E1DA Cosmos tests

[horias2000]

Hey all and a Happy New Year!

I've been playing around with the E1DA Cosmos and I want o share some results. I think that the Cosmos is a great tool and together with REW and a good DAC, it makes a great amplifier/DAC/preamp measuring tool-chain.

I measured my D50s and my SMSL SU9-n DACs and the THD numbers match with what Amir measured but the THD+N numbers are way below what Amir had in his posts.

Below you can see the measurement for the D50s. The second harmonic is just below -120dB and this is similar to what Amir had. The THD measurement in the upper left box says -116.9 dB and this seems OK by looking at the graph itself but the THD+N measurement of -95.8 dB does not seem correct. BY looking at the graph, it should be somewhere around -110dB. The nise floor of the measurement is similar to what Amir showed in his review.

Next I measured the SU9-n (below) and the THD was at an incredible -129 dB, again, very similar to what Amir was showing, but again, the THD+N measurement is around -95 dB and by looking at the graph, there is indication it shouldn't be anywhere close to -120 dB.

Overall, both DACs measure incredibly well and the Cosmos is a really great ADC. Is there something I'm missing regarding the THD+N measurement? It looks like REW is not able to measure below -95 dB.

For measuring, I powered the D50s from a battery pack and the Cosmos from another battery pack. Measurements were performed using a laptop n battery power. For the SU9-n, I obvirously used it's internal power supply powered from 230Vac.

 

[pkane]

Hey all and a Happy New Year!

I've been playing around with the E1DA Cosmos and I want o share some results. I think that the Cosmos is a great tool and together with REW and a good DAC, it makes a great amplifier/DAC/preamp measuring tool-chain.

I measured my D50s and my SMSL SU9-n DACs and the THD numbers match with what Amir measured but the THD+N numbers are way below what Amir had in his posts.

Below you can see the measurement for the D50s. The second harmonic is just below -120dB and this is similar to what Amir had. The THD measurement in the upper left box says -116.9 dB and this seems OK by looking at the graph itself but the THD+N measurement of -95.8 dB does not seem correct. BY looking at the graph, it should be somewhere around -110dB. The nise floor of the measurement is similar to what Amir showed in his review.
View attachment 339278


Next I measured the SU9-n (below) and the THD was at an incredible -129 dB, again, very similar to what Amir was showing, but again, the THD+N measurement is around -95 dB and by looking at the graph, there is indication it shouldn't be anywhere close to -120 dB.
View attachment 339284

Overall, both DACs measure incredibly well and the Cosmos is a really great ADC. Is there something I'm missing regarding the THD+N measurement? It looks like REW is not able to measure below -95 dB.

For measuring, I powered the D50s from a battery pack and the Cosmos from another battery pack. Measurements were performed using a laptop n battery power. For the SU9-n, I obvirously used it's internal power supply powered from 230Vac.

You can't directly compare noise floors by looking at FFT, especially those produced with different size FFT. Larger FFT size lowers the apparent noise floor, this is known as FFT gain. Also, your chart is not scaled to 0dBFS, so the appearance of the noise floor is about 8 - 10dB below where it should be compared to Amir's, even with the same FFT size. Try using dBc units.

All that said, the noise (N) parameter appears high, so something isn't right with your setup or configuration. First thing I'd check is that you are using more than 16 bits in your audio driver/REW chain configuration. Try using the ASIO drivers to make sure.

 

[horias2000]

You can't directly compare noise floors by looking at FFT, especially those produced with different size FFT. Larger FFT size lowers the apparent noise floor, this is known as FFT gain. Also, your chart is not scaled to 0dBFS, so the appearance of the noise floor is about 8 - 10dB below where it should be compared to Amir's, even with the same FFT size. Try using dBc units.

All that said, the noise (N) parameter appears high, so something isn't right with your setup or configuration. First thing I'd check is that you are using more than 16 bits in your audio driver/REW chain configuration. Try using the ASIO drivers to make sure.

I did use dBC for the D50s but not for the SU9-n. I will check that I use more than 16 bit and I will try to use ASIO as well.
Thanks!

