lose soon.
I was talking to them recently about a headphone application. Same for ESS. Neither were that well suited. XMOS is a power hog. ESS wants you to use there solution from end to end. And I found where Apple's arrogance went to die. However at some point I may need to add MQA which brings a lot of stuff I do not want to deal with.
I didn't buy PR only CR/SR but they have all models as I know. Try to ask Aliexpress shop if it is the same people what behind Taobao shop, they are a distributor company from SHZ.
So the CM6631A is a obsolete part?
I was just using it in a dotnet Windows application.
Alternatives?
I was just using it in a dotnet Windows application.
Alternatives?
Back to the ES9822. The AVCCL and AVCCR power supplies need to be very low noise for the best performance.
I've tried the ADP7118 and the regular, over the top standard, the LT3045.
Both give you the same dynamic range and, as far as I can see, distortion performance. But the increased noise of the ADP7118 (with noise lowering circuit in my case) increases the noise in the THD+N calculation when working at high signal levels.
Without the noise lowering circuit all I could get was 0.0005% THD+N, obviously being dominated by N. With the noise lowering circuit that fell down to 0.00018% and then with the LT3045 0.00011% THD+N.
0.00018% THD+N is still -117.44dB, which is within an acceptable degree of error, in my opinion, for the ES9822s actual specification of -118dB for mono mode.
Nevertheless 0.00011% is still better. -119.54dB for the LT3045. These are all A weighted.
I don't have access to any sophisticated test equipment as IVX does. I'm only working with a mono AK4499 as a signal source. With a proper low distortion/low noise source I'm sure I'd get better than the 0.00011% THD+N as my AK4499 outputs slightly higher distortion than it is capable of when driven close to 0dBfs. Nevertheless the point of this post was to highlight that you cannot skimp on the main analogue power supplies if you want the most out of the ES9822.
I've tried the ADP7118 and the regular, over the top standard, the LT3045.
Both give you the same dynamic range and, as far as I can see, distortion performance. But the increased noise of the ADP7118 (with noise lowering circuit in my case) increases the noise in the THD+N calculation when working at high signal levels.
Without the noise lowering circuit all I could get was 0.0005% THD+N, obviously being dominated by N. With the noise lowering circuit that fell down to 0.00018% and then with the LT3045 0.00011% THD+N.
0.00018% THD+N is still -117.44dB, which is within an acceptable degree of error, in my opinion, for the ES9822s actual specification of -118dB for mono mode.
Nevertheless 0.00011% is still better. -119.54dB for the LT3045. These are all A weighted.
I don't have access to any sophisticated test equipment as IVX does. I'm only working with a mono AK4499 as a signal source. With a proper low distortion/low noise source I'm sure I'd get better than the 0.00011% THD+N as my AK4499 outputs slightly higher distortion than it is capable of when driven close to 0dBfs. Nevertheless the point of this post was to highlight that you cannot skimp on the main analogue power supplies if you want the most out of the ES9822.
5th element, Cosmos ADC reads my Cosmos DAC proto (with actual THD+N -123-124db) as THD+N -121db in mono mode, and -118-119db in stereo. Cosmos ADC reading THD+N -123db in mono mode if Cosmos DAC has turned On its 1kHz LPF(actual THD+N -130db). The actual THD+N is obtained with Cosmos APU notch filter. All measurements were performed at 10Vrms 1kHz, the unit of Cosmos ADC has DR 125/128db(A-weighted, but the THD+N measurements are not weighted).
Last edited:
Are you using a large gain in the I/V stages, how did you get 10Vrms out of the DAC (that's 14Vpp differential/balanced)? Usually the output level is set to some 4Vrms.
You must get a lot of IMD ESS hump from the 9038 DAC at those insane levels, the hump is never completely removed, all we can do is to keep it buried in noise, that's why countermeasures (like lowering the I/V stage gain) come with some SNR penalty.
You must get a lot of IMD ESS hump from the 9038 DAC at those insane levels, the hump is never completely removed, all we can do is to keep it buried in noise, that's why countermeasures (like lowering the I/V stage gain) come with some SNR penalty.
My AP SYS2522 has no analog IMD option so I used AP's digital analyzer with -100db residual IMD, this way Cosmos DAC IMD is near to -100db as well.
I did try to use Cosmos DAC - > Cosmos ADC - >ARTA(ARTA is one of 2 FFT apps I know calculating THD+ and noise correctly, I hope IMD as well), and I see -116db at -10dbfs, -107db at -.5dbfs SMPTE 4:1 i.e. 2 times less than your figure, however, I see the hump now. I'm not sure if the hump is the result of DA or AD conversion.
