Thanks for quick response and good questions!
The pulses we want to catch are "analog" with a rise time of 10-20nS and a fall time of 100-200nS. The 50% width varies with amplitude with an average of 50-100nS, roughly. The amplitude as well as the period are arbitrary. Some occur almost overlapping in time, others are up to 1mS (milliS) apart. Some times they occur simultaneous on both ch, some times not. We only need to catch the ones that come simultaneous. It is sufficient to catch pulses that come 10-100uS (microS) apart. We will obviously loose whatever comes during this dead time, but that's ok.
Further info re your questions:
1) The peaks occur within 20nS of each other and the top (flat part) is less than 10nS. We can, if necessary, add an analog peak detector and hold the peak up to 500nS or, preferrably shorter time.
2) According to the info above, we would like to measure both ch every 10-100uS, faster is better. The sampling of both ch should take less than 500nS. If the saving to HDD keeps up with the 10-100uS period, fine, if not, the data has to be buffered up to 10 minutes.
3) We don't need a time stamp.
Interesting application, right?