Abstract
According to recent activities related to the DAC noise estimation, output noise of high power coil driver (HPCD) and input noise of whitening filter (WF) for measuring DAC/HPCD noise seemed to be larger than DAC noise with the suppression by multiple stages of de-whitening filter.
So I tried to measure the HPCD with better sensitivity in order to estimate realistic output noise to coil-magnet actuators.
Engaging 1 de-whitening filter seems to reduce total output (DAC+AI+HPCD) noise as a factor of 0.1.
On the other hand engaging 2 de-whitening filters seems to be limited by HPCD noise though total noise can be reduced as a factor of 0.03~0.05.
Details
In the latest DAC noise measurement in the mine, we confirmed that DAC noise was 2~3e-7V/rHz (~4e-4 ct/rHz) even if DAC output was non-zero value [klog#31738]. Output noise of HPCD with old implementation of BIO control which engaged former stage first was 0.6~1.2e-7 V/rHz (1~2e-4 ct/rHz) [klog#31763]. After these work, possibility that changing engagement order of de-whitening filter was pointed out by Ushiba-kun [klog#31771] and some improvement could be seen by changing engagement order [klog#31774]. But we couldn't conclude the actual noise level of HPCD because measured noise after improvement was limited by the input noise of whitening filter as 2~3e-8 V/rHz (~4e-5 ct/rHz). So I tried to measure HPCD noise again with better sensitivity by using SR560 instead of WF.
All results are shown in Fig.1. Top panel show ADC+AA+WF noise (left) and ADC+AA+SR560 noise (right) with various configuration of whitening and pre-amp gain. In both cases input port was shorten by 0Ohm. In the case of using WF, measurements are limited by ~3e-8 V/rHz which is input noise of WF. On the other hand, using SR560 allows to measure with the level as ~5e-9 V/rHz. Note that 0.3Hz high pass filter (12dB/Oct.) were applied in order to avoid saturation around DC. This is the reason why noise level below 1Hz with SR560 looks better than one with WF. So please see only above 1Hz in these plots.
Bottom panels show the HPCD noise by using WF (left) and SR560 (right). In this case, input port of HPCD was also shorten by 0Ohm. In the case of using WF, measurement of de-whitened HPCD (cyan, magenta, and brown curves) were limited by the noise of WF (black curve), though some improvement can be seen from the non-de-whitened HPCD noise (yellow curve). Green curve represents the HPCD noise with engaging the 1st de-whitening filter by old BIO implementation. This result suggest that in the case of engaging 1 de-whitening filter with old implementation, HPCD noise (~1e-7V/rHz) was larger than DAC noise (2~3e-7VrHz * 0.1).
In the case of using SR560, HPCD noise with engaging de-whitening filter reaches ~1e-8V/rHz which is larger than the measurement limit with SR560 (~5e-9 V/rHz). These measurements were done with ad-hoc connection as shown in Fig.2 because I couldn't increase preamp gain by saturation of 60Hz series when I used Dsub-BNC converter. And also this ad-hoc connection seems to induce additional noise (when I tapped the BNC cable, noise level is slightly increased) and my measurement is, in fact, slightly worse than Fig.1 in klog#31771. So I need to confirm measured noise level is really output noise of HPCD instead of the some noise limitation by this ad-hoc connection. On other hand HPCD noise is at least smaller than ~1e-8V/rHz even if some additional noise limited my measurement.
These results suggest following things.
1) engaging 1 de-whitening filter should reduce total output (DAC+AI+HPCD) noise as a factor of 0.1 above 10Hz
2) engaging 2 de-whitening filters should reduce output noise as a factor of 0.03~0.05 and HPCD noise limits a total noise level
3) engaging 3 de-whitening filters are no effect.
