I have different experience, moreover, if you pull out the male jack unexpectedly it will short the amplifier output to ADC ground (during moving inside the female connector) which is not welcome. I believe you in digital circuits design, you should believe me in amplifier testing.
Well, finally SMT is started and I brought home 1 panel(8 PCBA) for the test.
No mistakes, no problems. 2pcs exhibit H3 -160db@1kHz@4.5Vrms, one -149db, others -152-155db.
H2 -152-155db(maybe lower, I didn't spend enough time to compensate H2 <-155db) but one sample has quite high H2 -141db, so I put it in the drawer to investigate its unsymmetry later. Also, need to think about the rate of such unsymmetrical units, I really want to avoid testing each unit with -160db harmonics setup, it is crazy hard work..
All these units Sunny sold out in China so far, in the first minutes. Next few days I'll get bigger butch of assembled PCBA.
No mistakes, no problems. 2pcs exhibit H3 -160db@1kHz@4.5Vrms, one -149db, others -152-155db.
H2 -152-155db(maybe lower, I didn't spend enough time to compensate H2 <-155db) but one sample has quite high H2 -141db, so I put it in the drawer to investigate its unsymmetry later. Also, need to think about the rate of such unsymmetrical units, I really want to avoid testing each unit with -160db harmonics setup, it is crazy hard work..
All these units Sunny sold out in China so far, in the first minutes. Next few days I'll get bigger butch of assembled PCBA.
Ok. The Scaler is ready for production, however, after 8 recursions of the PCB layout, I can't get the crosstalk better than -127db@1kHz for the L2R, and -137db for R2L. The problem is too close relay's stereo groups of contacts + a slightly unsymmetrical L2R layout. I asked our discord people about should I stop my PCB layout polishing or keep it up to the <-130db result for both channels.. They said -127db is about Ok, so I decide to think that Scaler is ready.
The distortions are unmeasurable at 1kHz up to the gain = +15db or so. The residual noise is 250nVrms(A).
BTW, about the noise, in our discord I've got the question about the broadband noise FFT for a combo Cosmos ADC+APU, and the result looks like that:
As you can see, THAT1510 1nV/Hz^.5 is really there, and 1/f corner is about 100Hz. Better to see that with AVG(please note, the FFT above is for another APU sample, the next two FFT are from a bit less noisy sample):
The distortions are unmeasurable at 1kHz up to the gain = +15db or so. The residual noise is 250nVrms(A).
BTW, about the noise, in our discord I've got the question about the broadband noise FFT for a combo Cosmos ADC+APU, and the result looks like that:
As you can see, THAT1510 1nV/Hz^.5 is really there, and 1/f corner is about 100Hz. Better to see that with AVG(please note, the FFT above is for another APU sample, the next two FFT are from a bit less noisy sample):
Last edited:
Oh, I did add that function to the Cosmos Scaler in the last 2 versions, now Scaler works as a modest auto-ranger with coarse 16 gain steps yet. I made that clearly by the way because I had an idea of using a variable potentiometer for the gain setting but they are bad for THD or huge and bloody expensive. Hence, I did go with relay+MCU. The MCU has a lot of free resources and I decided to add 3 color LEDs indications of the output level. After that, the MCU still had resources and I did implement auto-ranger as well. The same day I got an idea to add a calibration procedure for the auto-aranger, so now you can set any of 3 thresholds for LEDs Blue(by default 1.7V)/Green(by default 4.5V)/Red(by default 10V), where Green is the auto-ranger target threshold. Simple solution and virtually free.We need autoranging input divider.
Ahh, Ok, it is L7 measurement/setup problem, Cosmos ADC has no such THD+N 40Hz bump at all. The corner from 200Hz .0001% to 20Hz .0003% is real, I'm not sure about the reason yet. I did adjust the lowest H3 at 1kHz -.5dbfs, you can try to readjust the lowest H3 at 20Hz instead, but pls do not lose the original H2/H3 Tweak_Cosmos app setting.
