E1DA Cosmos ADC

[sseppala]

@IVX , reagarding the input voltage ranges of Cosmos ADC, could you comment about their break-up voltage limits. I mean, how easily you can brake up the inputs, for example if you accidentally choose a wrong input range.

 

[IVX]

@IVX , reagarding the input voltage ranges of Cosmos ADC, could you comment about their break-up voltage limits. I mean, how easily you can brake up the inputs, for example if you accidentally choose a wrong input range.

Cosmos ADC with 1.7V gain setting withstands 10Vrms input.

 

[Rja4000]

RME ADI-2Pro FS DAC +19dBu (with THD pre-compensation) --> E1DA Cosmos ADC @192k, 6.7v. I obviously need a better sine generator

View attachment 159429

Just for fun
DAC: RME ADI-2 Pro fs R - Sum of 2 channels output - 19dBu range +0.5dB - HD2-3-5 compensation
ADC: E1DA ADC Grade A (128.2dBA measured) 6.7V range Mono mode

 

[Rja4000]

Just for fun
DAC: RME ADI-2 Pro fs R - Sum of 2 channels output - 19dBu range +0.5dB - HD2-3-5 compensation
ADC: E1DA ADC Grade A (128.2dBA measured) 6.7V range Mono mode

View attachment 162403

With better HD compensation, I even reached 120.9dB SINAD

 

[pkane]

With better HD compensation, I even reached 120.9dB SINAD

It looks like SINAD is noise-limited, so HD compensation isn’t going to do much to help improve it.

 

[IVX]

Guys, some Cosmos ADC users complain regarding the DC offset is too high. I have no idea why they care about that, anyhow it is an audio ADC i.e. inherently AC machine but what do you guys think about that?

 

[mansr]

Guys, some Cosmos ADC users complain regarding the DC offset is too high. I have no idea why they care about that, anyhow it is an audio ADC i.e. inherently AC machine but what do you guys think about that?

I'm not too bothered by it. It's quite small, and I typically remove any DC offset anyway.

 

[Mike123]

what do you guys think about that?

How many DC offset? When recording an ADC signal and playing it back through a DC DAC and DC amplifier, the speaker systems may not like DC offset of more than 50-100 mV.

 

[IVX]

1mV/1V but seems I can improve that 100x times i.e. about 10uV/1V

 

[Mike123]

1mV/1V

This is not much for me. But there are different people

 

[IVX]

This is not much for me. But there are different people

ok, however, you called 50mV as a margin DC offset, so only the DC-coupled amp 312W@8ohm will reach 50mV, correct? A bit surrealistic combo to me ADC->DC_DAC->DC_amp>50Vrms output.

 

[Mike123]

There is no relationship between DC offset and power. DC offset produces a constant offset for the woofers. What is the output offset without signal?

 

[mansr]

There is no relationship between DC offset and power. DC offset produces a constant offset for the woofers. What is the output offset without signal?

On mine it's about 0.0008 times full scale.

 

[Mike123]

I wrote that this offset is not large. But there are people who don't like this offset.

 

[Rja4000]

ok, however, you called 50mV as a margin DC offset, so only the DC-coupled amp 312W@8ohm will reach 50mV, correct? A bit surrealistic combo to me ADC->DC_DAC->DC_amp>50Vrms output.

Problem I see is when this offset fluctuates
Does it ?

 

[IVX]

it is floating of course, if the device supposes to be AC, no reason to care about its DC properties and make it more expensive in production.

 

[sseppala]

I'm not familiar with all the details of the use cases that this DC offset affects, but it got me thinking. Would it help any use case if there was integrated digital high pass filter that you could enable or disable through the tweaking app? This probably wouldn’t be too difficult to implement?

 

[MC_RME]

A high pass to get rid of the DC is part of ANY audio ADC out there, so DC is typical below -100 dBFS. Nobody wants to have DC in their recording. I already explained to Ivan (no clue why he wrote 'I have no idea why they care about that' above):

Quote: the Cosmos ADC has a basic problem - the high amount of DC (typcially -56 dBFS). This sometimes causes measurement data to be wrong (for example in HpW the RMS value always includes DC), and for audio recording it causes pops/clicks on start/stop and edit of a recording. /Quote

It is just bad practise and against the standard, will only cause unexpected issues sooner or later.

Now the issue with ESS ADCs is that their included High Pass, which they dared to rename DC Block in their latest data sheets, is a sloppy implemented 25 Hz filter, see this graph:

And improves DC only to -85 dBFS. It's unusable for both measurements and audio recording. ANY other ADC out there will give you a steep 1 Hz filter and crunch DC down to -130 dBFS!

So I suggested to Ivan to try the integrated DC offset option, a unique feature of the ESS, which would allow to compensate the DC similar to the THD compensation, and could be added easily to the Tweak app. Ivan managed to reach -120 dBFS this way, but as we are talking microvolts this offset is a little bit different per device, and resulted in -90 dBFS on my unit. This ended in the idea to add a basic DC offset to all units, to get better than the default -56 dBFS, and additionally add the DC offset settings to the Tweak app.

Now its on Ivan what he does.

 

[pkane]

A high pass to get rid of the DC is part of ANY audio ADC out there, so DC is typical below -100 dBFS. Nobody wants to have DC in their recording. I already explained to Ivan (no clue why he wrote 'I have no idea why they care about that' above):

Quote: the Cosmos ADC has a basic problem - the high amount of DC (typcially -56 dBFS). This sometimes causes measurement data to be wrong (for example in HpW the RMS value always includes DC), and for audio recording it causes pops/clicks on start/stop and edit of a recording. /Quote

It is just bad practise and against the standard, will only cause unexpected issues sooner or later.

Now the issue with ESS ADCs is that their included High Pass, which they dared to rename DC Block in their latest data sheets, is a sloppy implemented 25 Hz filter, see this graph:

View attachment 163106

And improves DC only to -85 dBFS. It's unusable for both measurements and audio recording. ANY other ADC out there will give you a steep 1 Hz filter and crunch DC down to -130 dBFS!

So I suggested to Ivan to try the integrated DC offset option, a unique feature of the ESS, which would allow to compensate the DC similar to the THD compensation, and could be added easily to the Tweak app. Ivan managed to reach -120 dBFS this way, but as we are talking microvolts this offset is a little bit different per device, and resulted in -90 dBFS on my unit. This ended in the idea to add a basic DC offset to all units, to get better than the default -56 dBFS, and additionally add the DC offset settings to the Tweak app.

Now its on Ivan what he does.

I assume that explains this in my measurements:

So, yes, while I can filter that part in software, it would be better to have a much smaller DC component in the ADC used for measurements, to answer your question, Ivan (@IVX ).

 

[Grooved]

A high pass to get rid of the DC is part of ANY audio ADC out there, so DC is typical below -100 dBFS. Nobody wants to have DC in their recording. I already explained to Ivan (no clue why he wrote 'I have no idea why they care about that' above)...

I'm not sure, but maybe @IVX was referring the "no idea" to the fact of "needing better than 1mV/1V", more than something like "no idea why they care about having nothing done on DC offset"