Guide to Repairing a PDK Transmission
#286
This is good info. Thanks for sharing and for the awesome work researching and putting the info together for the community.
Maybe putting up a web form to capture failure data would be helpful in getting more attention. I'm willing to see what our options are. "Porsche PDK failure log" or something of the sort, socialized among PCA, forum, and FB members. Its disturbing to think that possibly "6% in first 100k miles and 27% by 200k miles" fail and most could be resolved at much lower cost. If Porsche wants to maintain its reputation for having a high percentage of vehicles remain on the road, they need to address this. At the bottom of their depreciation curve a lot of these cars will become a parts bin if they have a PDK failure. We really need to reverse engineer the distance sensor and establish a 3rd party source. I wonder if the same or similar part is sold for DCT/DSGs under different part #'s without the exclusivity agreements.
Maybe putting up a web form to capture failure data would be helpful in getting more attention. I'm willing to see what our options are. "Porsche PDK failure log" or something of the sort, socialized among PCA, forum, and FB members. Its disturbing to think that possibly "6% in first 100k miles and 27% by 200k miles" fail and most could be resolved at much lower cost. If Porsche wants to maintain its reputation for having a high percentage of vehicles remain on the road, they need to address this. At the bottom of their depreciation curve a lot of these cars will become a parts bin if they have a PDK failure. We really need to reverse engineer the distance sensor and establish a 3rd party source. I wonder if the same or similar part is sold for DCT/DSGs under different part #'s without the exclusivity agreements.
The following 3 users liked this post by David Barber:
#287
Good read, that's related to DCT/DSG tranny from VAG..
https://www.team-bhp.com/forum/techn...t-gearbox.html
I spoke to a Porsche contact he suggested that PDK failure rate on 911/Cayman is around 6% in first 100k miles and 27% by 200k miles.
https://www.team-bhp.com/forum/techn...t-gearbox.html
I spoke to a Porsche contact he suggested that PDK failure rate on 911/Cayman is around 6% in first 100k miles and 27% by 200k miles.
Yeah I might need a clutch job but at least there aren't random $5 sensors that will make the car undriveable and cost $20k to fix.
I would love to help make a fix/replacement/rebuild of the sensor. Im a Mechanical Engineer but I have 10 years in controls and electronics for work and Im familiar with hall effect sensors, they are simple devices. And the temp sensors are even simpler if they are normal thermocouples.
How do we get our hands on some of these distance sensors?
#288
"How do we get our hands on some of these distance sensors?"
Maybe we need a trusted Indy that's willing to pull apart a used PDK and partner on the R&D. In terms of cost, a few of us could pitch in and find something through PCA or buy the cheapest eBay PDK available (which is still a couple grand). It wouldn't be cheap, but it wouldn't be absurdly expensive if we have enough donors. And we could eventually recoup the cost. I don't know what kind of IP battle we'd be fighting if we actually developed (or had developed) an after-market product. Repair instructions and parts sourcing might be a better option.
This makes me worry about my little 987.2, and I was really starting to warm up to the PDK...time for an oil cooler I guess. I work in tech and also have a little experience in automation and controls.. I can't imagine fixing the sensor array would be terribly difficult if we can get parts.
Maybe we need a trusted Indy that's willing to pull apart a used PDK and partner on the R&D. In terms of cost, a few of us could pitch in and find something through PCA or buy the cheapest eBay PDK available (which is still a couple grand). It wouldn't be cheap, but it wouldn't be absurdly expensive if we have enough donors. And we could eventually recoup the cost. I don't know what kind of IP battle we'd be fighting if we actually developed (or had developed) an after-market product. Repair instructions and parts sourcing might be a better option.
This makes me worry about my little 987.2, and I was really starting to warm up to the PDK...time for an oil cooler I guess. I work in tech and also have a little experience in automation and controls.. I can't imagine fixing the sensor array would be terribly difficult if we can get parts.
#289
The gear oil diagnostics has been done, attached. Since the gear oil change I have not had any codes. I have not had a chance to run any DE laps or travel more than 45 minutes. I'll report after a longer road trip or any code occurrence.
#290
Three Wheelin'
Thread Starter
"How do we get our hands on some of these distance sensors?"
