I think JPI are running what is called a finite impulse response filter – a fancy name for taking the last 10 (or whatever number) ADC readings, adding them up, and dividing the total by 10. So you keep the most recent 10 and run this filter on them.
When a spike arrives and trashes one of them, it takes 10 sample periods for the trashed value to work itself out of the output reading.
The interference seen in the video shows that sort of thing. The strobe pulses are very brief and would trash only one reading.
What they should have done is tested each reading for sanity, and discarded it if it is say more than 20% above/below the current output value.
Peter
On my Twincom, the readings on one side are rock solid, and on the other I see some sawtooth variation. It’s a long time since I looked, but I recall a similar magnitude/periodicity to your report. I’ve always assumed an obscure interference or earthing problem and left well alone. Twins are good for cross checking and also for controlled experiments!
Interestingly, in terms of response time, the JPI is more damped than the old analogue EGT. We sometimes get vapour in the fuel lines, particularly with the tip tanks in certain temperature conditions. You see the fuel flow dip and then rise – but without hearing anything or seeing the JPI move – but you can see a little twitch on the analogue EGT. The fuel flow always stabilises after a while in the cruise.
Alan
