mskar

So... it happened again. Went back to VIR having done a PDK fluid change, installed an LN Engineering oversized PDK cooler and having shaken it down as much as I could over the last few weeks (including driving it 3.5 hours out to VIR). It has run perfectly since I got it back in December. After my second session, my PDK failed with the same codes noted in my first post. It was actually worse this time, the car was totally inoperable, could not shift into any gear. Last time it was shifting like it was haunted until I reset the codes, then it would run about 20 minutes before Casper-mode kicked in. Hot off the track (and with no gears), I had the same codes:

• 18B4 – Software Error
• P1990 – Positive engine torque intervention not possible
• P1991 – Engine torque intervention not possible
• C152 – Communication PSM control unit (drive)
• C150 – Communication DME control unit (drive)
• 5400 – Fault Sport Mode
• P1629 – Fault manual transmission control module

By happenstance the tech who performed my repairs at my local Porsche dealer back in December was at VIR instructing and brought a REAL PIWIS with him. He remembered my car well and went down the same path as last time (reprogram + calibration) but this time the PDK calibration aborted, time after time at the same point (we did like a half-dozen attempts). I had my spare PDK computer with me, and swapped it in, neither computer would take a calibration. Later on I decided to try a calibration using my "PIWIS clone", and it wiped the programming from my computer entirely and just said "duh, error" (I'm paraphrasing). Confirmation there that the PIWIS clone is garbage for calibrations, at least mine is. We continued trying to calibrate the computer that hadn't been blanked to no avail. You can hear the shift levers moving during calibration, everything seems on track, then it just gets to a certain point and aborts. Relevant codes raised during calibration:

• P18B4 Software Error
• "CAN signal error, gateway - Abort condition or fault has occurred"
• P1691 - Communication, transmission control unit (no signal)
• U0101 Communication, transmission control unit
• P1628 CAN interface, sgs no communication
• U0418 CAN fault, brake

I've never seen the U0418 before and the brake switch has not been an issue with starting the car, HOWEVER... When I hold the brake down with the car started, I get a slight clack-clack-clack-clack from the shifter area. Maybe relevant? I have a brake switch in my emergency spare parts box that I'll put in when I get my car back. I hit the lottery this time, and a group of folks I met on Friday night had space on their trailer to bring me and my car back on their enclosed carrier. This time I only need to retrieve my car from 20 minutes away!

I'm going to keep posting to this thread until I've resolved the problem, hopefully soon. If anyone has any ideas or input I'm open to any and all suggestions.

Marc

A432

Sorry about the PDK gremlins. Shot in the dark, but anytime I see unclear results it starts pointing to potential electrical connection issues.
Went through this with another poster and here's the thread about cleaning the plugs/connectors.
987 Spyder PDK troubleshooting (and repair?) - Rennlist - Porsche Discussion Forums

Also, in the 987 and 997 stuck PDK threads above, there might be some newer relevant info.

PDK Repair Summary - Rennlist - Porsche Discussion Forums

Guide to Repairing a PDK Transmission - Rennlist - Porsche Discussion Forums

mskar

Thanks A432, I'm reading that first thread right now. A harness issue or bad connection is high on my suspect list: the issue has been intermittent, and either vibration or heat brings it on.

harveyf

Man, I feel your pain. This will not be much help but after recently removing and tearing down the engine in my 07 Cayman S, one thing I included in my new parts list was a new engine harness and as many sensors as I could economically justify. Regarding the harness, there are two things. Obviously, it has lived in a fairly hot, harsh environment for 16 years. It is really hard to evaluate insulation integrity. Also, it seems that every Porsche terminal connector is designed differently, which is frustrating as you try and crack the code on how to release them. Which means that some break when they release.

