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.
Cause could be as simple as a weak solder joint for one of the components on the sensor. Constant heating/cooling, expansion and contraction. Add the harsh environment of an automobile. Could explain the intermittent nature before it fully fails.
You are absolutely right, all of this is speculation which is a lot of fun, but we just don't know. Plus (Sandwedge's experience notwithstanding) the failure rate on these must be pretty low when put in context of the number of PDKs out there.
It can't be made failure-proof but we can take prudent precautions to reduce failure likelihood. These include high temp rating components, good workmanship during assembly, but most importantly a simplified design. If every solder joint is a potential single-point failure then it makes sense to have as few solder joints as possible. Same with the components. That's the basis of reliability analysis when you have parts in a chain; the failure rate of one part is multiplied by the failure rate of the next part and so on down the chain. The more parts in the chain the higher the failure rate, and it can grow quickly.
That's why my approach is to provide a simplified, inexpensive replacement coupled with the knowledge needed to minimize repair costs. We can make the part as reliable as possible but it will never be perfect.
Right now we are in a situation where people are grateful replacement was "only" $6k when compared to a $20k dealership quote. That's the wrong way to think about it IMO. If it costs $200 to make the part, and eight hours to install it, that's a $2k repair maximum. What the dealer quotes to replace the transmission is irrelevant as it's a false comparison. The dealer (Porsche actually) has intentionally distorted the market by causing a false scarcity.
Anyways, this is all handwaving without a replacement sensor but it does keep the thread alive while waiting for parts. I've now been waiting for 11 days for my circuit boards to travel from Oregon to Southern California, the damn post office misrouted them to Texas. It's like the country can't do the simplest things anymore.
That's why my approach is to provide a simplified, inexpensive replacement coupled with the knowledge needed to minimize repair costs. We can make the part as reliable as possible but it will never be perfect.
No disagreement on that. My take on what would seem like a simple solution to the expensive PDK replacements for people out of warranty, no matter how rare they are would be to allow dealership mechanics to open the PDK and work on them as they do when they take an engine apart. Which they do everyday all over the world. Internal engine parts needed, readily available.
Last I heard, Porsche or indy mechanics still...14 years after the PDK was rolled out are very restricted to what they're allowed to do with a problematic PDK. Completely taking it apart is off limits. Again, last I heard. Could have changed. Which is why Porsche won't provide internal parts for the PDK even to their own dealerships. The "logic" being....."what do you need that part for. You're not supposed to be in there to begin with".
So the story continues. Any internal PDK issue with a few exceptions that can be reset without taking it apart, the whole PDK is shipped to Germany and a rebuilt unit is sent back. As previously said, my last replacement on the GTS came to $20,000, covered by extended warranty. All due to a failed distance sensor according the fault code the dealership pulled.
It can't be made failure-proof but we can take prudent precautions to reduce failure likelihood. These include high temp rating components, good workmanship during assembly, but most importantly a simplified design. If every solder joint is a potential single-point failure then it makes sense to have as few solder joints as possible. Same with the components. That's the basis of reliability analysis when you have parts in a chain; the failure rate of one part is multiplied by the failure rate of the next part and so on down the chain. The more parts in the chain the higher the failure rate, and it can grow quickly.
That's why my approach is to provide a simplified, inexpensive replacement coupled with the knowledge needed to minimize repair costs. We can make the part as reliable as possible but it will never be perfect.
I have full confidence in your approach. There are no guarantees in life, at least when it comes to these things but sourcing high quality components and using proper installation techniques at least raises the confidence level in the end product. I guarantee you have put more thought and effort than the engineers who originally designed the DS whether they be ZF or one of their vendors. In the end for mass production, it's who can supply us with the cheapest parts that meet our spec. You have the same limitations but on a smaller more deliberate scale, you can also be more selective with your components cause there's no bean counters looking over your shoulder. Well, maybe your wife.
