One pulse is two measurements

The payload has to determine when a gamma ray arrived and, separately, the approximate energy its pulse amplitude represents. I had been thinking of the analog chain as one path ending at the converter, with the timestamp attached afterwards as a label. That arrangement gets the energy roughly right and the time badly wrong, because by the time a converter has finished digitising an amplitude, the moment the pulse arrived is several stages in the past.

I split the chain in two from the comparator onwards. The timing path is detector pulse, comparator, STM32 hardware timer capture, calibrated event timestamp. The energy path is detector pulse, amplifier, shaping or peak hold, ADC, calibrated pulse height, approximate photon energy. The comparator serves the first; the converter serves the second; both start from the same pulse.

The two paths have different acceptance criteria, which is what makes the split real rather than cosmetic. Timing is to be verified against reference equipment: electronic variation below 1 microsecond, absolute time within 10 microseconds of UTC. Energy is to be verified against known gamma-ray energies at a licensed radiation facility: pulse-height error below 5 per cent, measurement within 2 ADC counts after calibration, calibration peaks initially within 10 per cent of their accepted energies.

Neither calibration establishes the other. A perfectly calibrated energy scale on a badly timed event is a photon of known energy at an unknown moment, which for a burst is close to useless — and the reverse is a precise time attached to a number that does not mean energy.

Rev A has the comparator and the converter, but the comparator is not shown reaching a timer-capture input and the shaping stage is absent. Both paths are incomplete, in different places.