We should try to get you interested in these 4 cam V8's we play with here @ Lextreme
That might not be a bad thing, I'm very impressed with the development that's been done on the IS-F. But for that to happen I'm afraid someone's going to have to offer me a job that will pay for the toys. As it stands right now, old technology is about all my budget will stand. This new toy (the AA80E from an IS-F) has really stretched things badly out of shape.
Are PLC's fast enough? You know, that's a very good question. Although I don't have the answer right now, just the fact that they've been in existence longer (arguably) than automotive computers leads me to think that they probably are. Since it wouldn't have to deal with spark or injection events I would think that a cycle time under 10 ms might be enough to get the job done and that's not much of a demand on the electronics when component times are typically in the microsecond range and sometimes even less. So the answer is probably yes. Can it be as sophisticated as the stock controller? Perhaps, but making that happen would be quite a job. This also brings up an interesting possibility. In "M" mode the stock system requires 200 ms to initiate a shift. The shift itself requires a scant 100 ms to execute, which is really pretty fast considering that F1 trannys are just half that. So how much of that 200 ms is processor time? A good air solenoid valve 15 years ago could switch in 12 ms or less so I have to think that the "Fast acting linear solenoids" used in the transmission's hydraulics are at least that fast. This leads me to think that a fair number of processor cycles are being taken up with things like checking the conditions of all the sensors, switches and other inputs, consulting tables, accessing memory, and being fitted in between engine requirements, and finally calculating the optimal gear selection and gear change parameters before the ECM ever gets around to issuing a shift command. Don't get me wrong, getting all that done in 200 ms is a laudable feat, but there ought to be quite a bit of time there that could be saved in a simpler system. It could mean the difference between having 1/3 second shifts (including shift lag) and 1/4 second shifts or less. To me, that is a very significant difference.
I got a good deal of information from the local Lexus dealership. The IS-F uses a separate TCM which has a dedicated cable to the transmission and is connected to the ECM via CAN bus. The TCM controls all shift parameters but the ECM determines shift points, firmness, and mode using numerous inputs. So to use the stock TCM requires a CAN bus equipped controller with the appropriate outputs, whatever they are. I think that either the TCM and ECM should be used as a pair, which will require hacking the ECM at least to the extent of phantom inputs, or neither should be used, although it may be possible to use the CAN bus enabled Megasquirt to control the TCM, provided the code is modified to provide the needed outputs (which I don't know how to do at this time). I don't know if James Murray or one of the other Megasquirt gurus can be persuaded to pursue this or not, considering that they are strongly biased towards GM. All I can do is ask.
This tranny has 5 of the linear solenoids, all of which require a PWM control signal which is what gives control of the shift firmness. The TCC solenoid also used PWM for intermediate or partial engagement for economy, not fundamentally a necessity for this application. Also there are a couple of other solenoids used in this manner, one for line pressure reduction and at the moment I forget what the other function is. But since there are 9 solenoids I might as well just expect 8 of them to require PWM signals. That means a PLC controller would have to have at a minimum 8 PWM outputs (TCC Lock-up could be controlled manually) and about a dozen inputs at least. I do now have the solenoid identifications as well as truth tables for on/off conditions, so the only remaining parameters to be determined are actuation interval/PWM slope, and delay/overlap plus addition of any programming for automatic mode. I don't think this is an insurmountable problem by any means.
Jim