Although VIR is a tough test, it is only April/May and as such not a really hot VIR track day, in the great scheme of things (I remember going one summer when the daytime temps were 100 plus). So I would lean towards a vibration issue. Which is a tough one to identify. But in that regard, knock sensors might be something to replace as a matter of course. There are also the Hall sensors. Another thing is that on track we tend to stay way up in the rev levels for an extended period of time, at least relative to street driving. Again, it points to a potential vibration issue. How are your motor mounts? Is your serpentine belt OEM? There is a lot going on in the DME of these cars. Makes troubleshooting a bear.

stjoh

You may want to take a look at the pins in the TCU connector. I had some issues with these recently causing some weird and intermittent faults. Turned out that some of them were not making solid contact with the pins in the TCU. Easy and cheap replacement. I'd start with the pins for the CAN bus as the codes are indicating a communication issue. Secondly, I'd check the pins for power. See post #1245

https://rennlist.com/forums/997-foru...l#post18769254

mskar

Good call and exactly what the Porsche mechanic working on it was doing. He pulled the CAN bus pin(s) and gave them and tightened them up best as he could with the limited tool set that 3-4 of us could cobble together. I then pulled the front cowl and exercised the connector a bit, all the pins looked clean but I would've killed for a bottle of deoxit before putting it all back together. That is excellent advice and I will plan on checking the rest of the TCU connectors. Thanks for your help!

mskar

Totally agree, I've been at VIR in mid July, temps were in the triple digits every day of the 3 day weekend. The first time this happened was last November, definitely not the hottest month on the calendar, nor is April. I can say that both the DME and TCU run HOT!!!! I don't know if its because they are generating that heat or if its because they are attached directly to the outside of the engine compartment, but as someone who deals with computing devices for a living, they were way hotter than anything you see in a server room. Btw, the ground next to the DME was shiny and clean, I reseated it anyway, but it made no difference and I suspect those units have a good ground.

This car has about 55k on it, everything in the engine compartment is new looking, paper stickers are still in good condition. I have considered, as a nuclear option, hail-Mary, spec'ing a new transmission harness. I imagine dropping the engine and transmission at that point is inevitable.

mskar

Using a borrowed PIWIS-II adapter and having created a nice working VirtualBox PIWISII vm, I was finally able to reprogram the TCU I blanked using the crappy PIWIS3 clone's "PDK Calibrate" function last weekend. I don't know what is wrong with the PIWIS3 but it erases any TCU that I attempt to either program or calibrate. I'd love to hear confirmation that the PIWIS-3 clone either *does* usually work, or is broken for everyone. Maybe its just the version I got from my hacker, er... I mean... "vendor"? I'd thought of picking up another adapter from AliExpress, they appear to be about $75. Maybe its a matter of crappy hardware?

I installed the erased TCU, and hit program. It took about 15 minutes and I noticed 2 passes, maybe one program and one verify? Anyway, at that point I could view the extended information from the TCU, all values were empty
I installed the still-good TCU and used the "Replace Controller" functionality to read the values off of the still-good TCU, then disconnected the TCU and connected the one that I was trying to reprogram, and wrote the values.
The final message was that a CAN bus errors would have been created and to clear them
I tried to clear them (ton of C152s) but most of them just came right back. Turned off car, restarted the PIWIS-II, same thing. Tried to clear with PIWIS-3 and Foxwell, same thing.
I put the still-good TCU in and cleared all of the errors, this time they cleared.
I put the half-programmed TCU in and tried to clear the errors, filled up with C152 errors again.
AllDataDIY manual suggested that "Program" occurs AFTER the "Replace Controller" procedure so I ran a reprogram, same thing.
Since I couldn't edit most of the "?" entries under the PDU Extended Information tab, I backed out to the main PIWIS-II menu and selected mode "E" and went back in.
I now had additional options for programming, one of which (translated from german") referred to Program w/o reflash. That page pulled up a bunch of information that I had no idea what it meant, mostly referring to data blocks, I'm assuming these were individual item programming options. There were two sections, one referring to a hex code and the second referring to defaults. I chose the most "defaulty" options (I did not take screen shots since I was just monkey punching buttons at this point). Lo and behold, this worked. I was able to clear the CAN bus errors at that point and the unit showed as programmed.
I switched my PIWISII back into "V" mode (and English) and ran a Calibrate, this time the calibrate ran flawlessly

