E1DA Cosmos APU - Teardown and personal thoughts and measurements

[MarcosCh]

@IVX I believe it would be cool if you open a thread on cosmos adc as phono "digitalizer" to discuss and Q/A besides the main cosmos thread. I guess many people like me get lost there with the main technical discussion.

 

[IVX]

@IVX I believe it would be cool if you open a thread on cosmos adc as phono "digitalizer" to discuss and Q/A besides the main cosmos thread. I guess many people like me get lost there with the main technical discussion.

I did not offer the FW with RIAA officially but you can get that in our discord, and found the details regarding usage. I don't want if all 800 Cosmos ADC users will try to reflash their ADC without a serious reason(serious vinyl geeks involved), you know, reflashing sometimes corrupts the flash content(if that process is interrupted) and the user will need to open the ADC case and short two golden pads ob the PCB.. I worrying..

 

[Rja4000]

I meant a frequency sweep, such as the usual 20 Hz - 20 kHz.

The notch is useless for frequency response measurement, where only the high level fundamental level is of interest.

It is only usefull when measuring very low amplitude distortion or noise, triggered by (and therefore mixed with) a very high level signal.
Typically, THD, SINAD and IMD measurements.
(NB: APU doesn't work for standard IMD, like SMPTE IMD Amir is measuring)

Mutitone is the exception: for obvious reason, you can't use a notch, even if it would be useful.

 

[Rja4000]

any example?

QA480 1 kHz Oscillator + Notch

Update 12-July 2021: The QA480 is on backorder, with new order arrival TBD. The issue, as with our other products, is the silicon shortage discussed here. The QA480 is a very precise 1 kHz sine wave generator combined with 63 dB step attenuator (1 dB step size) and a low-noise notch. This...

quantasylum.com

 

[AudioSceptic]

What one should undersand is the limitations of AP meaasurements:

The notch is useless for frequency response measurement, where only the high level fundamental level is of interest.

It is only usefull when measuring very low amplitude distortion or noise, triggered by (and therefore mixed with) a very high level signal.
Typically, THD, SINAD and IMD measurements.

Mutitone is the exception: for obvious reason, you can't use a notch, even if it would be useful.

Where did I say frequency response? We're talking about measuring distortion over the audible frequency range, not just at a fixed frequency.

 

[Rja4000]

Where did I say frequency response? We're talking about measuring distortion over the audible frequency range, not just at a fixed frequency.

You didn't say it, but since I wasn't sure what you meant by "sweep", I covered both options ;-)

 

[Spkrdctr]

Actually I got it. I have had both, active and passive. Passive with -60dB at base frequency, the main disadvantage was the dependence of passband attenuation on source and load impedance. The active with -80dB at base frequency, as shown in post #3. Opamp selection is crucial (noise, distortion). This notch approach has pros and cons. After I bought your Cosmos ADC, I stopped using the 2T notch. I do not like the single frequency limitation, not mentioning the inevitable effect on phase of distortion component. Now I would only need the switchable divider/gain selector software driven. But frankly, it is just playing the game. All this is inaudible.

Thank you!

 

[Rja4000]

In terms of gain, the +60 dB pre-amplification will match the max. gain of my Motu M4 and will even outperform Focusrite Clarret’s +57 dB of gain

You have to be carefull comparing gain figure from an analog mic preamp to a digital interface.

What I do is always compare the gain delta vs the maximum gain where 100mV rms (-17.8dBu) gets you <0.1% THD and/or 0dBFS.
As an example, this reference gain is obtained just below "36dB" gain mark for the RME UCX II, at -37 gain marking on a Yamaha AD8HR, or even at 46dB gain marking on an analog Millennia HV-3C, where you may push the output to almost 28dBu before distortion gets greater than 0.1% (or even than 0.001%).
Also, on my venerable Motu 828mkIII FW, you get distortion (saturation) before you reach 0dBFS.

Then you're comparing things that can be compared, since, in practice, your mic will always give you (more or less) the same voltage for a given SPL, whatever the preamp you plug it into.

