Topping D10s + E1DA Cosmos ADC as a measuring system

[somebodyelse]

Unfortunately:

Update Jan 2023:​

At this point, I have no more autorangers available.. There are no plans for another batch.

https://linearaudio.nl/la-autoranger

 

[SIY]

OK thanks - I will investigate a bit more this evening.

So you are saying that the Focusrite Scarlett solo (or its bigger brothers?) + the Autoranger + REW etc would be good enough for some initial DAC measurements?

If you’re just doing DACs, you may not need the Autoranger. What it does is extend your capabilities to power amps and low level circuits. But again, it’s a matter of how low in distortion you’re trying to measure- good DACs will have distortion far lower than you can easily measure unless you’re willing to sink in a few thousand dollars. Frequency response is easy, though.

 

[SIY]

Unfortunately:

That’s a pity, and it makes life harder for hobbyists and small labs.

 

[Jeromeof]

If you’re just doing DACs, you may not need the Autoranger. What it does is extend your capabilities to power amps and low level circuits. But again, it’s a matter of how low in distortion you’re trying to measure- good DACs will have distortion far lower than you can easily measure unless you’re willing to sink in a few thousand dollars. Frequency response is easy, though.

Yes, well as per the other post the autoranger isn't available.

With DAC measurement I would like to see if some of my cheaper DAC's have better SINAD than say an Apple USB-C Dongle (which I believe 99 - I have an European one of those also I would like to measure).

I was under the impression I would need something considerable better to get a reasonable accurate value?

 

[SIY]

Yes, well as per the other post the autoranger isn't available.

With DAC measurement I would like to see if some of my cheaper DAC's have better SINAD than say an Apple USB-C Dongle (which I believe 99 - I have an European one of those also I would like to measure).

I was under the impression I would need something considerable better to get a reasonable accurate value?

Yeah, if you're trying to measure SINAD or distortion spectra at that level and want a reliable number, you're going to need something like the RME ADI-2 FS or better which is not cheap.

 

[pma]

E1DA Cosmos is good enough for the task. It is sometimes limited by its low input impedance. For the skilled DIYer, it is possible to design and build a stepped attenuator based divider and make the THD+N measurements vs. level per partes. But this would be rather a topic at diyaudio forum, here at ASR the audience is not that much oriented at DIY skills.

Focusrite 2i2 is not bad bad its use is highly limited, by its much higher noise (both DAC and ADC) and worse, bad linearity at high frequencies. It is usable at 1kHz, but not above 5kHz. It is also not usable for 19+20kHz intermodulation measurements.

With E1DA Cosmos ADC, one can get near -130dB THD and about -120dB THD+N in say 10dB range of levels. With switched input divider the dynamic range of reliable low level THD+N measurements is increased considerably. Autoranger would be the best solution, however all the effort would be spent to this and, still, the DAC/ADC is a tool and not the goal.

 

[KSTR]

Focusrite 2i2 is not bad bad its use is highly limited, by its much higher noise (both DAC and ADC) and worse, bad linearity at high frequencies. It is usable at 1kHz, but not above 5kHz. It is also not usable for 19+20kHz intermodulation measurements.

One can use the same trick that is used to measure microphone IMD, play 19kHz sine on one channel and 20kHz sine on the other and then mix them passively which does not create IMD products. Noise can be fought against by averaging (in the time-domain, preferably).
But in general, yes, the 2i2 is limited as is any other interface based on CS4272 and any time/money effort spent into workarounds should better go into a better interface to begin with.

 

[pma]

Some more measurements of D10s/Cosmos

 

[Dumdum]

So a few (lot of all around the same items) questions and something I’ve encountered

@pma and @IVX when configuring the e1da in mono how do you set this up? Is it in software somewhere or literally just using both inputs, I’m only measuring with one at the minute, I’ve also remade cables to have the sheild and cold tether at the rca plug rather than the small link I made in the xlr previously

I am using the isolator from olimex also

I was initially trying to use asio with the devices, and despite being able to select different devices for in and out I wasn’t able to get anywhere near the performance shown here, but I was using one input, using Java I can get better noise floor and therefore cleaner measurements (but maybe down to Blackman Harris 4 windowing so I need to revisit this) do you use asio4all in the set up part where you select inputs?

Can you take pictures of the settings in room eq wizard rta that enable you to get the excellent readings as I am 6db off with it connected as a loopback and I am also unable to get anywhere near 0dbfs from the topping (set to 0dbfs output) using a single stereo input to the cosmos, perhaps this is why… I have several xlr plugs and cable for splitters arriving soon

I get two unique readings for input 1 and two when I connect a splitter to both inputs, hence me asking about settings above

I figured out the Blackman Harris 4 and you use 8 averages

I was getting a noise on the signal which would ramp up the noise in the measurements ruining it… it turns out if you use a Lenovo laptop with ssd hard drive when it writes a good amount of data to it it makes some rf or effects grounding and makes noise go +30 or so db from where it should be… I switched off system noises and in the end listened to the other output of the topping to see if it was some kind of warning, it was very strange, and not until I watched the hard drive write speed pick up and coincide with the ‘noise’ that I worked out I needed to put the e1da and topping 18” away from the laptop and I could measure as normal once more

I also have an amplifier rig I was using with a Scarlett audio interface for larger voltages

It has a similar layout to your setup, but also a pair of 10uf foil caps on the input, unsure why these are required but it works and let me measure several car audio amplifiers

The e1da is for measuring dsp’s and the like as I have nothing bar one amp that came close to ruling out the use of the Scarlett

I have also measured some ‘very bad’ imo amps that shouldn’t be so bad for their cost, one is a rare class A that I suspect isn’t class A up to 30watts as they say and just has harmonics sprayed everywhere when measured over 0.5-1watt, very strange! But maybe down to my rig perhaps

I also have some very good measuring amps from a few manufacturers I expected to be decent…

Various attachments are attached with settings and measurements I’ve been able to acheive

Any help given is welcomed!

