When you click on links to various merchants on this site and make a purchase, this can result in this site earning a commission. Affiliate programs and affiliations include, but are not limited to, the eBay Partner Network.
I am replacing the distance sensor on my 987.2 Cayman and have read just about every thread on the topic and watched every video I could find. I have the car stripped down, I have built a suitable puller for removing the case, and I have a Design-9 distance sensor in a box, ready to go in. With the black caps removed, my car does not match what I'm seeing in the writeups and videos, probably because it is a first year PDK (2009). Rather than two large external circlips, my car has a "nut" on the bottom shaft. Rather than a hex nut of some sort, this is a round ring, with 4 notches in it, and it is locked by punching-down the rim into two alternate recesses in the shaft.
I have cleared the lock "tabs" out using a modified flat blade screwdriver and attempted to the the ring to move using a pair of needle nose vice-grips and a long screwdriver, also a set of 90 deg needle nose pliers, then trying to gently hammer directly on the notches with a screwdriver. All I'm doing is damaging the nut:
I am replacing the distance sensor on my 987.2 Cayman and have read just about every thread on the topic and watched every video I could find. I have the car stripped down, I have built a suitable puller for removing the case, and I have a Design-9 distance sensor in a box, ready to go in. With the black caps removed, my car does not match what I'm seeing in the writeups and videos, probably because it is a first year PDK (2009). Rather than two large external circlips, my car has a "nut" on the bottom shaft. Rather than a hex nut of some sort, this is a round ring, with 4 notches in it, and it is locked by punching-down the rim into two alternate recesses in the shaft.
I have cleared the lock "tabs" out using a modified flat blade screwdriver and attempted to the the ring to move using a pair of needle nose vice-grips and a long screwdriver, also a set of 90 deg needle nose pliers, then trying to gently hammer directly on the notches with a screwdriver. All I'm doing is damaging the nut:
Any help would be greatly appreciated!
Marc
Hi Marc, I have two early PDK boxes, which fingers crossed, stay working. However I wanted to thank you for documenting all of this in case I do need it.
Re the nut, it is was me, I would modify a suitably sized socket to engage in all four tabs. I’d probably use a grinder or dremel to do that. I reason that you would want to know the torque to undo so you could replicate that when refitting the nut. Good luck.
@Pinkawa to the rescue! He did this job several years back and went through this already. He lent me a tool that he modified from the closest thing he could find that fit. Apparently this stupid prong nut is in between "standard" sizes.
I created a castle nut socket by using an angle grinder to modify a 36mm socket. I tried finding one online with no luck. There were some from China that appeared to be correct but I didn't want to chance it. Since the lock tabs aren't reusable I used Loctite 263 to secure it. They are a good reference for torquing it when putting it back together though.
Interestingly, there is a groove for a clip just like the 981 box. You could try to source one of the correct dimension as an alternative to the lock nut.
I used an angle grinder to modify a 36mm socket. Since the lock tabs aren't reusable I used Loctite 263 to secure it. Interestingly, there is a groove for a clip just like the 981 box. You could try to source one of the correct dimension as an alternative to the lock nut.
Yep, looks about like what Pinkawa loaned me, he went down the rabbit-hole of trying to source an actual tool from Germany, so his base tool was that rather than a socket. Good to know it is 36mm.
I was having an internal debate as to Loctite Blue or Loctite Red, I was thinking the red would be really hard to get off w/o heat if I ever had to get back in there?
Part of my problem here is my complete unfamiliarity with this bizarre little piece of hardware, @stjoh called it a Castle Nut Tool. My familiarity with Castle Nuts is on wheel axles, then they usually have 6 or 12 points, but searching on that, I came up with "Spindle Nut Tool", specifically "4 Lug Spindle Nut Tool". I'll look around, if I see anything that looks like a fit, I'll post the link. Thanks again stjoh!
Transmission is back together, it still won't calibrate, tried 5 or 6 times. I'm getting nonsensical errors that point to nothing, the TCU is seemingly just going MIA at some point, flagging itself as a CAN error in basically every other system. TCU is still running SUPER hot. I have two TCU's so I've tried them both, same BS with each. There are 3 states to the calibration: shift rod (7 minutes), hydraulic (8-9 minutes) and Cluches (says it takes about 8 minutes).
I have run primarily Calibration with Parts Replacent, but have also tried the without parts replacement version as well. neither works: "Abort condition or fault has occurred"
The first calibration I ran failed 3 minutes into the clutch calibration. I took a picture about 5 seconds before it failed
Everything seemed to be working, I was watching the values on screen and they seemed to be consistent across gears and shafts.
During the first stage, prior to the distance sensor replacement, my 1-3 shift rod was settling out at 3mm, looking at the other rods, and the current state, I assume the goal is to have them settle as close to zero as possible. I see a basic range of -9mm to 9mm as each rod is calibrating.