If you plug the APU Preamp to the E1DA ADC, you'll get best results with setting the later to 10V (22dBu) range.
Therefore, the 60dB gain will get you 20dB more amplification than you'd need for 100mV.
Equivalent to 56dB gain marking on the RME.
With 100mV, and 34dB APU gain, you'll get 5V, so you should probably use the 6.3V range on the ADC.

 

[oal]

I agree with Ivan that "you don't need that if have no plan to measure distortions, noise, or a dynamic range at finer levels than $30K Audio Precision analyzers do." Of course, Cosmos APU is designed to measure dynamic range and THD+N of cutting edge DACs with high accuracy. But I use it as an input buffer for the Cosmos ADC to measure not so good DIY DACs. The input impedance of the Cosmos ADC is 640ohm for input range 1.7Vrms, unbalanced impedance is about 30% less (450ohm). It is too low for some unbalanced DACs. For example, a $10 DAC board generates much harmonic distortion because of the low input impedance.

While input impedance of Cosmos APU is 10k ohm (for unbalanced) which is equivalent to most of audio interfaces. Cosmos APU + ADC work fine with the $10 DAC.

However, an input buffer with variable gain is best suited for this purpose. (Cosmos Scaler?)

 

[Rja4000]

an input buffer with variable gain is best suited for this purpose. (Cosmos Scaler?)

Several of us are looking forward for the scaler, but Ivan seems not convinced he'd get his money back. Will there be enough sales ?
That's a fair concern, IMO.
If you want one, maybe you should tell him directly.

 

[Ron Texas]

https://e1dashz.wixsite.com/index/cosmos-apu :

"Please, before you decide to buy Cosmos APU, make sure that you completely understand what is it. Analog Processing Unit is the first in its kind device on the market, hence, you need to understand do you really want:
a) to do the measurement of your DAC/AMP [email protected]/10kHz at <-150/-130db harmonics level, or [email protected] down to -132db. In that case, you need the Notch part of Cosmos APU.
b) measuring your DAC/AMP DR/SNR with the residual APU's noise 130nVrms(A) with the Preamp of Cosmos APU.
c) to rip vinyl records with the Preamp of Cosmos APU + Cosmos ADC with built-in DSP RIAA EQing. Please, check Cosmos ADC page to find the CosmosAdc_v13_RIAA.hex.
d) to use a Pro condenser mic with 48V phantom power with the Preamp of Cosmos APU."

Thanks, because now I know I don't need this gadget.

 

[maxwell3e10]

The first stage (from right) contains a +20 dB gain stage, followed by the adjustable notch filters (@ 1KHz or @10 KHz), then the output stage with a selectable 0 dB/+6 dB gain. There’s also a +34 dB/+60 dB pre-amplification stage that can be used for measuring the background noise or DUT’s dynamic with any ADC you have around.

I think this description is backwards. The input of the circuit is on the left, with a differential amplifier that has a gain of 1 or 2 (6db). Then there is the notch filter and finally a 20 dB amplifier with single-ended output. It would make no sense to put an amplifier before notch filter.

 

[Rja4000]

I think this description is backwards. The input of the circuit is on the left, with a differential amplifier that has a gain of 1 or 2 (6db). Then there is the notch filter and finally a 20 dB amplifier with single-ended output. It would make no sense to put an amplifier before notch filter.

Definitely !

 

[trl]

maybe, however, Cosmos ADC has a gain range 1.7-10Vrms so you have an additional 15db to adjust it 34db and 60db is compromising between 3 different tasks:

Yep, definitely, everyone could do the 1.7-10V RMS ranger from the Cosmos ADC to handle to final gain of the APU+ADC combo properly. As the matter a fact, here's a test voice recording I did few days ago with an SE Electronics X1S condenser microphone, with Phantom power turned ON, ADC set to its max. sensitivity @1.7V RMS and APU set to its lowest +34 dB. Comparing to Motu M4 this recording was about 1 dB lower in peaks, but RMS levels were about the same and after saving with "Normalisation" (from Audacity, GarageBand, Ableton etc.) the peaks and entire's track volume are automatically adjusted anyway.

