I thought I had these all under control, but it appears I was wrong.
The CLT sensor data log is full of noise and it spent most of the time on the rail (101C) during my last run. There is a small amount of noise on the MAT sensor too, but nothing like the amount on the CLT. The temperature gauge in the car shows a completely sane stable temperature.
(It's possible that both of the noise magnitudes are the same when expressed in LSBs though - I haven't done the sums to prove or disprove this)
It's a new CLT sensor and the grounding is much improved after I discovered there was only one ground from the chassis to the engine. There's now 3 - one from each head to a point close to the battery and one from the block to the chassis at the front. The VEMS ground wires go to the LH head while the coils are grounded to the RH head. The rest of the grounding is pretty much as Rob's guide, give or take mods to make it fit.
Apart from the additional engine grounds, this setup has run well for a lot of miles.
The noise starts about half way in to the run. It seems to coincide with turning the lights on, but I can't see why this should be the cause - lights aren't the noisiest of things.
No other datalogs since the car was returned to the road show this noise problem. And it doesn't seem to affect the way the car runs other than the number the CLT gives and therefore the fans are always turned on. It then went away while I was looking for it!
I'm stumped. There appears no logical reason for this intermittant fault - no dry joints I can find, no broken wires etc...
Anyone have any suggestions on how to sneak up on it and catch it?