I'm also happy you found and went with the aluminum housing. I'm leery of plastics for long term use, especially when they can't be inspected. While simulated testing can be done, there's really no substitute for thousands of heat cycles, vibrations, etc. over years of actual use. I've just seen too many plastic parts over long terms get old, brittle, and fail in the field. All that said, the field testing by @jjrichar is doing is impressive. The cricket bat takes the cake. The phrase "takes a lickin and keeps on tickin" comes to mind. That's what high quality parts and production techniques get you.
With less kind words I'm still waiting for my updated circuit boards after now waiting 13 days since they were shipped from Oregon (800 miles away). Taking pot shots at the post office is like shooting fish in a barrel, but what a comedy of errors. The package was first misrouted to Texas, then once they figured that out it got shipped to LA. It's now been in LA for four days, first being sent to the regional distribution center, then out of it, then back to it, then out of it again, and finally this morning it's supposedly in Palos Verdes (where I live and only 25 miles from the LA center). The post office has changed the estimated delivery date three times and now they aren't even listing one anymore (I don't blame them).
With any luck I'll get them Monday and have two updated prototypes assembled and programmed within a few days. Assuming they weren't damaged in all of this.
I also got the 3D printed test articles using the machined aluminum CAD model and they look pretty good. I'll post photos soon.
Shipping anything these days can be frustrating. The lead role in the comedy of errors gets traded back and forth, so it's hard to predict who will provide the most reliable service at any given time. I've seen the same incompetence from FedEx more than once over the past couple of years. Overnight packages being routed far out of their way, then tossed back and forth between hubs like a radioactive potato, followed by a random walk that ends in some shrubbery in the addressee's general vicinity.
UPS seems to have their act together, by comparison... at least at the moment.
Have you noticed that carriers are now regularly logging deliveries as “attempted- but no one at home. Will try again next day”, when in fact my home office overlooks the front driveway and my cameras didn’t record anything coming up it apart from stray dogs and cats!!
Cheeky beggars !! - the carriers AND the animals 😀
Taking pot shots at the post office is like shooting fish in a barrel, but what a comedy of errors. The package was first misrouted to Texas, then once they figured that out it got shipped to LA. It's now been in LA for four days, first being sent to the regional distribution center, then out of it, then back to it, then out of it again, and finally this morning it's supposedly in Palos Verdes (where I live and only 25 miles from the LA center). The post office has changed the estimated delivery date three times and now they aren't even listing one anymore (I don't blame them).
With any luck I'll get them Monday and have two updated prototypes assembled and programmed within a few days. Assuming they weren't damaged in all of this.
Originally Posted by Noah Fect
UPS seems to have their act together, by comparison... at least at the moment.
Off topic of course but since it's been started I'll follow up. As PV997 says, the US post office has been a sad joke for decades now. If it was a private business we probably wouldn't have heard about them since decades ago. But credit where it's due. As Noah says, UPS seems to have it together and I have to give credit to Amazon also. Never had a problem with them. Often the item purchased arrives earlier than promised.
A new photos of the test housings I got from Xometry on Friday. These are printed glass-reinforced nylon but they were fabricated using the CNC machined CAD file that we would use for aluminum housings. Have to say I'm pretty happy with how they came out and they validate our CAD (which we wanted to do prior to placing a CNC order). Factory sensors are in the background for comparison.
Next step is to put the updated PWB in incorporating all of our lessons learned from the first prototype. Should have them potted and working in a few days.
Thanks @LynxStarAuto and I get it. It's really easy to fall into the penny-wise pound foolish trap. The goal here is not to squeeze every last nickel out of the cost and a lot of what we throw out is just spitballing in hopes that knowledgeable people like yourself will chime in. That's how we all learn.
All that being said, the aftermarket parts are around $2k and I know for a fact they are about $100 in material and maybe two hours labor. On any other normal car it would be a $500 part.