One of the other things I did was pull the connectors from the TCU and spray Deoxit on the pins. I also pulled the big round connector from the PDK at the back left of the trans and sprayed deoxit on that. I think there is another big round connector on the PDK, I'm going to get to that one as well and make sure it is clean. I started familiarizing myself with the wiring diagrams, the factory ones really suck, AllDataDIY has a really good color one though. Especially frustrating is that the factory diagram pins #s are all out of order, with a lot of the "n/c" pins not even listed so I spent a lot of time scratching my head and scrolling between the cutoff pages displayed thinking I was missing a page. It was also a bit of a challenge to read the #s on the pin connectors. I created an Excel diagram of the BMW connectors that the TCU uses. One is Gray and one is Black. I concentrated on the Black connector and tugged/pushed at each pin to make sure each was making contact. This entire thing has struck me as electrical, and I'm not done yet. I'm at the track again on Memorial Day weekend, I have the tools to do a track side repair this time but remember that didn't strictly work last time so I want to eliminate the possibility of wiring and connectors cause the PDK side are a b*tch to get to. I'm going on the premise of heat/or vibration.

Speaking of heat, it would be helpful if someone else could pull their right luggage compartment cover and check the temperature of their PDK TCU at running temp. After a calibration, TCU DISCONNECTED from the firewall (laying in the middle of the luggage compartment) the TCU was HOT, like burn your fingers hot. Before anyone suggests that it must be mounted in order to be grounded, the unit is electrically isolated from the body when mounted, it sits in a plastic carrier. The pins carry multiple grounds to a bolt about 6" away from the unit (which were previously checked for corrosion or loose connection). I'm kind of amazed at how hot this thing is running, and when I had the full panel off, it seemed like the DME ran equally hot. It doesn't seem right. It also doesn't seem right that they both be mounted to the hottest metal panel in the car (firewall behind engine air flow).

mskar

Here is the TCU connector pinout document I created in Excel. Hopefully this will help someone out with PIN numbering. Let me know if I got anything wrong and I'll correct it

Note that there are two BMW Connectors plugged into the TCU, one gray and one black. Each of those connectors has two slim connectors that release from the shell and slide out, one gray and one black. The diagram I created is for ONE shell connector, with each slim connector labeled gray/black.

In terms of the large plug assemblies that plug into the TCU, those are th "A" and "B" connectors that you see in the table below the diagram. I hyper-linked the Pin # from the A/B tables to the connector diagrammed for quick reference.

mskar

The condition is getting worse, which may be helpful for diagnostics. Today on my way to my shop 15 miles away, I went into Transmission Emergency mode after less than a mile. Codes were:
P17D0 - Hydraulic Gear Selection: shift rod moved w/o contr. request (No fault symptom available)
5400 - Sport Mode Fault (no fault symptom available)

This happened multiple times, I kept clearing the codes, then disappeared and I could not reproduce the problem over about 30 miles of driving

At my shop, I again went over the wiring an connections then went on a nearly 2 hours spirited drive in the country, 90% of that in Sport+ mode. Shortly after switching to manual mode, I completely lost drive and had to pull over and reset, this happened about 3 times, codes:
P1990 - Positive Engine torque intervention not possible (no fault symptom available)
P1991- Engine torque intervention not possible (no fault symptom available)
then
P1764 - Engagement block, synchronization or gear skip (no fault symptom available)
P1733 - Shift Rod 3 displacement sensors (signal implausable)
P177D - 1st gear not disengageable (no fault symptom available)
P1854 - Software Error (no fault symptom available)

The appearance of the shift rod and P17xx codes are now pointing to the distance sensor. Not good. Distance sensor replacement is an expensive repair and I doubt any of my local shops have any experience doing them. I was preparing myself for dropping $2k on a valve body replacement, at least I could have done that one myself. This Cayman has turned out to be the biggest mistake of any car I've ever owned, and I've owned a lot of cars.

