t-design

We sold quite a lot in last 2 month, most of sensors already installed. No new PDK control units were required on any of installations, and nobody reported any calibration issues.

Thats why I agree with PV997 - this is a unique case related to your gearbox and your PIWIS software. PIWIS verion/software/use being biggest suspect imo.

Niklas Vemdal

Mine is a month older. December 4th 2008. Transmission and vehicle is marked with December 16th 2008. It has done 95000 Miles, still transmission looks new inside.

Pinkawa

May 30th 2008 - 60000 mls

Kuro Neko

Any benefit in maintaining this?

amet

honestly, I don’t think so.

I can add at least 5-6 lines in there and we still won’t have any usable data. Sensor from my 991.1 is 2012.5 with 240,000 km ….

amet

I have installed 5, on 4 car started just fine and didn’t need any calibration. On one, that someone else played with before, we ended up having some trouble with 1831 error that was hard to clear… just wouldn’t go away. Took 2-3 “calibration with part replacement “ until it went away. In my head it was attributed to hydraulics needed to be primed and all previous errors cleared.

Kuro Neko

Yeah agreed, with 2012 in range, it seems as if this is a wider fault mode, possibly extending all the way through boxes with this same module...
Is it still used in present day models?

amet

I have not seen any 991.2 failures, only one 718 2017 cayman… my money is on 991.2 having the same sensor

PV997

Hi Pinkawa - When I was collecting data for the main post I found multiple claims that shops needed to replace the TCU after replacing the transmission with a junkyard PDK. There's also a comment way upthread from a guy who had his PDK replaced by Callas Rennsport (a respected shop in California) that told him the same thing and charged him for it. There are a few reports from Porsche dealerships doing the same. None of them had a good explanation for it but the shops told them that that using the old TCU often caused problems.

A few people stated they thought the TCU was somehow serial number matched to the PDK. Knowing the PDK itself has no smarts but just solenoids and sensors, I knew this could not be true. That's when I started digging into the whole TCU adaptation value tables and realized just how sophisticated this system is. The TCU uses a crude form of AI that adapts PDK commands to driver habits and transmission characteristics (shift points, speed of shifting, clutch engagement, etc.). These values are stored in the TCU in non-volatile memory and aren't cleared in normal operation. When I ran the VAL on my car I found them and there are literally thousands of hexadecimal codes that make-up the adaptation value tables.

I don't know for certain (and Porsche won't tell us) but I strongly suspect this is the reason some cars supposedly needed "new" TCUs. They really didn't need new ones, but the shops didn't understand the old adaptation values in the TCU were incompatible with the new PDK (particularly if the old adaptation values had been influenced by PDK malfunctions). They could have just cleared them but since no one really understands how this works (and Porsche won't tell us) they just tried a new TCU and it fixed the problem.

This is informed speculation rather than fact but it makes sense. I think amet is exactly right that using "calibration after part replacement" will likely set the TCU back to the state of being a "new" TCU or pretty close to it. One way to check this would be to run a VAL with the PIWIS both before and after running the calibration. This would not only show if the adaptation values changed but also if the software version changed. If anyone does run this check please let us know how the VAL changes. People providing these little puzzle pieces are helping unravel the mysteries of the PDK despite Porsche's efforts to keep them from us.

Edit: Adding a note that this may or may not be the cause of Pinkawa's issue, but I'm putting it here for context as we are figuring this stuff out. Porsche and ZF never intended for us to do any of this stuff and there will be hiccups along the way. As we learn more I'll post it and make corrections as needed.

jbaker136

Given the vibration, heat, and fluid exposure inside the PDK, would mechanical crimping of any wire repairs, including installing the t-designs distance sensor, be preferred over solder joints?

Jeff

amet

IMHO, any soldered joint is better than crimped. Vibration can loosen up mechanically crimped ones, no way soldered joints are going to shake loose

Kuro Neko

I made the same observation, and T-Design suggested due to issues with incorrect crimps - as they require specific tolerances and tools - the soldering was preferred.
We are advised, the components used are well within their heat and chemical exposure limits, and this includes the T-Design provided heat shrink.

For what it's worth, my 356 had many soldered connections (those little bullet pins for example), and it was built in 1957.
Provided a soldered joint is made properly - flux, nice flow, and not cold of course, then heat shrinked - they can be resilient.

Time will tell, and with a least one recent T-Design part destined for the track, we have a viable test bed.
Notably as the number of devices shipping is increasing, and crimps of the proper type could be readily retrofitted if necessary.

The point now though, is the proof of concept with a large population of examples to verify overall process integrity...

Kuro Neko

Crimps are often preferred in high-vibration environments (aircraft, OEM auto, etc) as soldered joints can fracture under vibration, and crimps maintain flexibility.
Their tolerance in making though, as you imply, need to be maintained to remain effective (see above for T-Design view on the same).

A further point in favor of soldered here though, is simply these joints are in-line, and not butted to a fixed element.
Thus, in suspension are less likely to be subjected to direct stress if properly supported and not flapping about.

t-design

Exactly. The wire provides kind of suspension, so vibration should not be an issue. We feel like incorrect crimping could be much more error-prone.

jbaker136

Good to know. Thanks for the clarification.

Jeff