SE Electronics X1S condenser microphone, Phantom power turned ON, E1DA ADC @ 1.7V, E1DA APU @ +34 dB​

There is unperceivable background noise at all induced by the ADC nor by the APU in all audio recordings / test I've done, but if there's any noise heard with sensitive headphones on the final recording then it's due to the microphone itself. This makes Cosmos (APU + ADC) combo my best audio recording interface I've ever had. Although, I'll still keep the M4 and Clarett 2Pre USB for a while, as I have too many desktops and laptops around than might still benefit of these interfaces.

 

[trl]

I think this description is backwards. The input of the circuit is on the left, with a differential amplifier that has a gain of 1 or 2 (6db). Then there is the notch filter and finally a 20 dB amplifier with single-ended output. It would make no sense to put an amplifier before notch filter.

Thanks for noticing, although I'm unable to modify that paragraph now, but usually the audio signals travel from left to right in most electronic schematics / diagrams. Such errors happen when writing docs at night.

 

[AudioSceptic]

https://e1dashz.wixsite.com/index/cosmos-apu :

"Please, before you decide to buy Cosmos APU, make sure that you completely understand what is it. Analog Processing Unit is the first in its kind device on the market, hence, you need to understand do you really want:
a) to do the measurement of your DAC/AMP [email protected]/10kHz at <-150/-130db harmonics level, or [email protected] down to -132db. In that case, you need the Notch part of Cosmos APU.
b) measuring your DAC/AMP DR/SNR with the residual APU's noise 130nVrms(A) with the Preamp of Cosmos APU.
c) to rip vinyl records with the Preamp of Cosmos APU + Cosmos ADC with built-in DSP RIAA EQing. Please, check Cosmos ADC page to find the CosmosAdc_v13_RIAA.hex.
d) to use a Pro condenser mic with 48V phantom power with the Preamp of Cosmos APU."

But you need 2 APUs for ripping vinyl as it's single channel, unless of course you have a mono cartridge!

 

[pma]

The input impedance of the Cosmos ADC is 640ohm for input range 1.7Vrms, unbalanced impedance is about 30% less (450ohm). It is too low for some unbalanced DACs. For example, a $10 DAC board generates much harmonic distortion because of the low input impedance.

Exactly. There are many situations when we need input impedance at least 10kohm and more. I understand that @IVX or Topping use BJT input opamps and low resistor values to get the lowest noise, but for measuring purposes we would really need much higher input impedance, like 100kohm or 1Mohm (oscilloscope standard) and such requirement would lead to the circuit design with JFET input opamp. For these purposes, I have used this, now for 20 years.

Audio Buffer in english

The concept is still lively and some degradation in noise and THD is often negligible compared to the measured device.

 

[Herbert]

@IVX Make it portable with batteries and there will be a use in the semiprofessional broadcast/film market.
Almost all DSLRs that provide stunning image suitable for cinema or Netflix lack good, noise free mic preamps.
+48V Phantom Power is mainly available as expensive accessoires.
As one example, the self noise of my Panasonic S5' preamp is
about -53 dB from full scale (Ramped all up to +6dB gain) - the best I can get out of it is about -70dB (-12db gain)
but a headroom between -9 and -12db has to be considered during recording - so the best is about -55 to -60dB.
Boosting the signal of a good condenser Mic and providing it with Phantom power would help to overcome these limitations.

 

[oal]

For these purposes, I have used this, now for 20 years.

Your audio buffer looks nice! Before purchasing Cosmos APU, I made a simple imput buffer using an OP AMP (Analog Devices AD797). I have adopted the noninverting configuration of the OP AMP, and its impedance is enough high for the DAC. However, I could not lower its noise level as Cosmos APU's. Unfortunately, I am a software engineer, not a professional circuit designer.

 

[trl]

@IVX Make it portable with batteries and there will be a use in the semiprofessional broadcast/film market.

But it's small enough to be called portable, right? We're talking about a PCBA of 90x50x29mm or 104x61x39mm with the original case, so pretty similar with a power bank containing a couple of 18650 or 21700 batteries inside, like the one I've used in my test. APU is using a case with the same dimensions as E1DA ADC, see below:

​