With the new Distance sensor, the calibration process isn't as violent sounding. The shafts were really clunking when I was running this a few weeks ago.
During the Clutches phase, at multiple points, it seems like the revs are brought to about 1000 and it sounds like the clutch is engaged to a near stall point. It is one of these little stall procedures that ends my calibration
I've managed to run two full calibrations to the 3rd phase, if I immediately try again, the calibration fails on the first phase every time. I can't say for sure but its almost like if I allow the car to cool down a bit, run it until it hits the magic-minimum of 60C, it then runs through most of a calibration.
I have reprogrammed the computer multiple times, my P/N is not in one of the listed so I have not done the "due to customer complaint" program.
Here is a list of codes that I've pulled across multiple calibrations:
P1871 - Data Record for Clutch Calibration is Invalid
U0418 - CAN fault, brake (get LOTS of these)
P18B4 - Software error (saw this on each TCU, but may have been left over from prior to the Distance sensor repair
P1991 - Torque Intervention not Possible (again, this may have been a left over, only saw it once per TCU)
C151 - Control unit communication (tons of these)
What about the CAN bus pins, did you confirm they are in good condition? If it was me, I'd replace the pins for the CAN bus to ensure you're not having trouble with that connection.
Not sure if you have done the reset or if it will help, I hope it's helpfiul
Yes on both, I have quite the library of documents on PDK, many courtesy of nhtsa.gov. Porsche basically disavows these transmissions (other than selling you another one). The second link is to the driver adaptations reset. I believe this data is nuked when you reprogram the TCU. My understanding is that you have the following layers:
Hardware (Transmission)
Software (TCU Programming) - overwritten by TCU Program Procedure
Firmware (TCU/PDK Calibration) - overwritten with defaults by TCU Program, and at beginning of calibration
User Settings (Adaptations) - To be safe I have done this "no-tools" procedure multiple times, but I'm pretty sure this user-adaptation data is wiped out by any of the above procedures.
What about the CAN bus pins, did you confirm they are in good condition? If it was me, I'd replace the pins for the CAN bus to ensure you're not having trouble with that connection.
Yes on both. When I broke down at the track in April, a mechanic from my local dealership was there and following the codes he pulled the CAN pins (hi/lo) and checked them for connectivity, sliding the pin onto the connector to check for "snug" when done.
A few weeks ago, I started pulling individual pins and checking them for grip. Every one of them was nice and tight, and with the trouble I had removing a couple of them, became convinced I was running down a road that would eventually result in the breaking one of the tiny pin retainers. I decided instead to pull the connectors from their shells, and watch the PDK live values while I wiggled and tried to simulate the worst vibration or jostle possible. All of the values were rock-steady while I pulled and wiggled the connectors
I have still have the rear-end of the car disassembled which gives me way better access to the PDK plugs. I'm going to see if I can do some continuity test between the PDK and TCU ends, but I would really expect the PIWIS live values to be the better indicator of whether the computer is picking up all of the sensors, and able to control the internal actuators.
I have one more weekend day to try to diagnose, failing any resolution today, I'll try to put my screenshots together in some logical way that might present this case better. Anyway, I appreciate you trying to help. You previously sent me the link to Mauser, where I can purchase pins so it may come down to repinning the entire harness.
Just as an aside question, since you have already figured out what pins are needed for the TCU side, do you have ANY idea what size pins would be used on the round PDK plugs? The T-Design distance sensor is a really nice solution, and he offers a lifetime warranty on it, but it would be GREATLY enhanced, in my opinion, if the new wires could be pinned into the connector rather than cut/spliced with the old internal harness. My existing (internal) harnesses Yellow Wire had broken insulation (NOT melted) in 3 places. It almost seems like Porsche had a spool of defective wire that ended up in my transmission. The conductor was perfectly fine so I spliced it as near to the round plug as I could and sleeved it with a shrink tube. I'll include a picture:
...
What about the CAN bus pins, did you confirm they are in good condition? If it was me, I'd replace the pins for the CAN bus to ensure you're not having trouble with that connection.
,,,
My intuition, right or wrong, is telling me that the multiple CAN bus errors that I get, coupled with things like "P18B4 Software Error" (a fall-through exception-handler code if I ever saw one) are are result of the TCU basically falling through its main exception-block and resetting, possibly even a hard-crash reboot. When the TCU gives up/falls through its exception-block/reboots, it notifies the CAN bus.
