DAC AD1862: Almost THT, I2S input, NOS, R-2R

@Hidy did some measurements #2823 ... I am not sure if adding analog post filter can improve something, maybe 🤔
The input digital filter can give a nicer measurement for one frequency - good for testing, but many listening tests confirmed, that finding a good digital filter is difficult and ofter it is better without :tilt:
 
Great to see some other measurements - thanks for sharing! I’m going to make a new match shortly and optimize a few things. I want to squash as much as I can, verify a 0db test tone, and match ADC sample to test sample.

Hope to share within the hour.
 
Ok, just made a couple more measurements based on the tips and suggestions given over the last several posts.

  • I turned off all unnecessary equipment and optimized test lead runs
  • Verified the full-size test tone

Measurement 001 was sent a 44.1khz test tone that I had made previously. Focusrite was set to sample at 44.1khz. Noise is definitely reduced, proving that noise in the first batch was largely self-induced.
Miro1862_FFT_001.png

Measurement 002 was sent a 96khz OS test tone generated from the REW signal generator and saved as a WAV. Focusrite was set to sample at 96khz. I felt that there was enough variation in the processes and test tone origin that it didn't make sense to include distortion numbers and invite a direct comparison. My purpose here was to see of the spikes around 45khz in the previous measurements were testing error or part of the DAC output. Appears to be the former.
Miro1862_FFT_002.png
 
Hello,
Each time I read this thread I’m reminded how little I really know about electronics. That’s not necessarily a bad thing but I tend to get very cautious as I don’t want to make mistakes. I’m getting close to finishing my dac and hopefully some here will help me get it done. Please take a look at the pictures and check for mistakes. Specifically the location of all the jumpers. I’m using the jlsounds pcb not flipped soldered directly to the dac pcb. The jlsounds pcb calls for a a jumper on H5 at J4, is a wire jumper ok? Is there a need for resistors at A1 and A2 on the back of the pcb? What about SJ3 , a resistor there?I’m going with a psu1 , please also check the wiring from it to the dac.
What am I missing? What other details?
Thanks so very much
Also, I intend to solder most components to the back side of the dac to allow more room around the dac and opamp ic’s
 

Attachments

  • 8E8329BF-D844-42F3-A8A9-E7E77BF0563C.jpeg
    8E8329BF-D844-42F3-A8A9-E7E77BF0563C.jpeg
    453.8 KB · Views: 182
  • FA637655-81DA-4BC0-B38F-30972CA280BB.jpg
    FA637655-81DA-4BC0-B38F-30972CA280BB.jpg
    108.8 KB · Views: 171
  • 904C6029-5D21-4020-83FC-B9463DB1754A.jpeg
    904C6029-5D21-4020-83FC-B9463DB1754A.jpeg
    287.4 KB · Views: 183
Last edited:
I've built at least 10 boards by now and each time before I plug in chips I verify correct power at the power supply, correct + to + and - to - on the power connections, then power up the board and test voltages at every pin on the DAC and op amp sockets. Haven't had anything too wrong since the first one where I reversed an electrolytic cap (bang!) but seeing the smoke come out of an 1862 would be very sad.

IMO you should connect the ground wires from DAC to power supply.
 
  • Like
Reactions: miro1360 and Jcris
What about PC oversampling and filtering? There are powerful and adaptable software options that might be better then ICs?

From some point it still be about patterns. When something is missing in the image, it is difficult to reconstruct it (or guess it) for any software. The result may be something that does not belong there (then it's better when the information not there at all and is smoothed).
 
When something is missing in the image, it is difficult to reconstruct it (or guess it) for any software.

Good point.

Some would say the filter isn't about trying to recreate something missing but suppressing noise, both ultrasonic and in the audible range.

Th cool thing about software filters is they are so easy to try. Some of the best are even free. Instead of spending hours soldering its just a few minutes investment to see how the sound changes.
And now corrected

.... everything is more correct with a satndard poodle looking on. Gorgeous dog.

1645572926871.png
 
I finally got around to installing and playing with the MSB adjustment. I can definitely confirm this is doable while taking a real-time FFT. Definitely cleaned up some of my noise and higher order harmonics. Kinda wish I'd taken a screen grab video of the adjustment.

Miro1862_FFT_MSB_003.png
 
  • Like
Reactions: Painkiller