E1DA Cosmos ADCiso Review

[Blimple]

Why didn’t you make a THD vs frequency measurement? It's very important imho

 

[staticV3]

Do you mean that the measurements aren't automated WHEN THE NOTCH FILTER IS USED?

You have to load the Notch correction file into REW once. After that, you can make manual RTA measurements and automated stepped measurements like normal. No special workflow required.

It should be said though that the notch has a fixed frequency of either 1kHz or 10kHz. Can't measure any other frequencies using it.

 

[Rja4000]

Does anyone know how the E1DA APU improves the SINAD that you can measure?

You may look at my review/measurements of Topping D50 III

 

[capslock]

There is no unbalancing, it is a result from the used setup for the ADC. With its volume control at max the left output channel is the average of left and right input channel and right output channel is the right input channel.
That's where the 3dB better noise values of the left channel comes from, but actually it's the same as the right channel (give or take a fraction of a dB).

Assuming Amir fed both ADCiso inputs from the same signal source, effectively paralleling them (not sure if the AP has two independent signal generators).

 

[capslock]

An alternative version with spdif-in would be fantastic...

How would that be useful? Almost any other ADC you could connect would be worse.

 

[capslock]

Tried to buy one. Not available. Contacting the developer page eMail address gave no reply. What I saw is that the input impedance is pretty low for some measurements like preamps and specific tube preamps.

Their Aliexpress shop no longer sells to Germany because some expensive certification is needed that they will pay for recycling. But in my experience, Sunny (Ivan's wife) replys within 24 h and I have bought directly from them. Linsoul and Hifiexpress had them in stock, too, a few days ago when I last looked.

 

[alekksander]

ayee… this is so good! exactly what i need to digitalize one of my two vinyls.

 

[capslock]

Amir, what kind of signal generator did you use to test the AP's ADC in high performance mode? To the best of my knowledge, the AP's analog signal generator does not have HHD < 160 dB, nor do the digital channels.

Do the generators used for the IMD tests change when subjected to a low impedance load? There was some discussion about the ESS hump in Cosmos ADC either here or on diyaudio. The ADC was driven from another ESS DAC. It turned out the hump decreased substantially when a Cosmos Scaler was added. The interpretation, AFIR, was that Cosmos ADC was loading the DAC so much it brought out the hump in the DAC.

 

[KSTR]

Assuming Amir fed both ADCiso inputs from the same signal source, effectively paralleling them (not sure if the AP has two independent signal generators).

From the Dashboard it seems clear that both inputs were fed, with the same signal. The older AP's only have one analog generator but two fully independent output stages, the APx probably similar.

 

[KSTR]

How would that be useful? Almost any other ADC you could connect would be worse.

I assume @Groom most certainly meant SPDIF-Output.

 

[peniku8]

APx555 analyzer attempts to justify its 100 times higher price with this level of performance:

View attachment 365918


Note again that AP's measurement is with 200K Ohm input impedance with automatic maximum input voltage of 160 volts

Judging by the last sentence, if I understand this correctly, this is the analyzer's DR including the auto scaler?
How much gain does AP's scaler apply?
There is the Cosmos Scaler to extend the capabilities of the ADC with this exact feature (and 200K Zin), which I've measured to come in at a dynamic range of ~140dB (from an extremely low level signal with the Scaler's maximum gain of 26dB to 0dB gain 4Vrms @-0,7dBFS).
For those who haven't seen it, the ADC can be set to mono, which lowers THD+N by about 3dB (but also halves its input impedance!).
For super precise measurements, there is also the Cosmos APU with a 1KHz notch and a low-noise preamp, so comparing the ADC by itself to the AP analyzer with its scaler enabled is a bit of an apples to oranges comparison imo, since the Cosmos ecosystem has more to offer than just the ADC

 

[Sokel]

Their Aliexpress shop no longer sells to Germany because some expensive certification is needed that they will pay for recycling. But in my experience, Sunny (Ivan's wife) replys within 24 h and I have bought directly from them. Linsoul and Hifiexpress had them in stock, too, a few days ago when I last looked.

Available at Audiophonics as well:

E1DA - Manufacturer of DAC, ADC and audio signal processing units - Audiophonics

E1DA is a company specialized in the design of DACs, amplifiers and DSPs to accompany your headsets and mobile devices. It offers very high quality products at fair prices and with incredible performance.

www.audiophonics.fr

 

[capslock]

I assume @Groom most certainly meant SPDIF-Output.

Thanks, that makes more sense! I wonder if this can be used to make synchronous measurements. One would need a soundcard / pro audio interface that accepts the wordclock signal from the ADC in ordner to not have to depend on the PLL in the interface for synchronization.

