Reports 1-1 of 1 Clear search Modify search
CAL (Pcal general)
takahiro.yamamoto - 12:40 Friday 24 February 2023 (24131) Print this report
OFS loop of XPcal path2 doesn't work properly

Abstract

A current operation point of the path2 control of XPcal seems to be out of linear range of VGA.
Input voltage to VGA should be reduced as at least 0.3~0.6V in order to obtain the design OLTF.
A circuit design seems to be difference between XPcal and YPcal.
So they should be replaced to a new version (V4) made by AEL.
 

Details

Because XPcal becomes available again as reported in klog#24013, I resumed to characterize XPcal.
I found at first an open loop transfer function of path2 was different from a design one.

Fig.1 shows the measured OLTF of path2 of XPcal (red curve) and a model function (black curve).
Zero/pole seems to mismatch between the measured OLTF and the model.
Because I haven't know the AOM and PD efficiency yet, overall gain of the model function was decided to adjust with measurements.
Zero/pole were decided from the circuit diagram (JGW-D2113413) as p3k:3k;z30k.

On the other hand, OLTF of path1 of XPcal and both paths of YPcal match with the model well as shown in Fig.2-Fig.4.
By the way, contorls of both path of Xpcal seems to oscillate if we believe measured OLTF.

In order to check what happen on path1path2 of XPcal, I measured OLTF with decreasing VGA gain as shown in Fig.5.
Because design OLTF can be available with decreasing VGA gain, it seems to saturation on somewhere or be out of linear range with a current operation point of XPcal (1.7V) to VGA input.

As just in case I checked relation between VGA input voltage and VGA gain for all four paths as shown in Fig.6.
Then I noticed that the relation was quite difference between XPcal and YPcal.
Fitting function is defined as
dB = [0] + 4.4 + 49.9 * V / [1]
where [0] is variable preamp gain decided by R39 and R40, "4.4" and "49.9" are fixed parameters of VGA from spec. sheet, [1] is a voltage divider gain decided by R37 and R43 (see also JGW-D2113413).

Maybe R39 and R40 are quite difference between XPcal and YPcal.
Because of historical reasons, OFS circuit has been improperly modified before O3GK.
For the easy characterization, OFS chassis should be unified as V4 version made by AEL.

postscript
The vertical axis of Fig.6 is "VGA gain" in correctly not "OFS gain".
I removed 1/3 gain by U10 (R38, R34 and C31) from the measured raw TF from OFS_ERR_V to OFS_TAOM_V.

Images attached to this report
Comments to this report:
takahiro.yamamoto - 12:53 Thursday 02 March 2023 (24232) Print this report
I changed default OFS gain in calparams.py for XPcal path2 from 1.7V to 1.4V.
Now measured OFS response is consistent with designed one because saturation at somewhere is removed.
(See also blue curve of Fig.4 in klog#24131)
Search Help
×

Warning

×