the theoretical time you need, with 9-bit words, 60 bytes (30 channels) per unit, 8 units and 400kHz bus is approx 9 * 60 * 8 /400000 = 10.8ms . However, given the overhead in-between the transactions with each device, I'd guess you are more likely to be in the 13ms : 15ms range. The Bela Pd backend attempts to read every 7ms. If it's still busy reading the previous batch, it skips one period, so you either get 7 ms or 14 ms or 21 ms, etc ... So you are possibly at risk of jumping from 14 ms to 21 ms. That may be easily changed so that it restarts immediately from the first Trill as soon as it's done scanning the last one.
So yes, you may in principle have a bandwidth issue, and yes you may see some advantage in using the EVT pin. An alternative would be to reduce the bit depth of the transmitted data to 8 bits (instead of the default of 16), which halves the transmission time. If all you are looking for is a threshold, rather than a continuous reading, that should be plenty and then you could continue using the current method.
If you want to do the EVT thing, and you have enough digital pins available, I would recommend implementing that in the C++ backend, it's just a few lines and I could help with that.
