E1DA Cosmos ADCiso Review

[pseudoid]

...When the user mixes up scaler vs scalar, I see that he's not an engineer...

I ran downstairs and got some popcorn when I read this above comment, that started the whole ruckus!
So, the cognoscenti engineer is demeaning someone (an engineer and/or an usher, alike) because they are not fully versed in the nuances of scalAr and scalEr.
I say BFD, it is not a word (concept) that is in daily usage for most *engineers and/or *ushers, alike!
But what I find totally amazing is the fact that even a self-appointed 'cognoscenti' engineer - who must use the logarithmic measurements daily - continues to spell "dB" as db.
"Give it a rest, Roy!" and "Throw the towel in, Howard!"

 

[restorer-john]

But what I find totally amazing is the fact that even a self-appointed 'cognoscenti' engineer - who must use the logarithmic measurements daily - continues to spell "dB" as db.

It always annoyed me that the tape dynamic range expansion system was called dbx. but the logo looks cool.

 

[KSTR]

But what I find totally amazing is the fact that even a self-appointed 'cognoscenti' engineer - who must use the logarithmic measurements daily - continues to spell "dB" as db

The dB (pseudo-) unit -- as it just denotes a log20'd ratio -- is often mis-spelled even in technical literature, plus it often gets mis-treated with prefixes while it already has one, the d (deci). 1/1000th of 1 dB is not 1 mdB (1 mili-deci Bel), it's 10mB (10 mili Bel)... similar to kg (kilo-gram, the mass SI unit and the only one with a prefix).
So I think we should give the guys some slack ;-)

 

[Dawsoo]

I have owned the E1DA cosmos ADC for more than a year or a year and a half and it is a great helper for me, I know exactly how to use it from the beginning and it helped me a lot, the HW is designed excellently in terms of the achieved parameters, however, even the criticism that was heard here is justified:
1. A classic manual sheet one-two in pdf, where everything would be clearly written, is after all a small thing that can be done without any costs.
2. I also have reservations about the HW interface (I prefer not to talk about micro dip switches and USB C about the 43V input on the small, atypical 2.5 jack), the first two wear out pretty quickly with use, and the minijack is the worst thing that exists for audio in general.
3. Switching between mono and stereo using win control is imho also nonsense which only complicates things, I would prefer a fixed mono/stero FW or HW switch.
4. The ISO version is a great thing and measuring with a PC is a must, it annoys me that I have an old version with which I always have to have a USB isolator.

just my 5c.

 

[PortalKeeper]

Is that IMD hump audible?

 

[PortalKeeper]

That's out of topic here.
Short answer is: in general yes, at low gain/high input level.

Sorry, maybe this question is more relevant than that last one. Would dynamic range be lost with a Neve or millenia mic pre even if using the balanced 43v 2.5mm port on the front of the cosmos?

 

[restorer-john]

The dB (pseudo-) unit -- as it just denotes a log20'd ratio -- is often mis-spelled even in technical literature, plus it often gets mis-treated with prefixes while it already has one, the d (deci). 1/1000th of 1 dB is not 1 mdB (1 mili-deci Bel), it's 10mB (10 mili Bel)... similar to kg (kilo-gram, the mass SI unit and the only one with a prefix).
So I think we should give the guys some slack ;-)

Bel as in Alexander Graham Bell. Like Farad(ay), Ohm, Henry, Volt etc. They get the capital letter.

gram (gramme), milligram, microgram was not a person's name.

 

[KSTR]

I have owned the E1DA cosmos ADC for more than a year or a year and a half and it is a great helper for me, I know exactly how to use it from the beginning and it helped me a lot, the HW is designed excellently in terms of the achieved parameters, however, even the criticism that was heard here is justified:
1. A classic manual sheet one-two in pdf, where everything would be clearly written, is after all a small thing that can be done without any costs.
2. I also have reservations about the HW interface (I prefer not to talk about micro dip switches and USB C about the 43V input on the small, atypical 2.5 jack), the first two wear out pretty quickly with use, and the minijack is the worst thing that exists for audio in general.
3. Switching between mono and stereo using win control is imho also nonsense which only complicates things, I would prefer a fixed mono/stero FW or HW switch.
4. The ISO version is a great thing and measuring with a PC is a must, it annoys me that I have an old version with which I always have to have a USB isolator.

Well, there is some price to pay for getting such levels of audio quality this affordable.

I think it is clear that this is a fragile device that you likely integrate in a larger, fixed setting, mechanically.

In normal operation you don't adjust the sensitivity switches (4.5V obviously is the range that fits most needs) and one better hot-glues a short adapter cable to the fragile USB-C port which I consider the most problematic issue (when it breaks repair may get difficult), so why, @IVX, would't a regular USB-B have done, which is known to last mechanically for years and years to come.

The software mode switching of stereo / stereo-RIAA / mono is indeed a bit awkward, as is the filter switching (especially as there no access from other OS's than Windows). But is is manageable.

I also have the non-isolated version but together with an (equally affordable) Topping HS01 it is perfect, and I can use the HS01 for other duties as well.

 

[Dawsoo]

Thank you very much for the information that it works with the HS-01, which is otherwise not very compatible with anything. I sometimes need two isolators when measuring class 1 (PE) equipment and this would help solve that.

