Hrec online uses a combination of B1_PD1 and B1_PD2 photodiodes:
0.992*B1_PD1_Blended+1.008*B1_PD2_Blended

Following a suggestion from Michal, I did a reprocessing of Hrec using as input only B1_PD1 photodiode and an other reprocessing using only B1_PD2 photodiode,
for 3 hours of lock between 2024-02-05-01:00:00 UTC and 2024-02-05-04:00:00 UTC.
The config files used are here:
/data/dev/hrec/hrectest/hrecpd1.cfg
/data/dev/hrec/hrectest/hrecpd2.cfg
The produced data are here:
/data/dev/hrec/hrectest/hrecpd1-1391130008-10008.gwf
/data/dev/hrec/hrectest/hrecpd1-1391130008-10008.gwf

Plot1 show hoft_raw spectrum for Hrec online (blue), HrecPD1 (orange), HrecPD2 (green) and the difference between HrecPD1_hoft_raw-HrecPD2_hoft_raw (cyan).
We can see that below 100 Hz, where mirrors longitudinal controls are the highest contribution to hoft_raw, the difference may be was is expected, while at higher
frequencies, the difference contains an additional noise sum of noises of PD1 and PD2.
We see also that the ASD spectrum of hoft_raw using PD1 or PD2 alone is sqrt(2) higher than ASD spectrum of the hoft_raw using the combination, as expected.

Plot2 and plot3 are zooms around 100 Hz and 2500 Hz
Plot4 shows the BNS range of Hrec, HrecPD1 and HrecPD2 over the 3 hours.

Plot5 shows the coherence between HrecPD1 and HrecPD2 and the coherence between Hrec and HrecPD1. The coherence below 40 Hz comes mainly from the common LSC_MICH and LSC_SRCL noises present in Hrec_hoft_raw and in HrecPD1_hoft_raw or HrecPD2_hoft_raw.