Looking at the viability of offering a manual/stick gearbox conversions for XK8/XKR (in the UK) either as kit or a service. Both options of a Tremec box or the much cheaper (used) S-Type box run into the same issue of the TCM having an electronic fit about the missing auto box and sending the car into limp mode + EML or gearbox lights (an MOT/Inspection fail here).

Considering two options to get around this - the alternative of a 3rd party ECU is not being considered due to cost.

1) fool the TCM into thinking there is an auto box there - feed it a speed signal or two, have some resistors for solenoids, then either mimic the feed back in Drive or see if it will be happy if I tell it it is in say 4th on the J gate. It will need to tell the ECU it is is N/Park to allow a start - most of these things are just Gnd/B+ on a particular wire. Some of the TCM output is picked up on the CAN bus and I want ABS and SRS to work.
Downside of this approach is that different box's/engines will need a different set up, different pin connections and more monitoring means more fooling.

2) let the TCM have a hissy fit but stop it from telling anything else by interfering in it's CAN output by blocking any error messages getting out. I think this method although initially more complex would be more portable across the range of engines and gearboxes.

I realise manual conversions have been done in the past by EPJ, TL Jag and championed by the late Norman D - it turned out to be too expensive for most hence my considering a different route. I also know that Paragon/XKracer has been doing this from many years ago, but he uses the 3rd party ECU route.

Would welcome technical opinions on the two alternative routes - or a 3rd way if you can think of one!

 

The TCM is inside the transmission so if you replace the ZF with something else, then you aren't going to have a TCM to throw a hissy fit. The path will likely be something that can stand in for the TCM and send all the right messages to all the other modules.

You may also try speaking to Jaguar Specilaties. Jaguar V8 Kit They have done LS conversions and have manual transmission options so have done all the work to decode the CAN messaging.

 

+1 on creating a TCM 'stand-in'

Fooling the TCM when you have no auto transmission does not seem worth the effort, even if the TCM is retained (easier with the Mercedes than the ZF, since it is not in the transmission but next to the ECM)

The TCM canbus messages are outlined in the appendix of the electrical guides - TCM sources messages to the ECM, the ABS, the INST, etc.
- so you would have determine and test which of those messages to generate with a custom pseudo TCM box, and what info to populate - heartbeats, fault messages, torque reduction requests, torque converter status, kick down, etc. to keep the ECM and other receiving modules happy and working the way you want them to
- I would expect a reasonable investment of time to develop and test a working solution, but I believe it should be possible

crbass has decoded most of the canbus messages, so you can search for those threads for more detailed insights into the details

 

It may be feasible to spoof the TCM CAN with a bit of electronics. There are only three TCM IDs with 20 total messages. One of these IDs and message is the heartbeat that must be transmitted. The electrical guide has the receive modules for each message, the CAN messages as far as I have decoded for the 2001-2002 (i.e. not completely, but mostly) are in this thread

​​​​​​https://www.jaguarforums.com/forum/x...-222041/page2/

I expect that the 03-05's have changed, at least some, there was a new Ford CAN spec then. I expect the 97-00 to be similar. The complicated thing is making it all consistent for the ECM/ABS.

 

Thanks all, The electrical guide was very helpful and I have taken a good look through the CAN decode thread (thanks crbass !), looks like I have a bit to learn.
So for a TCM stand in I need to look the physical wire information ( eg B+ on one pin to confirm P/N to allow start) on the plug. Then I would need to ensure that any data expected on the CAN bus is there, I can probably substitute CKP for input speed.

Looking at the '99 manual I get CAN transmission of


What I don't know how best to deal with is what gear it should be saying it's in, or if I should just take the engine speed and front wheel speeds from the ABS CAN, work it out and fire it back out. Sure the manual box will have a N and reverse sensor.

That's my thinking so far!

 

You need the network token, otherwise the other modules think the TCM is gone, no matter what the other messages are. That's pretty straightforward, just broadcast one byte every 200 ms on ID 7E9 with the hex value of 14. On the others, for example, if you don't send an oil temp, it's likely going to error, etc.

One potential way to check the basic functions is to bridge the CAN wires across the TCM so you have a whole CAN, but no TCM (i.e. take out the TCM). Turn the key, but don't try to start, and see what it complains about. That takes motion out of the equation.

 

Thanks, so the Token is the "heart beat", what is the PECUS flag ?
That's a good idea. I have been looking at the Error code list to see what the TCM is going to complain about - but now with the stand in route it will not be there to complain, so I need to look at the other modules error list and see what errors they have relating to data expected from the TCM.

Looking at the schematics I can see that the fluid temp, speed sensors, mode and kickdown switches are the only physical data that goes directly to the TCM, plus the "dual linear switch" which I assume is the later model shifter position sensor. So probably that, speed sensors and fluid temp data will need to be put onto the CAN network,

 

PECUS (= Programmable Electronic Control Unit System) tells whether a given module is programmed (properly, as in PECUS-T or PECUS-E). Don't know what happens if it's not there at all as opposed to 0 or 1.