Maybe we need a trusted Indy that's willing to pull apart a used PDK and partner on the R&D. In terms of cost, a few of us could pitch in and find something through PCA or buy the cheapest eBay PDK available (which is still a couple grand). It wouldn't be cheap, but it wouldn't be absurdly expensive if we have enough donors. And we could eventually recoup the cost. I don't know what kind of IP battle we'd be fighting if we actually developed (or had developed) an after-market product. Repair instructions and parts sourcing might be a better option.
This makes me worry about my little 987.2, and I was really starting to warm up to the PDK...time for an oil cooler I guess. I work in tech and also have a little experience in automation and controls.. I can't imagine fixing the sensor array would be terribly difficult if we can get parts.
Maybe we need a trusted Indy that's willing to pull apart a used PDK and partner on the R&D. In terms of cost, a few of us could pitch in and find something through PCA or buy the cheapest eBay PDK available (which is still a couple grand). It wouldn't be cheap, but it wouldn't be absurdly expensive if we have enough donors. And we could eventually recoup the cost. I don't know what kind of IP battle we'd be fighting if we actually developed (or had developed) an after-market product. Repair instructions and parts sourcing might be a better option.
This makes me worry about my little 987.2, and I was really starting to warm up to the PDK...time for an oil cooler I guess. I work in tech and also have a little experience in automation and controls.. I can't imagine fixing the sensor array would be terribly difficult if we can get parts.
There was a failed PDK offered about 6 months ago for $1k OBO and I was tempted but didn't bite (no way I could justify to my wife, she puts up with enough already). Unfortunately it does not look like any cars other than the 911/Boxster/Cayman use this particular distance sensor. The Panamera PDK (different but also made by ZF) does have a replaceable distance sensor but it's package looks nothing like ours. I suspect it's internals are similar (same Hall effect sensor approach) so it would be worthwhile to dissect one of those if we can. A shop that repairs the Panamera PDK may have a failed one we can get. A new Panamera distance sensor is pricey though so it's not worth it to buy one IMO.
I think we just have to wait until one of these fails on the right person's car. I'm an engineer, DIY everything (including things I probably shouldn't), cheap, don't trust the work of others, and wouldn't take my car to a Porsche dealership to save my life, so I'd give it a try. But I don't think I'm the typical Porsche owner so we may be waiting a while.
The following 2 users liked this post by PV997:
Kpadas (09-17-2020),
subaru335i (08-25-2020)
#291
Three Wheelin'
Thread Starter
Keeping our fingers crossed and let us know how it goes.
#292
The 6% failure rate sounds high to me but I have no hard evidence. If you google "PDK transmission repair" this post if the first thing that comes up, which shows how little info is out there. If that many PDKs truly were failing I think we'd be hearing more about it as lots of desperate people would be posting (being told your car needs a $20k repair is a pretty good motivator to sign up for Rennlist). Could be true but it doesn't sound right.
There was a failed PDK offered about 6 months ago for $1k OBO and I was tempted but didn't bite (no way I could justify to my wife, she puts up with enough already). Unfortunately it does not look like any cars other than the 911/Boxster/Cayman use this particular distance sensor. The Panamera PDK (different but also made by ZF) does have a replaceable distance sensor but it's package looks nothing like ours. I suspect it's internals are similar (same Hall effect sensor approach) so it would be worthwhile to dissect one of those if we can. A shop that repairs the Panamera PDK may have a failed one we can get. A new Panamera distance sensor is pricey though so it's not worth it to buy one IMO.
I think we just have to wait until one of these fails on the right person's car. I'm an engineer, DIY everything (including things I probably shouldn't), cheap, don't trust the work of others, and wouldn't take my car to a Porsche dealership to save my life, so I'd give it a try. But I don't think I'm the typical Porsche owner so we may be waiting a while.
