OK - I've now changed the ICV, changed the TPS, reset the ECU several times and the problem is just as bad as ever. I drove from Oxford to London the M40 last night and was actually shouting at the car by the time I got to London. It's not funny rolling along at 5mph in a heavy jam with the engine doing 3000 revs! The car actually accelerates all on its own if you get it rolling in 1st or 2nd gear. Not much but it definitley does accelerate up to about 3000 revs, even with the clutch engaged. This can't be right! The choice between silly over-revving or using the clutch as a brake for the engine is not a good one!
A bit more experimentation shows that dropping the clutch briefly drops the revs, but then they climb again if the car is moving. I guess the ECU must know the clutch is being dipped and obligingly drops the revs with it. So I can coast along at reasonable revs by sitting in neutral and pumping the clutch about every two seconds. Clearly I don't generally coast much but you know what it's like on a heavily congested motorway...
I am convinced there are no air leaks and I've checked the throttle cable and am sure it is not sticking and that the throttle is closing properly. Nothing else is obviously wrong so I am now rather stumped.
Disconnecting the speed sensor that sits on the gear box (where the speedo cable attaches) completely solves the problem and the car seems to drive perfectly at any speed and with just as much poke as normal. I am tempted to leave it like this but I'd rather get the problem fixed. The VSS must be there for a reason! I'm a bit loathe to change the VSS until I'm sure that's the problem (although the fact that disconnecting it cures it would seem to point that way). What do people think?
Is there a firmware upgrade for the ECU available that might solve the problem?
Thanks in advance, and sorry to keep boring you all with this.
Tony