View RSS Feed


Preliminary measurement results for the DSC1 DAC

Rate this Entry
I finally used couple of days to do small fixes for the first unit I put together and got first decent measurement results.

Results are still considered preliminary, because the DAC was measured without any shielding case, plain boards and transformers on a table, connected with open unshielded wire hassle...

I performed my usual set of measurements.

First was 0 - 22.05 kHz linear frequency sweep (44.1 kHz 32-bit PCM source file) with analyzer set to 5 MHz bandwidth and "peak hold" mode. This is a good measurement to see any digital or analog filter leakage as well as buildup of other ultrasonic noise.

Sweep with DSD128 output rate:

...and DSD256 output rate:

...and DSD512 output rate:

As can be seen, the ultrasonic noise bump is completely gone from the figure at 22.5792 MHz output rate, and very low indeed already at 11.2896 MHz. No correlated images visible at all, as those are completely removed by the oversampling digital filter to those high rates (analog filter cuts out everything way before reaching 11.2896 MHz).

Then the standard IMD measurement with 19+20 kHz tones:

Sine wave at 1 kHz:

THD is predominantly friendly sounding second harmonic.

24-bit J-test signal:

Nothing visible there...

Notes and conclusions:
  • Overall noise performance was better than expected, especially the audio band noise floor was lower than with couple of other DACs I compared against.
  • Low impedance (noise) design of analog stages is pushing the active parts a bit. DIstortion figures would go down with some amount of tweaking on the analog stages. AD797 could have been better choice for the filter stages based on earlier experience.
  • This one already sounds so good, that I probably need to make a DSC2 design this winter...


  1. Superdad's Avatar
    Looks nice Miska. It would be interesting to see a more expanded sweep down lower--say from 10kHz to 60kHz--to get a better picture of the just out-of-band response. At DSD128 and 256 only please.

    DSC2? So many possibilities… :)
  2. Miska's Avatar
    I made a new blog post of those results, because I cannot upload images to blog post replies...
  3. lkong's Avatar
    Im in progress of building my own dsc1.
    I was wondering if replacing the 2% resistor networks with 0.05% smd resistors would be helpful.
  4. Miska's Avatar
    At least it shouldn't hurt. I just didn't find anything better as through-hole components. Better tolerance has mostly impact on accuracy of conversion stage frequency response.

    I have made couple of fixes to the design that can be retrofitted on the original board design. I will make fixed schematic and board files as well as errata available as time permits.

    Changes are:
    1) Either change I/V feedback resistor R1/R2 to 1 kOhm (this is what I did), or drop resistance of the network (8k network would be optimal).
    2) IC11/IC12 non-inverting input pin 3 should be the point where DGND and AGND are connected, I added jump wires to the pin.
    3) IC11/IC12 output pin 6 should be connected to AGND with 100k resistor, I connected from C520/C530 input side.
  5. gionaz's Avatar
    if you have one available, I'll buy it now