 

[Dumdum]

Already replied several times, mini jack is too tiny and there is a risk that it will make short-circuit of amplifier output, when accidentally pushed in or pulled out from the female mini jack of the Cosmos. Remember 40Vrms/4ohm makes 400W power and 10A current, and almost unlimited current when shorted, depending only on level of protection circuit design. One cannot rely on amp protection circuits. I speak based on experience, not on debate.

Some more photos:

View attachment 240738

View attachment 240739

View attachment 240740
Quality control should have been better. There is a lot of metal dust filings (see top right corner, near the screw), which might be potentially hazardous to a proper function. They may be easily removed by compressed air. It should have been done before shipping, @IVX .


And some more measurements Topping D10s/E1DA

View attachment 240741
Best Topping D10s SINAD ever, I would say.

View attachment 240742

So this is pretty decent considering I’m using a single input to the cosmos?

My noise issue still exists though… I have a video I’ll attempt to edit down to the essentials

 

[IVX]

Here is about mono/stereo and more https://e1dashz.wixsite.com/index/cosmos-adc

 

[Dumdum]

Here is about mono/stereo and more https://e1dashz.wixsite.com/index/cosmos-adc

I have read that, so if I’m correct as long as the volume slider in windows 11 sound is at full 100 level I will have mono mode when using just the left channel for rew and the two inputs in parallel

It would maybe be helpful to do a video to show this, and also maybe some indication of if the device is in mono mode somehow? Maybe a name change for device input?

It also states you can use asio, I have tried with asio4all and can’t get as good results vs Java with exclusive mode for the e1da and topping d10s as per the above picture… that’s in Java, from memory, if I use asio4all the thd was 0.0004 not 0.0002

I am still getting some kind of interference from what I think is the ssd in the pc

See the video if I can attach it

 

[IVX]

Cosmos ADC has a mono(blue)/stereo(green) indication on the front panel. To take a screenshot pls use Alt+PrintScreen.

 

[Dumdum]

Cosmos ADC has a mono(blue)/stereo(green) indication on the front panel. To take a screenshot pls use Alt+PrintScreen.

I will if it’s on my pc, unfortunately that was a snapshot from a video from my phone

Good to know re the led thank you!

 

[IVX]

also, to use the volume slider for mono/stereo pls install the Comtrue driver.

 

[staticV3]

I was initially trying to use asio with the devices, and despite being able to select different devices for in and out I wasn’t able to get anywhere near the performance shown here, but I was using one input, using Java I can get better noise floor and therefore cleaner measurements (but maybe down to Blackman Harris 4 windowing so I need to revisit this) do you use asio4all in the set up part where you select inputs?

Using REW, it doesn't get any better than Java EXCL.
So if that's working fine, then there's no need to mess around with different ASIO drivers.

Here's how Mono/Stereo signal routing works on the Cosmos ADC:

Using the default Microsoft device driver, you can change modes only using the Cosmos_Tweak utility which you can download from the product page.

By installing the Comtrue driver, you can then change modes using the Level slider in the Sound Control Panel as well.

The D10s can easily clip the Cosmos ADC, as long as everything's set up correctly and the Cosmos is set to 1.7V

 

[Dumdum]

Which comtrue driver is it of the choice of four on their site? I have downloaded them all but still have the cosmos in device manager with a yellow triangle exclamation mark (labelled as iAP interface) on it and non of the drivers change that when you run there installer packages, currently I can’t run the thd sweeps without the trace jumping up randomly for whatever reason that is beyond me which the makes my data look worthless to anyone reading

I am trying to measure a Dsp currently and it just makes a mess of the measurements, I have tried a different pc and that did the same also, it’s very frustrating when I managed to measure the topping at -112.6dbfs earlier (so pretty much as clean as it should be) but I had to catch it to do 32 clean averages without it jumping up

If I hadn’t turned all system sounds off you’d swear it was a beep or something, but I listened to the second output of the topping and it doesn’t make a sound

 

[Dumdum]

So I have a driver listed next to the cosmos from comtrue? Do I need to use the asio driver for the cosmos? Or can I still use Java? I presume the driver is good enough just being installed for the e1da

 

[staticV3]

Which comtrue driver is it of the choice of four on their site?

The one from the Cosmos ADC's product page: https://e1dashz.wixsite.com/index/cosmos-adc

I have downloaded them all but still have the cosmos in device manager with a yellow triangle exclamation mark (labelled as iAP interface) on it

That's normal. The iAP interface is some kind of workaround from Comtrue for Huawei devices and has nothing to do with the Cosmos ADC's operation.
It can safely be ignored.

 

[Dumdum]

I should add I don’t get the levels of precision the e1da offers but I can measure amps just fine with the focusrite and a divider network, no issues with bouncing traces at all

Wondering if I need a full speed isolator for the e1da perhaps, if anyone wishes to jump on my pc with teamviewer you are more than welcome as long as I have a bit of notice and can set the gear up… it happens with just the topping and e1da looped so no device under test in the loop

I think I have the driver installed correctly, but the yellow triangle persists for two iAP interfaces now I have my usb input to the Dsp connected