Abstract

In order to check a reproducibility of DAC noise situation, I measured DAC noise on standalone system.
DAC noise level without de-whitening was reproduced on the standalone.
In addition to poor improvement of DAC noise by de-whitening filter, poor improvement of ADC noise by whitening filter was also seen.
In any case, measurement system and circuit noise should be evaluate more carefully.
 

Details

Because DAC noise check in the mine has lots of conflicts with other commissioning works, I constructed the measurement environment for DAC noise on the standalone system and then I measured DAC noise by a same manner as klog#24725. For some reason, the model file prepared for the circuit measurement was corrupted, and it took a long time to recover it, so I measured the noise only with zero outputs today. (There should be no difference between zero output and non-zero output. See also klog#31738.)

ADC noise (ADC + AA chassis + whitening filter chassis) without whitening filters and gains for the channel used to measure DAC noise is shown as brown curve in Fig.1. I was afraid that noise level on standalone was worse than one in mine, but fortunately it's almost same as theoretical level. In the past works in the mine, we estimated the whitened ADC noise level by using raw ADC noise level and the design transfer function of whitening filters. On standalone system, there is no concern of time sharing with commissioning, I decided to check also ADC noise level with various whitening configuration which are reference for the DAC noise measurements.

Green curve represents the ADC noise with 36dB whitening gain. It's ~60 times better than brown curve. So 36dB whitening seems to work well. But black curve which is measured with 36dB gain + 1 stage of whitening doesn't have a ~10 times improvement from green curve above 10Hz. And also, gray curve which is measured with 36dB gain + 2 stages of whitening doesn't have any improvement from black curve. These two measurements suggest that the whitening filter does not improve the ADC noise level. There may be some noise source upstream of whitening filter stages. If a same issue occurred when we measured in the mine, noise from the upstream of whitening filter may be one of the reasons why improvement by de-whitening is smaller than theoretical one. But if so, this fact means that noise of whitening filter chassis with large gain becomes larger than ADC noise. I need to ask AEL folks this fact tomorrow. By the way, during these two measurements, 1-4 pins of whitening filter input are shorten to 6-9 pins, respectively.

DAC noise and DAC + HPCD noise with 36dB whitening gain are shown as magenta and orange curves, respectively. Both of them are larger than the ADC noise level with 36dB gain (green curve), so the noise of DAC and HPCD seems to be seen properly. And also HPCD noise seems to be smaller than DAC noise because these two curve are almost same level. DAC noise measured in this time is same level as measurements in klog#31738. So there is no problem on reproducibility.

The situation that improvement by de-whitening filter is smaller than expected is also reproduced unfortunately. Blue and red curves are represents the DAC+HPCD noise with 1 stage and 2 stages of de-whitening filter, respectively (configuration of whitening filter is 36dB gain + 2 stages of whitening). These two curves are more than sqrt(2) times larger than the gray line, so the noise coming from the whitening filter doesn't seem to be a dominant noise but it may not be negligible. DAC noise should be ~10 times smaller than magenta curve. So the reason why blue and red curves are larger than gray curve seems to come from the downstream of de-whitening filter stages.

Other channels shows similar results as shown in Fig.2. So channel#0 is not a special situation. This issue is reproduced multiple channels.

Noise of measurement system (ADC + AA + WF) is now very close to the DAC (+HPCD) noise. So it might be better to improve the system noise at first.