Thanks so much for clarifying this. We made wait_vdd a parameter controlled by software and will try to experiment with it to find some compromise between deadtime and the offset added by the droop in VDD.
the posting you mention is still accurate. Any power supply will drop when you start the Domino wave, no matter how big your capacitor is. Unfortunately the output signal of the DRS4 scales with VDD. So if your VDD drops by 40 mV and you get a trigger and you immediately start the readout, the output baseline will also be shifted by about 40 mV. If you are sensitive to dead time, you can remove the wait_vdd state completely, but then you have to deal with varying baseline shifts. If you have narrow signals sitting on a broad baseline, you can correct for this by measuring the baseline outside your signal, then subtracting it before integrating your pulse. If you have lots of pile-up in your signals, it might sometimes be hard to evaluate the baseline on an event-by-event basis.
I am working with others at UCLA on a custom made board built around the DRS4. We are in the process of writing firmware so I am adapting the readout state machine from the evaluation board firmware.
I see in the state machine of the eval board firmware that after a trigger is received, the FPGA goes into the start readout state and then into "wait_vdd", where the FPGA waits "~120 us for vdd to stabilize" before reading out the ADC.
Our application is sensitive to deadtime and this wait_vdd state adds very significantly. I am trying to find anything explaining the necessity of wait_vdd in the documentation / elog and have only found so far your old forum posting, https://elog.psi.ch/elogs/DRS4+Forum/12
Does this forum posting explain wait_vdd or is there a another purpose that I have missed?
If this post is relevant to wait_vdd, does the advice of large capacitance and an LDO with fast transient response still apply or are there any new recommendations?