 

[Vict0r]

I have no use case for this, but I am very glad it exists. Especially for a price as accessible as this. Amir needs more well-equipped soldiers in his measurement regiment!

 

[pkane]

Does anyone know how the E1DA APU improves the SINAD that you can measure?

The APU notch makes things better by reducing the effect of ADC noise (in the +6dB mode) and eliminating ADC distortions from the measurement.

Compare this to @VintageFlanker 's measurement of DO300 above (this is the non-ISO version of Cosmos ADC):

 

[weme]

@amirm
I've been waiting for this - thanks for your measurements.

Just a small correction here:
I started with the default configuration of 4.5v range at 1.7 ohm input impedance:
->
... 4.5v range at 1.7k ohm input impedance:


And here is an addition to the technical data:

XLR input - max. voltages / impedance:
DIP switch attenuator for 8 sensitivity ranges from 1.7-10 Vrms. Each range has a different input impedance:
1.7 Vrms 640 Ω - (6.83 dBu)
2.7 Vrms 1 kΩ - (10.85 dBu)
3.5 Vrms 1.3 kΩ - (13.1 dBu)
4.5 Vrms 1.66 kΩ - (15.28 dBu)
6.7 Vrms 2.46 kΩ - (18.74 dBu)
7.6 Vrms 2.82 kΩ - (19.83 dBu)
8.5 Vrms 3.12 kΩ - (20.8 dBu)
10 Vrms 3.48 kΩ - (22.22 dBu)
Unbalanced impedance is approx. 30 % lower. All inputs are AC-coupled (-7.5 V to +2.5 V DC permissible).
Inputs are balanced, but can be used with unbalanced outputs in any combination. It does not matter whether +IN to GND or -IN to GND or IN+ to OUT+ or OUT-, the same applies to IN- etc.

AUX (2.5 mm) - max. voltages / impedance:
43 Vrms 13.6 kΩ
Do not plug this socket in or out when the amplifier is switched on!
I would advise against using the 2.5 mm AUX socket. Use the 10 Vrms XLR input with two upstream resistors, e.g:
2 x (560 + 2700 Ohm in series *1)) on pin 2 / 3 result in an impedance of ~ 10 kOhm / 31.16 dBu ...
For frequency response measurements at the upper range limit (~ > 200 kHz), the voltage drop across the resistors (*1) could be compensated for with a capacitor connected in parallel.
The correction files CosmosADC_FR.cal / CosmosADCiso_FR.cal contain too few sampling points.

 

[IVX]

It seems Cosmos ADC Grade B at 0dbfs i.e. 4.5-4.6Vrms will give equal to APx555 THD+N -121db, 4V input has lack -1.5db. AP isn't critical to such a difference due to autoranger onboard, but Cosmos ADC is straight ADC and nothing more. That's why it is correct to measure THD+N at -0.5..0dbfs level. BTW, L7 did measure Cosmos ADC 2 years ago and got -121.2db of sinad at 4.5Vrms.

 

[Grooved]

With the latest firmware it provides an SPDIF source signal at an internal pin... you "only" need to make your own little PCB and fit the output circuitry of SPDIF and/or TOSLINK there.

I must have missed this firmware then, where can I download it?
The firmware I updated the last time is one that offers the Masterclock on an internal pin, but it doesn't offer SPDIF signal directly

 

[KSTR]

I must have missed this firmware then, where can I download it?
The firmware I updated the last time is one that offers the Masterclock on an internal pin, but it doesn't offer SPDIF signal directly

Find attached the latest versions, for standard ADC and for ADCiso.

Install instruction here: https://www.audiosciencereview.com/...ds/e1da-cosmos-adc.27038/page-75#post-1959444

 

[KSTR]

Thanks, that makes more sense! I wonder if this can be used to make synchronous measurements. One would need a soundcard / pro audio interface that accepts the wordclock signal from the ADC in ordner to not have to depend on the PLL in the interface for synchronization.

I will be using RME ADI-2 Pro for this, it can sync on SPDIF input for USB connection, even at mutiples of the sample rate. From there I can connect a DUT on SPDIF out and have the whole shebang in sync.
As I have two ADI-2 Pros, I have all options to run the DUT at, say, 48Khz and the Cosmos ADC at 192kHz, still in full sync...
For higher sample rates on the Cosmos, while it does run the SPDIF pin correctly at these rates, it remains to be seen if the physical interface and the partner device can handle it. The other, much more complicated option would be to create a divided down (2x/4x/8x/16x) version from I2S (not using the SPDIF Firmware option) and use that to feed an I2S-->SPDIF converter (WM8804).