E1DA Cosmos ADC

[staticV3]

Maybe you need to glue a nut on the back (inside)?

The rear plate itself is tapped, so no need to add a nut yourself.

@danadam
FWIW, my Delock 84025 and 83717 properly and securely screw into the rear I/O plate.

 

[IVX]

Does anyone has the USB cable with screw for ADCiso? I bought one but it looks like the hole in the case is too big and the screw just passes through.

did you buy that in E1DA Aliexpress shop? The rear panel of ADC has the M2 thread(I measured that with a caliper from the inner side 1.71mm) suitable for the cable's screw(the caliper shows 1.93mm). If you can not measure these diameters, please show me pics of how the screw floats in the panel, I want to realize why it happened, and what I should send you a new panel or a new cable.

 

[danadam]

did you buy that in E1DA Aliexpress shop?

Yes, both the ADCiso and the cable were ordered there about 2 weeks ago. Here's the sticker from the package of the cable:

Here's the backplate with the screw next to it. I also recorded some videos, linked at the end.

If you can not measure these diameters, please show me pics of how the screw floats in the panel

I only have a manual caliper (admittedly not in its prime). To my eyes (also not in their prime):

• jaws closed shows 0.1 mm
• usb screw shows 1.8 mm
• screwdriver (which doesn't go through) shows 1.95 mm
• hole's inner diameter shows 1.65 mm

I made some photos of the measurements but I'm not sure if that will be helpful. They not always exactly agree with what I saw, I guess there could be some parallax problem here. I made them using iPhone XS Max and Pixel 3 (in RAW) thinking that maybe there will be some quality difference but I don't think there is. Here they are both anyway, cropped to 1200x1120:

jaws closed:


usb screw:


screwdriver:


hole's inner diameter:

 

[IVX]

hm, I'm still not sure if there is too small the screw or too big hole in the panel? Can you try any standard M2 screw with the panel?

 

[danadam]

Can you try any standard M2 screw with the panel?

My local kind-of Home Depot has only M3 and bigger. I could order online but the delivery costs for the ones I've found are about the same as the price of a new cable. So I ordered the Delock mentioned above. There's slight chance it will be here tomorrow, if not then next week.

 

[danadam]

Can you try any standard M2 screw with the panel?

So I ordered the Delock mentioned above.

I got 84028 but similar results.
I thought that maybe I can position the panel so that threads "catch" each other on one side but no luck. I can always basically pull out the plug without unscrewing it.

 

[IVX]

if the problem is the panel, please contact Sunny from our Aliexpress and she'll send you a new panel.

 

[Michael1]

I have what is probably quite a basic question regarding the Cosmos ADC, but I am struggling to find an answer that I understand amongst the many pages of information scattered across the internet.

So, the factory THD+N of the ES9822PRO is -117dB in stereo mode. The Cosmos ADC is -123dB for a grade A unit, if I am interpreting the spec sheet correctly?

Is this improvement over the factory values due to manually configuring the THD compensation registers as described on page 9 of the datasheet?

Is my understanding correct that this calibrated value is valid only at the specific frequency of 1KHz, 0.5dBFS and 27 degrees, and that outside this range the ADC will perform according to the factory-level performance figures?

 

[IVX]

There are 2 reasons:
1) each Cosmos ADC is calibrated i.e. H/H3 harmonics are minimized individually.
2) ESS used APx555 gen to test that chip, 555 is known as a too noisy analyzer to measure modern DAC/ADC.
BTW, Cirrus Logic offered some LNA circuit to test with APx555 and get the specified noise level for CS43131/43198 $3 DACs.

 

[Michael1]

There are 2 reasons:
1) each Cosmos ADC is calibrated i.e. H/H3 harmonics are minimized individually.
2) ESS used APx555 gen to test that chip, 555 is known as a too noisy analyzer to measure modern DAC/ADC.
BTW, Cirrus Logic offered some LNA circuit to test with APx555 and get the specified noise level for CS43131/43198 $3 DACs.

Ah, that's very interesting. And to clarify, the calibration is specific for 1KHz, i.e. that method of calibration does not calibrate THD at all frequencies simultaneously?

Regarding point number 2, are you talking about the generation of the 1KHz sine wave itself? Would the solution to be to use something like the Victor oscillator?

 

[IVX]

no, Victor's OSC(good for THD but be aware that the OSC's H3 is rising quickly from >4V), is even noisier and bad for THD+N. The option with less noise vs APx555 is my new THD stellar 9039S, some samples have THD -145-149db i.e. all harmonics are surely lower than -150db@1kHz. 10kHz H3 -135db is also perfect. The golden sinegen would be a decent DAC+LPF tuned at 1000Hz to cut all harmonics and noise.

 

[Michael1]

no, Victor's OSC(good for THD but be aware that the OSC's H3 is rising quickly from >4V), is even noisier and bad for THD+N. The option with less noise vs APx555 is my new THD stellar 9039S, some samples have THD -145-149db i.e. all harmonics are surely lower than -150db@1kHz. 10kHz H3 -135db is also perfect. The golden sinegen would be a decent DAC+LPF tuned at 1000Hz to cut all harmonics and noise.

That's insane... I guess that is using the same technique for THD calibration as used in the ADC?

 

[IVX]

Yes, 9039Q2M has separated R and L THD compensation, but you know what, all these coefficients are nearly zero.. In my mind, ESS has reached something revolutionary with that IC! A huge progress.

 

[LKA]

"no, Victor's OSC(good for THD but be aware that the OSC's H3 is rising quickly from >4V)"
what V ? only Vpp applies to 4
Output voltage without load 300mV - 2.7V (RMS)

 

[IVX]

as I know, does exists a balanced OSC with doubled voltage. You can check that discussion, I performed careful tests for about 20pcs APUs to prove that it is Victor OSC rising H3 at >4Vrms.

Cosmos APU a notch+LNA $70 to outperform APx555b for $30,000

Here is another report about high H3 in the notch section: Below is the notch section at 0dB position with 450mV, 1kHz input for 4.5Vrms output (the notch is at 10kHz). When the notch select is switched to the 1kHz position, the 1kHz fundamental is attenuated by 30dB, but the 3rd HD is...

www.diyaudio.com

 

[eeMGee]

"no, Victor's OSC(good for THD but be aware that the OSC's H3 is rising quickly from >4V)"
what V ? only Vpp applies to 4
Output voltage without load 300mV - 2.7V (RMS)

Balanced version takes negative signal from pin 7 of opa (see pic) - todays version use OPA1656 and 100 Ohm output resistance.
Than you have 2 x 2,7 V (5,4 V) on the output of oscillator.

 

[danadam]

if the problem is the panel, please contact Sunny from our Aliexpress and she'll send you a new panel.

Got the new panel and locking fits well now. Thank you.

 

[IVX]

Good to hear that! Sorry for the inconvenience.

 

[VIRTINS]

Did anybody encounter problem with Cosmos ADCiso at a sampling rate of 705.6kHz? No problems have been found at other sampling rates including 384kHz and 768kHz. Comtrue ASIO driver was used.

Sampling rate: 384kHz

Sampling rate: 705.6kHz

Sampling rate: 768kHz

 

[IVX]

Did anybody encounter problem with Cosmos ADCiso at a sampling rate of 705.6kHz? No problems have been found at other sampling rates including 384kHz and 768kHz. Comtrue ASIO driver was used.

Sampling rate: 384kHz
View attachment 385156

Sampling rate: 705.6kHz
View attachment 385155

Sampling rate: 768kHz
View attachment 385157

as I remember I have deleted 705k mode in one of earlier FW, and it is not supported. Next 3 days I'll back home and check that again.