• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Best spec ADC Chip currently.... ??

Wow this starts to be like the equipment at work, switch on and wait 2-3 hours until it stabilizes :)
 
Wow this starts to be like the equipment at work, switch on and wait 2-3 hours until it stabilizes :)
I mean, we are talking insane measurement-grade performance here, so yeah.

50°C sounds like a reasonable enough temperature. The idea of heating ICs actually is anything but new, I am reminded of the TCA530 voltage regulator IC with built-in thermostat heater from the 1970s that was quite popular for generating a nice and stable tuning voltage (FM tuners with varactor-tuned frontends as common in Europe) but turned out to be a bit of a reliability fiasco long-term. The part did apparently manage to achieve a tempco of -1 ppm/K once warmed up though.
 
I had the idea to use ES9822 GPIOs current to heat it up but finally took the cheap MCU for that.
2021-12-08_22-21-44.jpg


2021-12-08_22-23-17.jpg
 
I juste measured my E1DA Cosmos ADC 'A-Greade' with REW :D using SMSL M400 (AK4499EQ) as a signal source. In mono mode it shows THD+N below -121dB! Worst harmonics stays around -140dB o_O To be honest, this is first time I see such a values at home. Thank You Ivan :D

View attachment 173192

Wow! So you used the M400 as a signal generator with no additional filtering? You used the Cosmos ADC in mono mode but measured only one stereo channel of the M400? Did you select for the better channel? I noted that in Amir's review of the M400, one channels was considerably worse than you results. The other was kind of hard to see but may have been in line with your results. By the way, was in Stuttgart for Xmas shopping yesterday just when your message popped up.
 
ChUmI, well, I do design only what I like to use. Nobody design a decent ADC for audio measurement for $100-200 , hence, I have to do that ;)
BTW, not all 100% of Cosmos ADC has 3rd harmonic as low as -140db. Average is -138db, worst ADC unit had -134db, the best one -145db or less but it is rare.
Be aware, that 3rd harmonic noticeable depends on the ADC temperature, so try to keep it warm(I calibrate ADCs at 27C). Very soon I'll offer a tiny thermostat PCBA for ES9822 chip, which will keep the ADC chip temperature around 50C, if the ambient is too cold. The PCBA is 5x5mm heater with a thermometer onboard. To install the thermostat need to glue that on the top of ES9822 IC, and connect two wires 5VDC and GND.
View attachment 173194View attachment 173195
Do you calibrate at 27°C of room temperature with the aluminum case of the Cosmos ADC being pretty much at 27°C? This seems pretty hot, but may be customary where you are. How hot does the ADC chip's case surface get when the aluminum case is at 27°C?

If 27°C case means 45°C for the ADC chip, the noise penalty of going to 50°C will probably be small. If it means taking the chip from 30 to 50°C, maybe significant. I assume you would need to recalibrate the ADC for distortion either way when a heater is retrofitted?

I was wondering if it would be better to use a peltier element to cool the ADC chip to e.g. 10°C. This should give you better noise performance and leaves enough of a temperature difference to uncooled operation for the control loop to work. Of course, you would want to connect the other side of the peltier to the aluminum box thermally which might not be an easy thing to do.
 
Hi @IVX any plans for an unbalaced simple USB ACD? I bet there is a market for it.
People is using musicians interfaces with lots of innecesary functionality to just digitalize their records or stuff like that, but simple, small and decent ADCs aren't common.
Definitely a niche product but you would be the only one in the market i think...
 
Hi @IVX any plans for an unbalaced simple USB ACD? I bet there is a market for it.
People is using musicians interfaces with lots of innecesary functionality to just digitalize their records or stuff like that, but simple, small and decent ADCs aren't common.
Definitely a niche product but you would be the only one in the market i think...
You can plug unbalanced sources into the Cosmos pretty easily.
Here's the cable layout for stereo recording:
2021-11-08_11-09-17-1-1.jpg
And for Mono:
unknown-3.png

Furthermore, I don't know how Ivan could simplify the Cosmos any further.
It has analog in, digital out, and switches to control gain.
 
  • Like
Reactions: MCH
Hi @IVX any plans for an unbalaced simple USB ACD? I bet there is a market for it.
People is using musicians interfaces with lots of innecesary functionality to just digitalize their records or stuff like that, but simple, small and decent ADCs aren't common.
Definitely a niche product but you would be the only one in the market i think...
you can consider Cosmos ADC as unbalanced with no mistake, there is FDA input, hence, doesn't matter balanced or unbalanced source you need to convert. I don't believe if someone may complain that Cosmos ADC is a too big device(103x61x39mm .15kG). Or you are talking about a dongle-sized ADC??
 
Ah i didn't know you can plug unbalanced sources to the Cosmos. Yeah if so, that's exactly the device i was missing :D
Thanks!
 
capslock, the temperature of the thermostat selected to have no problem with recalibration. Just need to glue that to the ADC chip and tie 2 wires.
ES9822 uses so-called polyresistors which have minimal nonlinearity at such a temperature. 10C will create a lot more distortions.
Ok, from this I gather that at 27°C box temperature the ADC chip's case lives near 50°C already so that taking it up by a few degrees won't matter for calibration (probably not true, see addendum).

Yes, of course cooling the ADC chip will change the calibration, especially if it changes the die temperature more than heating to 50°C case temperature would. But unless the internal circuitry is optimized to work at an estimated 60°C die temperature, this should not hurt achievable distortion while helping with already excellent noise performance.

Addendum:

The data sheet

tells us typical power consumption is 4.5 V * 26.5 mA + 3.3 V * 15.2 mA = 169.41 mW.

