Thanks johnsfine, but still no success, I'm afraid.

At the risk of asking something dumb, if I look at the waveforms in the spreadsheet, are the timings for that initial burst accurate? It says, for example, 7000us for the first burst, but if you count the gridmarks (at 500us per mark, per the notation elsewhere), shouldn't that initial duration be 8500 us?

There's got to be something else wrong somewhere....

--Imre