There was a failed PDK offered about 6 months ago for $1k OBO and I was tempted but didn't bite (no way I could justify to my wife, she puts up with enough already). Unfortunately it does not look like any cars other than the 911/Boxster/Cayman use this particular distance sensor. The Panamera PDK (different but also made by ZF) does have a replaceable distance sensor but it's package looks nothing like ours. I suspect it's internals are similar (same Hall effect sensor approach) so it would be worthwhile to dissect one of those if we can. A shop that repairs the Panamera PDK may have a failed one we can get. A new Panamera distance sensor is pricey though so it's not worth it to buy one IMO.
I think we just have to wait until one of these fails on the right person's car. I'm an engineer, DIY everything (including things I probably shouldn't), cheap, don't trust the work of others, and wouldn't take my car to a Porsche dealership to save my life, so I'd give it a try. But I don't think I'm the typical Porsche owner so we may be waiting a while.
Also a surprising amount of these threads have a lot of 991's so im not sure im convinced that the 991 gear oil cooler helped anything unfortunately.
#293
The 6% failure rate sounds high to me but I have no hard evidence. If you google "PDK transmission repair" this post if the first thing that comes up, which shows how little info is out there. If that many PDKs truly were failing I think we'd be hearing more about it as lots of desperate people would be posting (being told your car needs a $20k repair is a pretty good motivator to sign up for Rennlist). Could be true but it doesn't sound right.
There was a failed PDK offered about 6 months ago for $1k OBO and I was tempted but didn't bite (no way I could justify to my wife, she puts up with enough already). Unfortunately it does not look like any cars other than the 911/Boxster/Cayman use this particular distance sensor. The Panamera PDK (different but also made by ZF) does have a replaceable distance sensor but it's package looks nothing like ours. I suspect it's internals are similar (same Hall effect sensor approach) so it would be worthwhile to dissect one of those if we can. A shop that repairs the Panamera PDK may have a failed one we can get. A new Panamera distance sensor is pricey though so it's not worth it to buy one IMO.
I think we just have to wait until one of these fails on the right person's car. I'm an engineer, DIY everything (including things I probably shouldn't), cheap, don't trust the work of others, and wouldn't take my car to a Porsche dealership to save my life, so I'd give it a try. But I don't think I'm the typical Porsche owner so we may be waiting a while.
There was a failed PDK offered about 6 months ago for $1k OBO and I was tempted but didn't bite (no way I could justify to my wife, she puts up with enough already). Unfortunately it does not look like any cars other than the 911/Boxster/Cayman use this particular distance sensor. The Panamera PDK (different but also made by ZF) does have a replaceable distance sensor but it's package looks nothing like ours. I suspect it's internals are similar (same Hall effect sensor approach) so it would be worthwhile to dissect one of those if we can. A shop that repairs the Panamera PDK may have a failed one we can get. A new Panamera distance sensor is pricey though so it's not worth it to buy one IMO.
I think we just have to wait until one of these fails on the right person's car. I'm an engineer, DIY everything (including things I probably shouldn't), cheap, don't trust the work of others, and wouldn't take my car to a Porsche dealership to save my life, so I'd give it a try. But I don't think I'm the typical Porsche owner so we may be waiting a while.
Jeff
The following 4 users liked this post by jbaker136:
#294
Three Wheelin'
Thread Starter
I bought the failed 997.2 C4S PDK for exactly this purpose (not an easy one to explain to the wife!). Once track season is over, I plan to tear it down. I would be happy to collaborate with you and others who are interested on the teardown and failure evaluation. PM me and we can discuss the best approach. Thanks
Jeff
Jeff
The following 3 users liked this post by PV997:
#295
Pro
Just to clarify the 6% was based on 997 and 987 cars (09-12) in Canada. Total sales figure was something around 3680 cars for period of 4 years, 220 replacements.
With 17 dealers in Canada at the time, it works out to 3 replacements per year per location. Hardly noticeable.
With 17 dealers in Canada at the time, it works out to 3 replacements per year per location. Hardly noticeable.
#296
I bought the failed 997.2 C4S PDK for exactly this purpose (not an easy one to explain to the wife!). Once track season is over, I plan to tear it down. I would be happy to collaborate with you and others who are interested on the teardown and failure evaluation. PM me and we can discuss the best approach. Thanks
Jeff
Jeff
#297
Jeff
#298
Three Wheelin'
Thread Starter
Decided to take some measurements on the distance sensor output to help figure out the internal components. I did this by accessing the TCU (a real PITA as Mr. Hobbiest noted above) and measuring the signal with an oscilloscope. Against my better judgement I pierced the distance sensor wires with a pin so I could connect the o-scope while still keeping the circuit intact.