 

If it was me, I would first on the unmodified car try recording the CAN data for the TCM as you accelerate up through the gears. Then write a bit of code to play the appropriate TCM data back at the corresponding speed/RPM value instead of the TCM. The ECM will think you have the original transmission fully working.

The advantage of doing it this way is you don't need to know anything about the data, just that it is from the TCM.

 

I have plenty of TCM/ECM/ABS/INST data like this for the 2001-2002, but it does vary by load, slip, kickdown, etc. It's not clear what the ECM considers 'bad behavior' from the TCM, if any (e.g. some combination of rpm in/out slip, etc). I do know that the Jag engineers were very concerned about these kinds of issues since running things 'by CAN wire' as opposed to actual wires was pretty new.

There's no information on ECM logic beyond a very basic functional diagram as far as I have been able to determine.

 

As long as there are no transmission faults shown when recording the data, it doesn't matter - there will be no automatic transmission to protect. All that's needed is to give the ECM "good" data so that it doesn't show faults or go into limp mode.

It would be so easy to try, compared with trying to work out what the ECM is expecting, I would do this first.

 

I agree on one hand, relatively easy to try, but the ECM isn't necessarily passive here, it reacts to what happens in the TCM. Depending on what the ECM thinks is happening in the TCM, it may well go outside the envelope of what the ECM thinks is okay, even if it was okay for the 'test run' because of differences in load/etc. So, 'good' may be situationally dependent on RPM, and other things in the engine.

If one were to do this, in general, one would probably want a range of responses in motion, with a response surface for ECM inputs based on all of the output parameters from the TCM. It may be that they all follow one curve, but I'd bet they follow a multivariate surface.

 

In essence, the path you are describing is to somehow either produce a (software) TCM CAN simulator or somehow keep running the existing TCM in a simulated environment (Matrix-style) without the original transmission hardware. Sounds like a challenge...

 

Thanks for the idea dibbit, I see what you mean, but not sure if it would be easier than a calculated response as you would still need to look at the engine RPM and the car speed then look up the relevant chunk of information.

Have to say I don't know what the ecm does with the informations as if you look at the documentation for the Manual box for the S-type (that is the likely replacement) it appears to have no feedback - but the Auto has the same sort of output as the XK8/R does.

I think the error codes will be the key to what is taken notice of.

 

That's it. Except as was pointed out on the ZF 5HP24 on the XK8 to 03 the TCM leaves with the gearbox so it's looking like the route of a ghost TCM - it can be seen but it's not there!
I like a challenge!
This is interesting http://www.thejagwrangler.com/upload...on_control.pdf

 

I knew I had read something like this somewhere:

"Overlap shifting is only possible using a high speed communications link to the engine management system so that shift energy management can be closed loop controlled in real time. Shift energy management allows the transmission to control the engine output torque during a gearshift, so that excess power is not dissipated in the transmission friction elements during the shift. The functions of shift energy management are:" Whiffin, 1996, as cited above.

I think this is where the problems come. The engine output torque is modulated during a 'shift'. If the fake TCM doesn't provide the 'correct' fake shift for the load, there's the potential for a bad feedback loop of unknown magnitude, especially when interacting with clutch/acceleration gear change. Here, one doesn't want it modulated at all.

Note added: It may be possible to skip the modulation based on the TCM output CAN values. If so, that would simplify things.

 

That was the bit that looked a bit scary to me too. Now I understand what the first message I listed was for " torque reduction request For shift energy management" - BUT if it is not sent is it an issue ? What about if we tell the ECM that the box is in "manual mode"

Nota added after your note - great minds think alike!

 

It's the other way around AFAIK. The TCM for the earlier 5HP24 is in the engine compartment. The TCM is packaged "inside" the later 6HP26.

 

I don't know, and that was one of the messages that I hadn't fully decoded (i.e. what number corresponded to what percentage, etc). But it's mostly zero (not shifting), and if no interaction is required, you may just be able to transmit zero at all appropriate times.

 

Thanks fmertz, I did question post #2 in my mind, but as now you mention it I checked in the w/s manual for 99 xk8 and R you are correct that the TCM is shown in the engine compartment for both. - CCfulton has an 03 so I suspect his would be with the transmission - as would yours.

Either way (certainly <2003) there is only the CAN wires and a B+ to an ECM pin to say it's in N/P so the engine can start, the ghost TCM seems to be the route to go as it should cover XK8/R 97 - 2006 with minimal modification.

Just looking at the ECM error messages in the attached and I am not seeing anything relating to the TCM output except Token message missing and sudden loss of P/N signal voltage whilst driving!

This is interesting from the DTCs
P1643 CAN TCM token message missing
When ECM detects fault: [M] – Limits throttle to approximately 30% – Inhibits cruise control (Torque reduction request data missing; results in harsh transmission shifts)