Plus what's the reasonable labor to replace one of these things? Like I said we saw 28 hours quoted and that sounds crazy, maybe ten hours sounds about right to me. There's no book time on this so do shops just get to make it up? I know there's a lot of overhead in running a shop but that gets rolled into the burdened labor rate. Shops don't get to pad the hours on top of it.
You seem like a straight-shooter and I really appreciate your willingness to engage. None of us think it's fair to compare DIY to the costs of an honest shop. What we are poking at is why some people seem to be getting quoted really high hours and why the damn part is so expensive. Seems like the prices are being set compared to a PDK replacement rather than the actual cost plus a reasonable profit. That's what we want to change.
There is no official repair time for replacing the distance sensor. No book time unfortunately. Interestingly enough, there IS a labor line, but no information available. I usually just quote dropping the transmission assembly, and reinstalling. That pays 13.6 hours. The labor time includes draining, filling, and running all procedures. So there is no reason why a shop would be charging additional for that. However, they are within their rights to do so, if they choose to. That’s part of the free market right? As a consumer, you have the right to decline and take it somewhere else. Or you can always discuss with the shop to see if they can work on the price. Communication is key on both sides. Being informed is always your best weapon. For example, on a 718/981 you can R&R the sensor with the unit in the vehicle. However, I also drop the unit out of those to do this. Simply put, it’s easier for me to do so. I’m 39, it’s easier on the body and my body is not getting any younger. Lol also, it’s just safer to do so if I have one of my guys work it on the bench, than it dangling on the car, on the lift. It also gives me a straight shot at the rear main seal, which maybe it’s 996 PTSD, but I’m always wanting to inspect that area when the opportunity arises to ensure everything is copacetic. So even if it’s possible to do it with unit in vehicle, I will drop it, and replace the sensor the traditional way. As such I would charge accordingly. Iirc, it pays slightly less to drop the unit on those. I think 11.8 hours or so.
Of course there is such thing as unscrupulous practices. 28 hours to R&R one of these is on the extreme side. I personally, wouldn’t have the courage to quote something that high. That’s almost charging to do it twice. Which is interesting, perhaps said shop is providing a warranty for this experimental part? And so they figure charge double to cover our ***. No idea, there are many different ways to approach it. Me personally, I charge the book time. If I do it in 6 hours, well I’m just really good at my job. If I do it in 15 hours, well I need to get better. I let the customer know this is the roll of the dice. I can guarantee the transmission will come out, open up, close, and reinstall to factory spec, but the sensor we are all figuring out. Again, communication is key as always, which seems to be a BIG problem with shops, and humans in general in this artificial blue screen device era. In Florida, we have very strict laws regarding labor, and how it’s quoted on an RO. Although it’s hard to enforce, but consumer affairs does have some requirements, which must be abided by yearly when we renew our licenses and certs. If you have a positive reputation with the city, they will extend you an biannual inspection, vs annually. Unfortunately, every state is different. But as a rule of thumb, I always encourage consumers to ask for itemized invoices. With all the part numbers, labor lines, etc. I’ve included the actual quote for this customer here for reference. I add labor lines for the pan and filter, but this is strictly for CARFAX reporting purposes so I deduct the labor times for those individuals items from the total time, and quote as 2 separate lines. However, the labor remains the same 13.6 hours.
The majority of electrical failures I’ve seen in my time as a mech is usually chalked up to improper assembly during the manufacturing process. Cold solder joints, loose or poor fitment on the circuit board. Other times it’s installation error as well. Wiring harness too taut, not enough insulation. Routing around sharp edges, or tight clearances causing chafing. Things of that nature. Heat and contamination I would say is the close second.
Can someone explain to me why book time is even necessary? I'm not aware of any other industry that uses book time outside of the motor vehicle repair business. I don't understand how a business that charges an hourly rate, charges that hourly rate, unless an arbitrary piece of paper says they should or can charge more. Why isn't it time and materials like every other hourly business. I'm in the repair business as well, and we don't have a "book". If it takes 3 hours and $X.00 in parts, we charge 3 hours labor plus the parts X 1.5 markup. We document our time. 8:00 am- 11:00 am completion.
