• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

E1DA Cosmos ADC

The Cosmos ADC was intended as a tool for E1DA customers to calibrate their 9038S and 9038D units, which have an output impedance of .1/.07Ω, and so input buffers would've only reduced the dynamic range.

The input impedance is clearly stated on the website and the AliExpress shop, so it should be easy to decide whether the Cosmos ADC is the right tool for you.

A future Cosmos ADC Pro with buffered inputs and gain control via UAC2 is planned.
And, hopefully, an SPDIF or AES/EBU output ?
 
@trl When you want to know if the ODAC has issues with the low load simply use the Motu and put a 1k resistor in parallel to (both) outputs and see what the Motu shows.
ODAC_v1_MotuM4_1kHz_resistor_.png

Motu M4 fed by ODAC v1 with 1 kOhm paralleled resistor

Thanks Frans, it's like expected actually, some DACs simply just can't drive lower impendances loads.
 
ODAC and Objective2 are the first interesting DIY audio projects in the world that proved audio transparency baked up by measurements done with AP equipment and these audio devices started a hole new generation of objectivists audiophiles! Based on AP's measurements made by its designer, ODAC is a transparent-enough DAC; in an A/B test, with FLAC songs, I can't tell which one is ODAC and which one is D90.

Ivan, don't get me wrong here, I DO LOVE COSMOS ADC and I consider it's the most affordable audio measuring equipment in the world, but given that AP, dScope and Quantasylum devices are all having a default input impedance of 100 kOhms what do you think of adding a switchable input impedance on a future Cosmos ADC, like dScope Series III has: "100kR, 600R or 150/200R (jumper)"?

However, looking for others to see if they encounter impedance issues, hope I'll be the only one having issues here. :)

Thank you for understanding!
I measured topping dx3pro+.

Headphone out is a good result, but rca out increases noise and distortion as you say.
Changing the cosmos adc input to 3.5v, 4.5v, 6.7v does not disappear.
So I thought. I will use topping l30 for output buffer.

I hope Ivan will make a scaler with a low output impedance operated by a battery... It will be a color match.
 
Last edited:
View attachment 163592
Motu M4 fed by ODAC v1 with 1 kOhm paralleled resistor

Thanks Frans, it's like expected actually, some DACs simply just can't drive lower impendances loads.

You could even check how that DAC would react to the 2.5k load from the Topping L30 when you really have nothing else to do.:)
 
ODAC rev. A, not rev. B, so 10 kOhms input impedance would be needed. DAC chip is ES9023.
 
Since ESS released (what they call) a datasheet for the ADC chip, I thought it might be interesting to probe the I2C bus to see what settings are applied, and now I'm a little confused. I'm guessing the ES9822 and ES9842 are actually (almost) the same chip with the former operating the four ADCs pairwise in parallel. That would explain the writes to registers that are "reserved" or not mentioned in the 9822 datasheet but documented as controlling ADC3/4 parameters for the 9842. So far, so good. What surprised me was that when setting the filter, the firmware applies the ADC1 setting to all four channels, ignoring the other three values. @IVX, is this intentional or a bug?
 
Could it be that HpW isn't using the correct driver, and so Windows is using software volume instead?

OK, one more about Mono/Stereo switching:

1. start tweak-1 app set to stereo & terminate tweak-1 app
2. play L&R with different tone and record = stereo = OK
3. start tweak-1 app (no connect) play as 2) = OK
4. connect using tweak-1 app, shows really MONO Mode
5) repeat 2) and it is STEREO, so tweak-1 app is false
6) tweak-1 app to STEREO and back again to MONO
7) play L&R with different tone now its MONO :rolleyes:

in other words the tweak app is not WYSIWYG :facepalm:

>> A future Cosmos ADC Pro with buffered inputs and gain control via UAC2 is planned.

Yes, so change & improve crappy oscillators too, different powering, USB isolation and may a SPDIF out (limits to 192kHz SR)
 
HpW, I sent you PM a week ago where was explained MONO/STEREO switch behavior:
The slider in Cosmos tweak app is only for MONO/STEREO switching. To send the command to switch the mode need to move the slider. At the same time, USB host(I guess Windows) sending UAC2 volume commands as well, so ADC responses for the last volume command received.
 
  • Like
Reactions: Hee
HpW, I sent you PM a week ago where was explained MONO/STEREO switch behavior:
The slider in Cosmos tweak app is only for MONO/STEREO switching. To send the command to switch the mode need to move the slider. At the same time, USB host(I guess Windows) sending UAC2 volume commands as well, so ADC responses for the last volume command received.
Posted now here, while IMHO of general interests to switch back to MONO while the gear is on STEREO!

Action to take:

open tweak app, connect, app says MONO (what is false), than switch to STEREO and back to MONO again...
 