As expected, the output signal was pulse width modulation (PWM) with a level from around 0 to 5 volts. For those who aren't familiar with PWM it's a square wave where the width of the pulse encodes the information (i.e. the shift rod location). The period from one pulse to the next stays constant whereas the time length of the pulse (the duty factor) varies depending upon the information being transmitted. PWM is used as it's largely immune to noise unlike a simple analog signal.
Here's a plot showing a typical Hall-effect sensor PWM scheme. The short duty factor on the left corresponds to one travel extreme, the long duty factor on the right corresponds to the other extreme.
The PWM signal starts immediately once the ignition key is switched to on, there is no initialization sequence or two-way communication. The carrier frequency of the PWM signal was 1 kHz (I actually measured 1.05 kHz but there may have been some error) and the duty factor ranges from around 10% to 90% depending upon the shift rod location. Pretty much a plain vanilla signaling scheme.
I've been researching Hall-effect sensor ICs with PWM and found a couple that match these specs and were available back when the PDK was introduced. The most likely candidate is a programmable chip so we'll have to figure out a few parameters ZF used to match them. Programming is pretty simple but it does complicate things. More to come as it's figured out.
As expected, the output signal was pulse width modulation (PWM) with a level from around 0 to 5 volts. For those who aren't familiar with PWM it's a square wave where the width of the pulse encodes the information (i.e. the shift rod location). The period from one pulse to the next stays constant whereas the time length of the pulse (the duty factor) varies depending upon the information being transmitted. PWM is used as it's largely immune to noise unlike a simple analog signal.
Here's a plot showing a typical Hall-effect sensor PWM scheme. The short duty factor on the left corresponds to one travel extreme, the long duty factor on the right corresponds to the other extreme.
The PWM signal starts immediately once the ignition key is switched to on, there is no initialization sequence or two-way communication. The carrier frequency of the PWM signal was 1 kHz (I actually measured 1.05 kHz but there may have been some error) and the duty factor ranges from around 10% to 90% depending upon the shift rod location. Pretty much a plain vanilla signaling scheme.
I've been researching Hall-effect sensor ICs with PWM and found a couple that match these specs and were available back when the PDK was introduced. The most likely candidate is a programmable chip so we'll have to figure out a few parameters ZF used to match them. Programming is pretty simple but it does complicate things. More to come as it's figured out.
Last edited by PV997; 08-30-2020 at 04:51 PM. Reason: added PWM plot
The following 5 users liked this post by PV997:
997ajk (08-30-2020),
Floyd540 (10-30-2020),
I am the Walrus (09-06-2020),
irnnr (08-03-2021),
mchrono (08-30-2020)
#299
.........Pretty much a plain vanilla signaling scheme.
I've been researching Hall-effect sensor ICs with PWM and found a couple that match these specs and were available back when the PDK was introduced. The most likely candidate is a programmable chip so we'll have to figure out a few parameters ZF used to match them. Programming is pretty simple but it does complicate things. More to come as it's figured out.
I've been researching Hall-effect sensor ICs with PWM and found a couple that match these specs and were available back when the PDK was introduced. The most likely candidate is a programmable chip so we'll have to figure out a few parameters ZF used to match them. Programming is pretty simple but it does complicate things. More to come as it's figured out.
Very interested to see where this goes!
Edit: Is there a better / more rugged sensor available today that was not available when the PDK was introduced?
Last edited by mchrono; 08-30-2020 at 07:43 PM.
#300
Three Wheelin'
Thread Starter
Thanks mchrono. I think we may be able to find a mil-spec part or a replacement with an extended temperature range, there do seem to be some available. The distance sensor signal is so simple we don't necessarily have to have the exact same part. It is a pretty dumb device, it simply outputs four PWM signals when powered, there's no handshaking or any other complex communication. BTW, the ICs I've found that match the distance sensor output are only about $2 each.
The following 4 users liked this post by PV997: