I have noticed that cylinder #3 (at least) is peaking a bit late so did a “GAMI test” flight.
The data collected is even more noisy than I have seen before
The curious thing about the noise is that the steps seem to always be quite large; about 5F-8F. I never see small steps. Yet, all the intermediate values are eventually visible in the data so this is not a case of the instrument being set to say 5F resolution.
Is this some sort of “data compression” where it only writes out new values once the difference to the old value is at least 5F?
I have had a reply from JPI:
That, however, is not entirely true, because the EDM display shows the temps to 1F resolution, and the recorded values also have a 1F resolution in so far as the least significant digit does contain all the possible values 0-9.
Yet something is happening to the data that is recorded which makes a specific channel change in 5F increments. I wonder if this is some kind of data compression scheme. It would be a pretty bizzare algorithm.
Is this some sort of “data compression” where it only writes out new values once the difference to the old value is at least 5F?
I think that is the case. A quick look suggests that there are no changes of less than 5F. So it probably records the new temperature and a time stamp, when the change is more than 5F from the previous one. No records are needed in between, saving memory space.
Good point!
I did wonder why the latest firmware seems to hold more data than the old one did. The feeling I get is that it holds 2x more, maybe. They changed it when it was sent back for the rebuild from the 2.25" version to the 3.125" version.
But it does mean that getting decent data for injector matching is not so easy. One can fit curves to the above data but it’s obvious that the fit is to a fair degree illusory. In Excel, using the polynomial fitting function (6th order worked best) and by downloading the EDM data (in addition to the data I wrote down during the flight) I get this
I am going to get 1-3 new injectors, for a very good price from GAMI…
I have noticed that I am getting “noise” on CHTs 5 and 6. It is quite hard to spot, but basically the sampling interval appears to be much shorter on those two channels.
Going back through the records, it goes back to the new engine installation 1.5 years ago.
I wonder if the EDM config has got corrupted and it is sampling those channels much faster… The actual values are correct, as far as one can tell.
It could also be a lack of grounding on those two channels.