 

[KSTR]

Sorry, maybe this question is more relevant than that last one. Would dynamic range be lost with a Neve or millenia mic pre even if using the balanced 43v 2.5mm port on the front of the cosmos?

If you really plan to max out a Millenia's max output level of +32dBu this also means you will run into occasional clipping just as well so that's not a meainingful use case.
If you back off and reduce max output by 10dB (a factor of only 3.16x) the 10V range of the Cosmos is the range that is appropriate and will fully retain dynamic range.

 

[KSTR]

Thank you very much for the information that it works with the HS-01, which is otherwise not very compatible with anything. I sometimes need two isolators when measuring class 1 (PE) equipment and this would help solve that.

I have to note that under Windows I occasionally have glitching problems with the HS01 but under Linux (Mint Xfce) it runs perfectly even at 768kHz on an 11year old small laptop.

 

[Rja4000]

Bel as in Alexander Graham Bell. Like Farad(ay), Ohm, Henry, Volt etc. They get the capital letter.

gram (gramme), milligram, microgram was not a person's name.

You forgot André-Marie Ampère.
And James Watt

 

[Rja4000]

I also have the non-isolated version but together with an (equally affordable) Topping HS01 it is perfect, and I can use the HS01 for other duties as well.

I wouldn't say perfect. But workable.

 

[PortalKeeper]

If you really plan to max out a Millenia's max output level of +32dBu this also means you will run into occasional clipping just as well so that's not a meainingful use case.
If you back off and reduce max output by 10dB (a factor of only 3.16x) the 10V range of the Cosmos is the range that is appropriate and will fully retain dynamic range.

Thanks but not according to what @Rja4000 said earlier:

“Note though that high end analog mic preamps usually have an output level that may reach more than this ADC will accept, even on max 10V range.
(A Neve or a Millennia, as an example, may push up to 28dBu on the output before distortion.
So you'll loose part of their potential dynamic range.)”

 

[KSTR]

The Comsos, in mono mode, has an audio band RMS noise of about -128dBFS which is industry-leading. The only ADC which could resolve better down low would be of the stacking type.
But, at first I'd really like to see the actual noise floor of a Millenia pre, even with 200R shorted inputs, at reasonable gain settings so that the actual peak voltage could be reached. I'm not convinced that this will show a lower noise floor (I think it's actually quite impossible with any serious amount of gain).

 

[Rja4000]

The Comsos, in mono mode, has an audio band RMS noise of about -128dBFS which is industry-leading.

My E1DA Cosmos ADCs (non ISO) have around -128dBFS (A) noise in mono.
That's A weighted.

But, at first I'd really like to see the actual noise floor of a Millenia pre, even with 200R shorted inputs, at reasonable gain settings so that the actual peak voltage could be reached. I'm not convinced that this will show a lower noise floor (I think it's actually quite impossible with any serious amount of gain).

I measured EIN (A) at 150 ohm on the Millennia HV-3C (I own one) for various gain settings.
At minimum gain (around 8.8dB), I get -110.6dBu (A) EIN, which is excellent.
Since the HV-3C is starting to distort (-60dB THD) for that gain (for any gain actually) at 28.0dBu output (ie 19.2dBu input), that gives us 129.8dB (A) dynamic range.
If you limit the ouput to 22dBu, to match the ADC range, the dynamic range is lowered to 123.8dB (A), whatever the ADC performance.
You'd get 2dB better with the RME ADI-2/4 Pro SE (24dBu range)
In both cases, the ADC (in mono) is not the limiting factor, which is pretty awesome.

Not that this difference at such a level of dynamic range matters.
It's excellent anyway, and, in practice, you won't go there.
But in absolute technical terms, those are the facts.

 

[KSTR]

In both cases, the ADC (in mono) is not the limiting factor, which is pretty awesome.

Yes. However, the elephant in the room is the microphone ;-) We won't find many that will achieve +28dBu out of the Millenia with only ~10dB of gain, even when allowing for 10% distortion from the mic alone. The moment we need to apply 20dB or more to even get there, the ADC is definitely not compromising the PreAmp's SNR even when we restrict its output to +22dBu (10Vrms).

 

[Rja4000]

Yes. However, the elephant in the room is the microphone ;-)

Fully agree

We won't find many that will achieve +28dBu out of the Millenia with only ~10dB of gain, even when allowing for 10% distortion from the mic alone.

On that, I'm not sure -at least theoretically:
An old version of the DPA 4006 is supposed to output 40mV/Pa, and to accept a peak SPL of 146dB for 10% THD.

So, 40mV, or -28dBV or -25.8dBu at 94dB SPL would give us 26.2dBu on the input of the preamp for 146dB SPL.

For maximum output level of 28dBu and a gain of 8.8dB, you need a maximum of 19.2dBu at the input. So 139dB SPL at the microphone.
And at 139dB peak, the DPA 4006 should have 1% THD.

What on earth could be a practical use of recording 139dB SPL with a high sensitivity condenser microphone, I have no clue, though.

 

[KSTR]

What on earth could be a practical use of recording 139dB SPL with a high sensitivity condenser microphone, I have no clue, though.

Well, record a snare drum hit while capturing a fruit fly's poop from nearby, haha!

 

[Rja4000]

Well, record a snare drum hit while capturing a fruit fly's poop from nearby, haha!

More seriously, they are probably mostly used to record grand Orchestra (with cannons !)