E1DA Cosmos ADC

[nanook]

@IVX: I'm still searching for a DAC that is capable to cover the "whole" Nyquist frequency spectrum.
The ADCiso has a remarkable bandwidth of ca. 320 kHz and I'm searching for an appropriate companion on the DAC side.
It should be ES9039q2m based because the ES9038 has a lot more "grass" in its spectrum at intermediate levels.

You were once planning a Cosmos DACiso as far as I remember, but I haven't read about it for long.

—> Do you have a recommendation for a DAC that can make reasonable use of 768 kS/s?
—> Is the Cosmos DACiso still on your ToDo list?

 

[IVX]

Well, to me 9039S is Cosmos DAC, especially with Cosmos LPF board

 

[nanook]

Well, to me 9039S is Cosmos DAC, especially with Cosmos LPF board

@IVX: Thanks a lot for your answer.
I'm aware that the 9039S has extremely nice performance. The 9039Q2M is great I agree and you obviously made a very nice composite OpAmp - the bare AD8397 does not at all live up to this performance according to the datasheet.

But: The Specification you give on the 9039S web-page sounds like it's restricted to 384kS/s PCM with the corresponding max. bandwidth of 192kHz. Even with slow roll-off FIR filters this would give 0.43 * FS = 165kHz @ -3dB so some analog filtering is involved as well.
Don't get me wrong - limiting BW is very reasonable for DAC/HPA but for a measuerment DAC this is a restriction I'd like to avoid. Of course I could get another SMSL D-6s, modify the filtering in the I/V- and RCA- output stages and add a USB-isolator, but I'd rather have a plain DACiso with 350kHz analog bandwidth @ 768kS/s ;-)

Cosmos 9039S specification:
Input: USB-C tighter grip version, PCM up to 32b/384kHz DSD256

Frequency response: DC-130kHz +.05/-3db @32b/384kHz, -.5db at 45kHz
DC performance is limited by DC protection

Edit: Ok., looking into the ES9039Q2M datasheet and knowing, that the 768kS/s do not have a fast roll-off PCM filter, -3dB @ 350 kHz is sporty. The slow filters have a passband (-3dB) at 0.43*FS = 330 kHz (without any LP-filter in the I/V-stage etc.). Getting beyond this would require a peaking in the analog filter.
You do need some LP-filter to get rid of the high-frequency content at the output of a noise-shaping SD- DAC and ideally this filter would get switched with the actual sample-rate (no need for a ca. 600kHz LP-filter when running at 48kS/s). A higher order LP-filter would be an advantage. I wonder what RME did in the "ADI 2 Pro fs SE", they claim a -3dB frequency response of DC - 285kHz for 768 kS/s.

 

[horias2000]

Hello all. Recently I tried measuring a device and it looks like my scaler is acting up. I have very high THD when I connect the DUT to the Cosmos ADC via the scaler. Below you can see what I mean.

This is when I measure my D50s using the scaler. If I connect the D50s output directly to the inputs of the ADC, I get a THD+N of -110dB but when I use the sclaer, I get what you see above. I tried several cables and several gain settings but the THD looks exactly the same. Has anybody seen this? This looks to me like one of the op-amps might be damaged. Most probably the OPA1656 that sits on the input. I do have several OPA1612 around but not the 1656. I'll order a few and I will try replacing them and I'll report back. Any suggestions are welcomed.

 

[jkim]

Hello all. Recently I tried measuring a device and it looks like my scaler is acting up. I have very high THD when I connect the DUT to the Cosmos ADC via the scaler. Below you can see what I mean.

View attachment 488135

This is when I measure my D50s using the scaler. If I connect the D50s output directly to the inputs of the ADC, I get a THD+N of -110dB but when I use the sclaer, I get what you see above. I tried several cables and several gain settings but the THD looks exactly the same. Has anybody seen this? This looks to me like one of the op-amps might be damaged. Most probably the OPA1656 that sits on the input. I do have several OPA1612 around but not the 1656. I'll order a few and I will try replacing them and I'll report back. Any suggestions are welcomed.

Do you have any other DAC to try? The scaler has extremely wide bandwidth that can react to some DAC's ultrasonic garbage (like > 1000 kHz). In my case, I also initially thought there's a problem with the scaler, but adding a passive RC filter (b/w DUT and scaler) resolved the issue. Of course, this is necessary only for certain DACs. In your case, the scaler might be defective, though. You just need to rule out possibilities.

 

[nanook]

The Scaler adjusts its gain according to the signal in the audible frequency range.
Did you try to manually set it to 0dB gain (gain poti fully left)
Do both channels of the scaler behave similar?

Cosmos Scaler | index

e1dashz.wixsite.com

16 positions controllable by the knob, in the "Auto" position Scalerr works like a ultrafast autoranger(100mS) with the left channel target 4.5Vrms.
The right channel output voltage is ignored. 20Hz-20kHz is the valid requency range for autoranger.

 

[nick_l44.1]

Any suggestions are welcomed.

First, it is always worth minimizing the signal and monitoring the signal voltage in the path. If you set the frequency to 50/60 Hz, the voltage can be easily measured with any multimeter. Clipping usually only occurs when the voltage increases. The scaler has two channels, and it is unlikely that they will both malfunction in the same way. How you connected the D50s RCA and scaler TRS and how you power it may also be important.