 

[horias2000]

I used FlexASIO and I unchecked the "use dithering" checkbox from the signal generator window (this made a difference) and the THD+N measurement improved with about 7-8 dB to -103.6 dB. It's still around 6 dB lower than what Amir had but I think this is the limit. I see there is some sort of noise in the low frequency region. I will try with another cable to see if there is any improvement. Another thing worth mentioning is that the Cosmos is in "STEREO" mode. I will make a cable that connects the L and R inputs together and I will use the Cosmos ADC in "MONO" mode for the best results.

 

[mdsimon2]

You should adjust the input voltage setting on the Cosmos so you are as close to 0 dBFS as possible, your current measurements are severely limited from a noise perspective as you are nowhere close to 0 dBFS.

Michael

 

[horias2000]

You should adjust the input voltage setting on the Cosmos so you are as close to 0 dBFS as possible, your current measurements are severely limited from a noise perspective as you are nowhere close to 0 dBFS.

Michael

At the moment it is set to 4.5V as I measured the SU9-n on the XLR outputs and it outputs 4Vrms. I will try to lower the input voltage setting and retest.

 

[mdsimon2]

Depending on the DAC output impedance you might be able to get away with using the 3.5 V input setting, but give how far away you are from 0 dBFS at the 4.5 V I think it is more likely you have attenuation somewhere in your measurement chain. I assume you have the generator set to 0 dBFS and the DAC volume control maxed?

Michael

 

[horias2000]

Depending on the DAC output impedance you might be able to get away with using the 3.5 V input setting, but give how far away you are from 0 dBFS at the 4.5 V I think it is more likely you have attenuation somewhere in your measurement chain. I assume you have the generator set to 0 dBFS and the DAC volume control maxed?

Michael

I changed the input voltage of the Cosmos to 2.7V and I increased the output of the DAC to 2.2V and this gave me the best result, a THD+N of about -106 dB. I guess this is pretty much as far as it can go. I still want to try with the MONO setting on the ADC and using the proper XLR cable that connects both inputs in parallel. Maybe I will get a few more dB and I will get even closer to the -110 dBs that Amir had.

 

[mdsimon2]

Which DAC are you using for this measurement? Either way, at -13 dBFS you are still a long way off of 0 dBFS on the Cosmos ADC and could still lower the input voltage setting.

In general, to get the best THD+N measurement you want your DAC at full output, or maybe -1 dB below full output. You say you are at 2.2 V, but have you confirmed this? You could play a 50 Hz tone and measure the voltage with a multimeter as a check.

You should have no issue replicating -110 dB THD+N at 2 V on the D50S, even in stereo mode. On the SU9-n you may struggle to replicate -121 THD+N at 4 V in stereo mode, but you should be able to get at least -116 dB THD+N in stereo mode. Going to mono mode will get nothing on the D50S and only a dB or two on the SU9-n.

Michael

 

[pkane]

I changed the input voltage of the Cosmos to 2.7V and I increased the output of the DAC to 2.2V and this gave me the best result, a THD+N of about -106 dB. I guess this is pretty much as far as it can go. I still want to try with the MONO setting on the ADC and using the proper XLR cable that connects both inputs in parallel. Maybe I will get a few more dB and I will get even closer to the -110 dBs that Amir had.

View attachment 339623

Try capturing ADC output in REW with nothing connected to the inputs. You should see noise level at around -126dBA or better in mono mode:

 

[horias2000]

Which DAC are you using for this measurement? Either way, at -13 dBFS you are still a long way off of 0 dBFS on the Cosmos ADC and could still lower the input voltage setting.

In general, to get the best THD+N measurement you want your DAC at full output, or maybe -1 dB below full output. You say you are at 2.2 V, but have you confirmed this? You could play a 50 Hz tone and measure the voltage with a multimeter as a check.

You should have no issue replicating -110 dB THD+N at 2 V on the D50S, even in stereo mode. On the SU9-n you may struggle to replicate -121 THD+N at 4 V in stereo mode, but you should be able to get at least -116 dB THD+N in stereo mode. Going to mono mode will get nothing on the D50S and only a dB or two on the SU9-n.

