By chance today, I noticed a difference in normalized yields depending on the deadtime. I found one run that had a deadtime > 30%. This run's normalized yield doesn't agree with others at the same setting with more reasonable deadtimes. It's hard from this quick analysis to determine a rate where this definitely becomes a problem but keeping the rate below the maximum DAQ rate which is probably a little under 8 kHz (I can't recall off the top of my head, but a maximum of 7 kHz most likely safe) and deadtime <20% should help.
Figure 1
HighDT.pdf