 

[IVX]

Hello all. Recently I tried measuring a device and it looks like my scaler is acting up. I have very high THD when I connect the DUT to the Cosmos ADC via the scaler. Below you can see what I mean.

View attachment 488135

This is when I measure my D50s using the scaler. If I connect the D50s output directly to the inputs of the ADC, I get a THD+N of -110dB but when I use the sclaer, I get what you see above. I tried several cables and several gain settings but the THD looks exactly the same. Has anybody seen this? This looks to me like one of the op-amps might be damaged. Most probably the OPA1656 that sits on the input. I do have several OPA1612 around but not the 1656. I'll order a few and I will try replacing them and I'll report back. Any suggestions are welcomed.

As I understand, your Scaler had worked properly, but now it has a lot of distortions? In that case is most probable that the 10ohm 0603 input resistors were burned.

 

[horias2000]

As I understand, your Scaler had worked properly, but now it has a lot of distortions? In that case is most probable that the 10ohm 0603 input resistors were burned.

Yes, it worked just fine for a couple of years. So the unit was good. I didn't use it for a while and now I wanted to measure a power amplifier I made and I had huge THD. I was pretty careful not to connect the measurement chain to the amp when the amp was powered ON and OFF in order to avoid the transients that happen at start-up and power down. I'll check the input resistors and I'll get back with the results. Thanks for the input.

 

[horias2000]

The Scaler adjusts its gain according to the signal in the audible frequency range.
Did you try to manually set it to 0dB gain (gain poti fully left)
Do both channels of the scaler behave similar?

Cosmos Scaler | index

e1dashz.wixsite.com

16 positions controllable by the knob, in the "Auto" position Scalerr works like a ultrafast autoranger(100mS) with the left channel target 4.5Vrms.
The right channel output voltage is ignored. 20Hz-20kHz is the valid requency range for autoranger.

Yes, I did try at other gain settings. I went through all the gain settings that are available and I also tried with very small amplitude input signal and the results were the same. Yes, both channels suffer from the same behavior. What I did notice, and I thought it was strange, is that sometime, when the scaler was set to "auto", the relays kept clicking as it was not able to set to a certain gain. Turning the input signal ON/OFF solved this.

 

[horias2000]

First, it is always worth minimizing the signal and monitoring the signal voltage in the path. If you set the frequency to 50/60 Hz, the voltage can be easily measured with any multimeter. Clipping usually only occurs when the voltage increases. The scaler has two channels, and it is unlikely that they will both malfunction in the same way. How you connected the D50s RCA and scaler TRS and how you power it may also be important.

I will check this as well. Usually, I have no issues measuring at 1kHz with the Keysight 34461A. Both channels seem to have the same issue.

As I understand, your Scaler had worked properly, but now it has a lot of distortions? In that case is most probable that the 10ohm 0603 input resistors were burned.

I checked the 0603 resistors and both are 10ohm (R42 and R51). I will check with another source just to make sure it's not the D50s that is acting up but I remember I measured it with this measurement chain a while ago.

 

[horias2000]

I measured the SMSL SU9-n DAC that I use in the main system and the results are the same. A lot of THD, even at low input signal level. The DAC volume was at 50 and the gain in REW was set to -7dB.

This is the measurement using the scaler:

And this is with the DAC directly connected to the Cosmos ADC (set to 4.5V):

It's worth mentioning that in the second measurement I increased the volume from the DAC to 99 and the gain in REW was set to 0dB.
I will order a few op-amps from Mouser and I'll replace them to see if the issue is resolved.

I checked both channels of the scaler and both behave in the same way. The Scaler is connected to the DUT via an RCA to 3.5mm jack and the scaler is connected to the ADC with two jack to XLR connectors. I use two connectors as the ADC is in mono mode.

@IVX I also checked the supply rails and both are OK at +/-15Vdc

 

[MaxwellsEq]

now I wanted to measure a power amplifier I made and I had huge THD.

Was the 4 or 8 Ohm load you were testing with purely resistive? Also is the scaler necessary for measuring power amplifiers?

 

[horias2000]

Was the 4 or 8 Ohm load you were testing with purely resistive?

I use an 8ohm 50W wire wound resistor. So, yes, mostly resistive load.

 

[IVX]

if all 0603 10ohm are ok, I would take off the IC10, it is part of the input opamp(1612) protection, 5A N/P BJT pair.

 

[horias2000]

if all 0603 10ohm are ok, I would take off the IC10, it is part of the input opamp(1612) protection, 5A N/P BJT pair.
View attachment 488361

OK, I will try removing IC10. I measured all 0603 resistors and two of them are not 10ohm. The ones close and perpendicular to the MELF resistors are around 6K. I can't tell the designator (see below circled in red). Not sure if those should also be 10ohm or not.

 

[IVX]

wery well, you need to replace these burned 10ohm resistors, and think about how to avoid input overvoltage in the future.

 

[horias2000]

wery well, you need to replace these burned 10ohm resistors, and think about how to avoid input overvoltage in the future.

If these are on the input, can I presume these are thin film as to not generate noise?

 

[IVX]

10ohm and noise? Come on.. any thik-film is ok.

 

[horias2000]

10ohm and noise? Come on.. any thik-film is ok.

true