Test conditions are Ta = 25°C. It does not contain graphs of performance vs. temperature, nor does it tell us the thermal resistance. For this chip in the same 40 QFN package,

we get about 23 K/W junction to ambient (about 13 junction to case and about 8.5 junction to board). So without the heater, the die will be a maximum of 4 K hotter than ambient. Assuming there are not a lot of other components generating massive heat inside the box, the uncooled ADC die will be at about 30 - 34° when the box is at 27°C.

The heater forcing the case to 50°C will result in the die sitting at about 52.5°, so we get about 20°C vs. unheated. Unless the temperature coefficient gets much worse below 25°C, cooling the case to e.g. 10°C with the die ending up at 12.5°C would result in the same 20°C delta, just with opposite sign.

If there is no noise benefit from active cooling, I would humbly propose controlling the whole aluminum box to 27°C. This would minimize thermal gradients between the ADC and other analog circuitry, and it would keep ADC, op-amps and resistors at their calibration temperatue.
 
I juste measured my E1DA Cosmos ADC 'A-Greade' with REW :D using SMSL M400 (AK4499EQ) as a signal source. In mono mode it shows THD+N below -121dB! Worst harmonics stays around -140dB o_O To be honest, this is first time I see such a values at home. Thank You Ivan :D

View attachment 173192
Looks great!

In my loopback test (Topping E50 into B-grade E1DA Cosmos ADC) the 3rd harmonic seems to dominate:
Topping E50 DAC (0 dBFS) into E1DA Cosmos ADC (4,5V sensitivity) - Balanced, mono mode (in and...jpg

This is with the Cosmos warmed up above ambient temp as suggested by @IVX - with the ADC running colder the 3rd harmonic is 3-4dB worse:
Topping E50 DAC (0 dBFS) into E1DA Cosmos ADC (4,5V sensitivity) - Balanced, mono mode (in and...jpg


To see how much of the THD is actually coming from the ADC itself I also measured it with Victor's 1kHz ULD oscillator:
E1DA Cosmos ADC (4,5V sensitivity, balanced, mono mode) - THD measured with Victors 1kHz ULD o...jpg

THD+N is less precise here due to the increased noise and skirting of the analogue oscillator, but distortion figures are more accurate - we see that both 2nd and 3rd harmonic are ~138dB down vs the fundamental, and the rest are even further down - below -140 or even -150dB.

Pretty solid for a hobby audio measurement rig I would say! :D
 
Because it is normal to somewhat hot for homes in the US or Europe in winter?
 
Sure, I use a hifiberry beocreate DSP board (just because I had it lying around) which has an analog devices ADAU1451 DSP and decent power supplies onboard.

I connect the Cosmos with I2S to the DSP board (I abuse the RPI pins for this)

The DSP can have the Cosmos as clock master or can use onboard samplerate convertors.
The Board has spdif in TOSLINK out (only goes up to 192kHz but thats fine for me), which I connect to my RME babyface pro
Hi, missed you post some weeks ago. It looks like you succeeded with a raspberry board, well done!
I had no luck with two boards, but they are different than yours.
I now need a bit of free time to finish the other way I've chosen to add digital outputs
 
Grooved, to be honest, I don't care about the studio-production market, there is brand-name needed but not tech specs. If you are The Name, you can sell the same model for 15 years for $2000-4000/pcs with THD+N -115db, and Crosstalk -115db as a TOP of technology. I have sold about 2000pcs DACs with THD calibration ability, Cosmos ADC for that. I think if someone in a modern studio aren't stupid, he'll buy my Cosmos for 10% of The Name price, maybe 10-50pcs of ADCs would be sold there, why I should care?

You have your business math wrong. Your FOB cost for ADC goes up, units remain low. DAC is a mass market item with wide appeal, ADC is niche. Margin is low on DAC, high on ADC. Also, in business, 80% of revenue is derived from 20% of product. So to maintain revenue streams, creating new SKUs is part of business.

You are actually witnessing this very thing with "THE NAME." So sell a new calibration ADC for $1500pcs which has better specs than THE NAME, also sell audio ADC to professional market. With time, YOU are THE NAME and charge $2-4000/pcs.

For DAC, RME is THE NAME. Topping is THE NAME.
For ADC, AP is THE NAME.

Also for instance. You see "high performance" air intake filter in automotive market. These are made by companies who are specialist in cotton production. Their big market are home air conditioner units. Making filters for Corvettes is just a niche market, but they'll charge hundreds of dollars for a plastic pipe with a filter on the end.

I think your approach and logic is too literal in this regard. e.g. Worrying about the dynamic range of microphone, when this isn't important to most who purchase professional audio products. You already have people contacting you interested in your converter design and turn them away because you are too focused on design.

Business managers make poor engineers. And engineers make poor business managers. Me thinks you should hire a business manager in your country Ivan!
 
...

Business managers make poor engineers. And engineers make poor business managers. Me thinks you should hire a business manager in your country Ivan!

At the same time, I totally understand Ivan on one point I see : he already build several DAC, but it's the ADC is launching, and having make a choice on how the product should be, now that he has launched it and has some complementary products coming, he has to finish that first.

Being on both sides of the job is not easy, and I would not call it exactly "Business managers make poor engineers. And engineers make poor business managers", because without generalize, and at my level, I already had to care of both sides, and with years of hindsight, I know that I was better at each when doing only one thing than when I was doing both.
So I would more view this situation like "You can't be good at both at the same time", and it's hard to promote a product the best way when you're the one who got the idea and build it. Not that you can't but it's harder.

And as there are far more audio ADC than measurement ADC in the market, it would certainly be harder and he may be right on the NAME part.
 
Back
Top Bottom