BTW, mono 8M FFT 64x overlap 48k VASAPI has 1/2FPS screen update. The stereo will be slower but still superfast. We planing the only correct THD, THD+N, noise calculations, ok IMD, no freq.response I think. Maybe Cosmos ADC automation for calibration i.e. one-click action and your ADC calibrated for the current frequency and magnitude. Good for me because I spend a lot of my time for calibration now at only 1kHz -.5dbfs. In the auto mode, it could be 10kHz -.5dbfs as well as a preset.
 
  • Like
Reactions: Hee
mansr, I did simplify that because I see no reason to set the different filter for each ADC of 4. And yes 9822 is the same thing as 9842 but has only 8 analog inputs instead of 16. All 4 ADCs need to set up for 9822.
I agree it's probably pointless, just found the discrepancy between the app and the firmware confusing.

Is there any chance you could briefly document the HID protocol used? I have figured out the basic structure, but I see the tweak app sending and requesting some reports without an obvious correspondence to any settings.
 
Since Day1 I had troubles measuring non-balanced signals. On Motu M4 and Clarett 2Pre I get very good results, while on Cosmos ADC I get crappy ones, see below:

View attachment 163423
ODAC v1 measured on Cosmos ADC (1 ch., Stereo mode) - SINAD of 45.5 dB

View attachment 163424
ODAC v1 measured on Motu M4 Input #1 - SINAD of 95 dB
What am I doing wrong? It doesn't matters if I'm using an RCA - XLR cable (V+ connected to hot, V- connected to GND) or the default Cosmos ADC cables with spades, I get the same huge harmonics. It happens on all DACs but Topping D90 MQA (which has a very low output impedance).

Any thoughts, please?
I also can't properly measure my Loxjie D30 V2 RCA output when I connect it directly to my Cosmos ADC. Also tried every XLR connection, harmonics are normal but noise floor is high.

I can only measure unbalanced connections using my interface (INA1651 + OPA1656 balanced out combo) though it has significant noise and THD penalty. Performance of my interface is limited by INA1651. Measurement below at 2Vrms out of D30 V2
 

Attachments

  • JLE Interface1.PNG
    JLE Interface1.PNG
    124.6 KB · Views: 166
  • jle interface2.PNG
    jle interface2.PNG
    119.4 KB · Views: 154
I also can't properly measure my Loxjie D30 V2 RCA output when I connect it directly to my Cosmos ADC. Also tried every XLR connection, harmonics are normal but noise floor is high.

I can only measure unbalanced connections using my interface (INA1651 + OPA1656 balanced out combo) though it has significant noise and THD penalty. Performance of my interface is limited by INA1651. Measurement below at 2Vrms out of D30 V2
I don't know how people here measure these type of things...
but just to give you an idea --- you should NOT measure UNBALANCED signal using just one computer.
There may be potential ground loop which boosts the noise floor.
The ground noise may come from one of your usb outlet that connects DAC to your other usb outlet that connects ADC.

One needs to have two computers to break the ground loops.
Example : Laptop 1 -> DAC -> ADC -> Laptop 2, both using batteries.
You can also replace laptop 1 or 2 with cell phones.

Measuring balanced is fine with a single computer, as balanced can avoid ground loops in general
 
I don't know how people here measure these type of things...
but just to give you an idea --- you should NOT measure UNBALANCED signal using just one computer.
There may be potential ground loop which boosts the noise floor.
The ground noise may come from one of your usb outlet that connects DAC to your other usb outlet that connects ADC.

One needs to have two computers to break the ground loops.
Example : Laptop 1 -> DAC -> ADC -> Laptop 2, both using batteries.
You can also replace laptop 1 or 2 with cell phones.

Measuring balanced is fine with a single computer, as balanced can avoid ground loops in general
Yup I'm using 2 laptops, 1 for the D30 and another for the Cosmos ADC. I've tried most configurations as I've said in my previous post.
Both laptops are running on batteries. I also have the Motu M4 before and it can measure unbalanced signals without problems.
 
Last edited:
To be fair, here's the measurement of iPhone 6P directly connected to E1DA Cosmos ADC. (hot and gnd of XLR)
It is floating and has a very low output impedance (around 5ohm compared to loxjie d30's 100ohm).

So I'm getting the same spectrum with or without the interface buffer.

IP6 direct.PNG


ip6 jle interfac.PNG
 
Last edited:
Cosmos_TWEAK is a temporary tool, we are close to finishing our 100.0% free FFT Windows app where the TWEAK functions will be integrated.

Look, HW as the Cosmos ADC & Setup application is one part of the business... so a MUST HAVE as a given tool for the HW to setup :rolleyes:

your other app's may have this TWEAK additional integrated what is opened to you
 
Back
Top Bottom