Michael

I was using the D50s for this test. I understand what you are saying about the -13 dBFS as it confuses me as well. My guess is that there is something in REW that I'm not doing correctly but I can't figure it out at the moment. There is nothing that might attenuate the signal. The RCA output of the DAC goes to the + of the XLR connector of the ADC. The only attenuator is the resistance the Cosmos has based on the input voltage selection. For an input voltage of 2.7V the ADC has a 1k resistor and for 4.5V it's 1.6k. There is a note on the official COsmos ADC webiste staing "Most FFT software may correctly show the levels of the harmonics but incorrectly calculate THD+N etc be aware.". I was thinking it might be this but I'm not 100% sure. In any case, thanks a lot for the support so far . Highly appreciated!

 

[horias2000]

Try capturing ADC output in REW with nothing connected to the inputs. You should see noise level at around -126dBA or better in mono mode:

View attachment 339632

I have about -128dB with no input on the left channel.

 

[McFly]

Comtrue USB driver?

 

[pkane]

I have about -128dB with no input on the left channel.

View attachment 339846

OK, that's normal and says that ADC is working fine and doesn't have any unusual self-noise. That must mean any high levels of noise are coming from the outside the ADC, i.e., the DAC or the connectors. How do you connect the DAC to the ADC when you measure in mono mode? What goes into the left and right inputs on the ADC?

 

[horias2000]

OK, that's normal and says that ADC is working fine and doesn't have any unusual self-noise. That must mean any high levels of noise are coming from the outside the ADC, i.e., the DAC or the connectors. How do you connect the DAC to the ADC when you measure in mono mode? What goes into the left and right inputs on the ADC?

At the moment I only use the stereo mode as I'm missing an XLR connector in order to make the L+R connection for the two input connectors. I'm using a cable I made, with an RCA plug on one end and an XLR male connector on the other end. The hot wire from the rca goes on the + of the XLR. For the SU9-n measurements, I used an XLR to XLR cable that I made as well. Good quality connectors and cable.

 

[sarieri]

Check your distortion setting. I remember there’s an option on AES17-2015 or something similar that can drastically improve the THD+N measure.

 

[hayabusaxps]

I ACTUALLY NEED TO GET MY COSMOS ADC grade 0 MEASURED.
im getting like 116 db on 1 channel input and 105 dB on its left channel input. while in stereo
and

i made a tutorial on mine but im getting lower measurements on all devices i measure.
any chance you have different results, i zipped my email to cosmos


im benchmarking a DX7 pro Plus
this is what i got for example

View attachment 345840

do you think running a band pass on 20-20khz will change the readings/calculations?

and shouldn't my 2nd,3rd, harmonics be closer to 116 instead of -102 for a DX7pr plus





and does anyone know who i could reach out to to get my COSMOS benchmarked?

 

[hayabusaxps]

also do you notice a 0.5 V drop on the DAC output voltage when connecting the cosmos?

that seems a bit high of a voltage drop. makes me wonder how much of a current draw the input adc is pulling.

 

[hayabusaxps]

even right now, benching something supposed to have a 106 dB im getting more like 85 dB

with a Volt meter showing 3.98Vrms on the +&- balanced XLR

and a DSP function generator is creating the 500hz tone at 3.95 V

64K FFT

4M FFT

64FFT mono setting

with the cosmos ADC open ended or disconnected from XLR input.
you said i should be seeing -128 dBfs for open. on mono
im seeing -125 dbFS A on a stereo (grade 0 ADC) was supposed to be 127 technically if im analyzing it right.

and mono its still -125.5 dbFS A


mogami 2549 shielded cables and XLR terminals i made them.
i dont think you need to add a 50ohm resistor or anything though......

 

[mcdn]

Couple of things:
* change your FFT length to 256k instead of 1M
* change your window to Blackman-Harris 7
* change your averages to 2 or none
* select “use 64 bit FFT”

Also power your Cosmos ADC from a battery power bank, not the wall. When measuring down at these tiny levels it’s really important!