Editorial:
I can't enumerate how many times I read, while researching the 987.2, how "bulletproof" the PDK is. This is otal nonsense. The PDK is a great shifting piece of technology that was not built to last. Yes, my car is 14 years old, but it has less then 55k miles on it. I have spent nearly 6 months now reading forum post after forum post about PDK problems, none of them straightforward, most poor owners like myself trying to figure out WTF happened to their car. Since Por$che thought it would be funny to make a disposable transmission, as owners we are forced to go to the aftermarket for parts and service. My experiences at the track have been similar, meeting people who told me how many weekends they lost to their PDK breaking down in mysterious ways. The only guy I've talked to who was eventually able to race his Cayman again had to shell out over 12k for a replacement PDK.

I have a track weekend next weekend, it looks like I will be bringing my 35 year old 928S4, the car I had planned on retiring because I didn't want to break it. Oh the irony.

clib

my experience exactly. The most frustrating was the wasted time and money and the porsche dealer inability to definitively diagnose the issue. Great cars but driven hard including some track use and i would guess pdk failures are not uncommon with higher mileage. So many never accumulate many miles the user experience is biased. Mine has been great after the new unit

jazztrip

@mskar I apologize as I've only skimmed the previous posts in this thread, but I wanted to share a recent issue with the PDK on my 987.2 S track car in the hope it's helpful to you. In September of '22, out of the blue, I got a Transmission Emergency error with the codes in the attached photo. The car would not move and the gear position indicator on the dashboard was not showing the selected position. I had the car towed to my shop and they narrowed the issue down to the position sensor unit. They replaced it and my car has been fine since, including five or six track days and about 1,800 road miles getting to the track(s). It wasn't cheap, but it worked. Best of luck getting your issue sorted out.

BadsSgt

Guys My Panamera is now running again. I paid a fellow 650 usd tonight to remote into my piwis III only to find that the seller did not set up my piwis III correctly he had to flash new stuff into my unit. after he did it calibrated thru the whole process my transmission errors are gone and my car now runs again. I am pretty pissied at autonumen.com right now because i told them it had to be something with there software and they told me they would only look at it if I paid for a tech to remote in and look. they are selling these like this on purpose to upcharge for more money. the driver light is out because i had the lower fender out looking for the transmission control unit. ill fix it tomorrow. its freakling calibrated and running now thats all i care about I can share my remote guys info if you like we agreed on 500 usd. i paid him 650 it was worth it.

mskar

The kicker is the "higher mileage thing". My car as 52,700 miles on it. It appears that the PO was a Silicon Valley guy who used the car for commuting and maintained it at every service interval. My current thinking (since a 173x code FINALLY popped up, is that the gear oil overheated on my first (50 degree) track weekend and cooked the distance sensor. That is the real birch of it, the amount of money you waste on towing, the few "cheaper" parts you figure you can "try", and the labor. Thankfully I can do my own labor, this distance sensor would be pretty expensive to have installed by an indy, and I doubt many of them have experience doing these repairs. Thanks for sharing your experience!

mskar

I can only WISH that my car was considerate enough to throw a 173x code to begin with, it has taken 5 months, two wasted track weekends, and two trailored rides back from VIR (one which damaged the roof a bit) for it to finally throw a P1733 (same as yours but 3rd gear). The 18B0 w/o a P173x cross references to TCU error according to Porsche's docs (I now have two programmed TCUs, both of which generate the same errors), and I cant find ANYTHING intelligible on P1990 and P1991 (except in the same "bad tcu" Porsche doc). I have a T-Design distance sensor in hand, I'm back from the track on Monday and have the day off so I'm going to start that project. I'm adding an LNEngineering Gear Oil cooler to keep my LNEngineering PDK cooler company. Thanks for sending your codes, P18B0 has been a CONSTANT.