Like I said, I'm in the repair business as well and I understand that things go sideways. I get that some things are experimental, like this distance sensor. I would explain to the customer that there's going to be a learning curve and maybe a few hiccups. We will bill accordingly. What I do know is, if I told a customer that something was going to take 13 hours of labor but I left after 6, the customer would correctly understand it really was a 6 hour job.
When something is new or a new process, then the customer should understand that there will be some extra time involved. But.....after the shop has a few under their belt, there should be some efficiencies gained in that repair process. I've been in my field a while. I know what does and doesn't need to come apart to effect repairs. We don't use a book to determine that I'm good at my job. When I'm good and efficient, I never run out of work because my customers know they're getting a value. My phone keeps ringing. Plus, charging 28 hours for a 13 hour book job isn't the free market, that's a rip off. That's preying on customers naïveté
Of course there is such thing as unscrupulous practices. 28 hours to R&R one of these is on the extreme side. I personally, wouldn’t have the courage to quote something that high. That’s almost charging to do it twice. Which is interesting, perhaps said shop is providing a warranty for this experimental part? And so they figure charge double to cover our ***. No idea, there are many different ways to approach it.
Given what the dealerships are charging for a failed PDK I guess it's open season for anyone who can do it for substantially less than $20,000. I don't know what the price is for the after market distance sensor but even at 28 hours at $150/hour you're at $4,200. Add the cost of the sensor and I'm only guessing here but you must be miles away from $20,000. Most likely less than half of that and likely less. And with your pricing and just guessing again, price of the the new sensor being $1,000 you'll be well below $5,000.
If I had another failure I think my biggest concern about the indy fix with a so far unproven after market duplicate sensor would be my biggest concern. As you say though, a warranty on parts and labor on an after market indy sensor job would no doubt add confidence to the process. Needless to say, the reputation and longevity of the indy doing the work would play a roll too as would the language of the warranty.
Just finished two new prototype distance sensors ready for in-PDK tests. These use printed GF nylon housings (instead of aluminum which is coming) but other than that are very close to what we think is the final incarnation. (Note molex connectors are for testing only and won’t be on actual sensors.)
Factory sensor in the background for comparison. I had a little trouble using the high temp potting for the first time as it doesn't spread as easily and is more susceptible to tiny bubbles, so there's a learning curve here. Plus these nylon housings stain and you can't just wipe the potting off if some gets on it. That problem will go away with the aluminum housings. Regardless those are cosmetic concerns, not functional.
Lots of improvements from what we learned on prototype #1:
.
Modified the HE sensor angle to minimize magnet height sensitivity and adjacent magnet interference. This is probably the most significant change as it significantly reduces variability.
Moved sensor DS3 to the opposite side to replicate the factory sensor inversion. This had been inverted via programming on the first prototype, but I wasn’t happy with the range.
Reduced sensor gain to better match factory sensor performance. The first prototype was consistently reading a few mm too far regarding magnet travel and caused us to fault on Reverse and intermittently on Third.
Set sensor clamping levels to never exceed 12-88% when operating. We learned on the first prototype the TCU considers values outside this range implausible and faults if it sees it.
Plus some changes to better match the final configuration:
.
Used the CNC machined housing design (still printed GF nylon for now but the form factor matches what we’ll use for aluminum).
Switched to 22 gauge, high strand count, silicone insulation wiring rated at 200 deg C
Switched to 832HT high temperature potting rated at 225 deg C
We came in just under $100 in material each on these and about 3 hours assembly time each. Working on some process improvements to get that down to our two hour goal.
Next is mailing them to @jjrichar and @stjoh for testing in a 981.1 and 997.2 PDK respectively. If anyone has a PDK on a bench in So Cal and wants to help please PM me. I’m tuning these in a simulated setup and we could really speed things up if I could bolt one into a PDK and take some manual measurements.