For non-coders, exception blocks were basically invented for real-time programming, as wrappers around code sections. With an exception mechanism, rather than a crash that ends execution (which would be bad on say, a Mars lander), a crash will instead "fall-through" and trigger an exception handler that tries to figure out what problem occurred, issue an appropriate code, and does whatever necessary to restart execution at an appropriate point rather than just terminate. Exception blocks tend to be nested one inside another, sometimes dozens deep to give a program the best chance of executing the least intrusive handler possible. What I'm saying is that something very generic sounding like P18B4 Software Error, sounds to me like a main execution loop error handler that might have "Reboot Computer" as its exception handler, which would then tell the CAN Bus "Hey I'm taking a union 5 over here, smell ya later".
try
{main TCU code}
except
{something like... issue a non-descript code P18B4 and reset}
The assumption I'm working from right now is that something that was never planned for has happened to my transmission. It came out of nowhere during an otherwise successful track weekend, on car that was running beautifully. From the videos I shot, it seems that the very first time it happened, the car was rapidly decelerating from about 130 down to 40 (lots of fast downshifting) and lurched during one of those down shifts. Things then got progressively worse as I went through the next 7 turns, seeking, slipping, then refusing to change gears. This says clutch-pack all over it, but there are a LOT of clutch codes that could be thrown, yet all I received were codes that pointed at the TCU (P1990, P1991, P18B4).
The most recent track breakdown happened on VIR's back straight, I let off to allow a pass, I think in 4th IIRC, and as soon as I started accelerating, again, seeking and then finally no drive whatsoever. I had to coast all the way through oak tree, roller coaster and hogpen. Roller coaster really felt like a roller coaster that day let me tell you!
Anyway, pedantic explanation of where I'm at, if you stuck with me, thank you!
Toward the end of yesterday I reached out to my Dealership mechanic friend, he told me to try running the Fill procedure from the PIWIS maitenance menu. Having run about a half-dozen of these, over two days, following the installation of my PDK oil cooler, flush and fill, I didn't expect too much. Because the Fill procedure requires the PDK temp to be between 30C and 50C, I could not run this yesterday, as it was too hot from all of the calibration attempts.
I warmed the car up to 30C, pulled the plug and started the Fill procedure. I watched the fill hole with a flashlight and saw the Bleed procedure bounce the level up and down, I was an oz or two overfilled and I let the extra run out over the process of repeating the fill procedure 5 times. I tightened the plug and turned my attention to running a calibration.
Recall that I have two TCU's, the original and a replacement that I purchased from LA Dismantlers after the first time I had issues (November) and received the combination of P1990, P1991 and P18B4, which pointed at a bad TCU. I'll call these TCU1 and TCU2.
TCU2 was currently installed and I decided to start small, running a Calibration without Parts Replacement. This is the shorter of the two calibrations, and executes 2 phases rather than the 3 of the Calibration with Parts Replacement (it does not run the middle phase Hydraulics Calibration). The calibration worked! Following the calibration I got in the car (on lift) and ran through the gears. Because of traction control kicking in, its tough to shift higher than 3 on a lift but all gears worked, all indicators worked, and the LED indicated the gears 1, 2 and 3.
Moving on, I put TCU1 back in (original). I cleared about a dozen codes across all systems, most related to the TCU not present (CAN) along with P18B4 Software Error, and tried to calibrate. This brought back the dozen or more errors. I ran the TCU programming (the non "customer complaint") version and tried to run a calibrate again. Once again, multiple codes. I put the PIWIS into E mode and navigated the German interface to do a manual Program. There were two secions, PDK 0xFF and then another section below that just said PDK something or another with no 0xFF. I chose the top o)ption whch seemed to be a default data block from 2009. This ran quickly (I think it is the 2nd part of the non-E-mode programming) and I reran, this time, a Full calibration. This time the full calibration went through all 3 phases successfully.
I ran another short calibration just to verify that multiple calibrations would work. I ran the car through all the gears I could and all indicators worked properly and no codes wre thrown.
I put TCU2 back in and ran a full calibration, again, successfully. Again, it ran through gears w/o throwing any errors, all indicators working.
so, it seems that maybe there was an air bubble in the system that was causing yesterdays failures. Both days were completely consistent, 0 successes and about 6 failures yesterday, 4 successes and 0 failures today.
P18B4 now strikes me as an indicator that the PDK has a bad calibration, at least in this circumstances.
Somehow the standard Programming feature did not allow me to run a calibration w/o using E mode, but I have no conclusive evidence here, maybe there was something else at work here (like restarting the PIWIS program or turning on/off the car).
One conclusion I will draw is that it is worth performing a PDK bleed procedure if you have trouble with calibrations, and you have no conclusive codes.
I feel good about replacing the distance sensor, previously I could not get shift rod 1-3 under a value of 2.97mm during a calibration, with the new displacement sensor all shift rods are within +/- 1mm. I *THINK* this is what normal is supposed to look like but I'd appreciate anyone's input that has done, and remembers their calibration data.
05-29-2023, 05:13 PM
PDK Distance Sensor Replacement 987.2
