A $100 3D-printed PDK distance sensor?
#106
Three Wheelin'
Thread Starter
Update: you may remember this photo of the prototype distance sensor (foreground):
Well here it is now:
This is after jjrichar baked it at 350 F, froze it over night at -30 F (thermal test), strapped it to a running chainsaw for an hour (random vibe test), wacked it against the ground several times (shock test), then teed it up for a few dingers using a cricket bat (just gratuitous violence at this point). The nylon housing didn't do well which is why we are switching to machined aluminum, but once he plugged the harness back in (there's a connector there covered by potting) it still worked just fine on all four channels. It's painful seeing it in that condition but I'm happy with the results. We'll post a video of our "environmental tests" soon.
Edit: My mistake it was 350 F, not 400.
Well here it is now:
This is after jjrichar baked it at 350 F, froze it over night at -30 F (thermal test), strapped it to a running chainsaw for an hour (random vibe test), wacked it against the ground several times (shock test), then teed it up for a few dingers using a cricket bat (just gratuitous violence at this point). The nylon housing didn't do well which is why we are switching to machined aluminum, but once he plugged the harness back in (there's a connector there covered by potting) it still worked just fine on all four channels. It's painful seeing it in that condition but I'm happy with the results. We'll post a video of our "environmental tests" soon.
Edit: My mistake it was 350 F, not 400.
Last edited by PV997; 01-25-2023 at 09:23 AM.
#107
Video here. Electrical and in-transmission testing at the start. 'Environmental' testing at the end. Very impressed with what this thing can withstand. Besides smashing it apart with a hammer or putting it on the BBQ, I don't really know how to stop making it work.
Last edited by jjrichar; 01-25-2023 at 12:37 AM.
The following 3 users liked this post by jjrichar:
#108
That's one of the most entertaining phases of product development. Congratulations!!!!!
#110
Nordschleife Master
Video here. Electrical and in-transmission testing at the start. 'Environmental' testing at the end. Very impressed with what this thing can withstand. Besides smashing it apart with a hammer or putting it on the BBQ, I don't really know how to stop making it work.
https://youtu.be/ZBrJ3YUhHXk
https://youtu.be/ZBrJ3YUhHXk
Very impressive and I hope it ends up being a good solution. Having had two sensor failures, the only concern I have is how it will work with the rest of the electronics in the car. Unless I misunderstood the service advisor at the dealership, the problem in both my failure cases was not "mechanical" failures of the sensor itself but a failure of the sensor to communicate with the electronics in the car necessary for it to operate as designed.
I guess you could call that a mechanical failure but neither case had anything to do with any kind of physical damage to either sensor. Strictly electronical in nature. Basically, one part wasn't talking to the one it should have been talking to which caused both failures.
#111
Very impressive and I hope it ends up being a good solution. Having had two sensor failures, the only concern I have is how it will work with the rest of the electronics in the car. Unless I misunderstood the service advisor at the dealership, the problem in both my failure cases was not "mechanical" failures of the sensor itself but a failure of the sensor to communicate with the electronics in the car necessary for it to operate as designed.
I guess you could call that a mechanical failure but neither case had anything to do with any kind of physical damage to either sensor. Strictly electronical in nature. Basically, one part wasn't talking to the one it should have been talking to which caused both failures.
I guess you could call that a mechanical failure but neither case had anything to do with any kind of physical damage to either sensor. Strictly electronical in nature. Basically, one part wasn't talking to the one it should have been talking to which caused both failures.
1. Component to output failure to conduct.
2. Connections at either end fail to conduct.
3. Contamination shorting out signals.
4. Mechanical device being sensed is sticking.
But looking at a sealed unit, it would be nearly impossible to know one from the other.
For instance ... In the case of the three sensor replacements that are currently available, swapping the component has repaired the PDK. This could cover the first three items above. If the shift rods were hanging up (case 4) it is possible that this was accidentally fixed while accessing the senser. But the fix alone can't absolutely define specifically what caused the cure.
Nonetheless, it is comforting that we have choices. Even with the Porsche tax, it appears that a PDK repair now is less than the MT synchro repair occurring in higher mileage cars.
#112
Wondering how they came to this conclusion. Most failures that start as intermittent, leading to complete inoperation, are one of four things ...
1. Component to output failure to conduct.
2. Connections at either end fail to conduct.
3. Contamination shorting out signals.
4. Mechanical device being sensed is sticking.
But looking at a sealed unit, it would be nearly impossible to know one from the other.
For instance ... In the case of the three sensor replacements that are currently available, swapping the component has repaired the PDK. This could cover the first three items above. If the shift rods were hanging up (case 4) it is possible that this was accidentally fixed while accessing the senser. But the fix alone can't absolutely define specifically what caused the cure.
Nonetheless, it is comforting that we have choices. Even with the Porsche tax, it appears that a PDK repair now is less than the MT synchro repair occurring in higher mileage cars.
1. Component to output failure to conduct.
2. Connections at either end fail to conduct.
3. Contamination shorting out signals.
4. Mechanical device being sensed is sticking.
But looking at a sealed unit, it would be nearly impossible to know one from the other.
For instance ... In the case of the three sensor replacements that are currently available, swapping the component has repaired the PDK. This could cover the first three items above. If the shift rods were hanging up (case 4) it is possible that this was accidentally fixed while accessing the senser. But the fix alone can't absolutely define specifically what caused the cure.
Nonetheless, it is comforting that we have choices. Even with the Porsche tax, it appears that a PDK repair now is less than the MT synchro repair occurring in higher mileage cars.
#113
Three Wheelin'
Thread Starter
Wondering how they came to this conclusion. Most failures that start as intermittent, leading to complete inoperation, are one of four things ...
1. Component to output failure to conduct.
2. Connections at either end fail to conduct.
3. Contamination shorting out signals.
4. Mechanical device being sensed is sticking.
But looking at a sealed unit, it would be nearly impossible to know one from the other.
For instance ... In the case of the three sensor replacements that are currently available, swapping the component has repaired the PDK. This could cover the first three items above. If the shift rods were hanging up (case 4) it is possible that this was accidentally fixed while accessing the senser. But the fix alone can't absolutely define specifically what caused the cure.
Nonetheless, it is comforting that we have choices. Even with the Porsche tax, it appears that a PDK repair now is less than the MT synchro repair occurring in higher mileage cars.
1. Component to output failure to conduct.
2. Connections at either end fail to conduct.
3. Contamination shorting out signals.
4. Mechanical device being sensed is sticking.
But looking at a sealed unit, it would be nearly impossible to know one from the other.
For instance ... In the case of the three sensor replacements that are currently available, swapping the component has repaired the PDK. This could cover the first three items above. If the shift rods were hanging up (case 4) it is possible that this was accidentally fixed while accessing the senser. But the fix alone can't absolutely define specifically what caused the cure.
Nonetheless, it is comforting that we have choices. Even with the Porsche tax, it appears that a PDK repair now is less than the MT synchro repair occurring in higher mileage cars.
The sensor doesn't really "communicate" as most of us would think of it (e.g. like a CAN bus) but simply puts out four signals when ever it is powered on. There is no communication from the PDK back to it. Those four signals from the sensor are a square wave where the width of the pulse is proportional to the strength of the magnetic field seen. That's all that happens, the sensor is as dumb as they come.
Normally that square wave from the sensor will be between 12% and 88% duty factor. In the 12% case this means that for each pulse it will be on 12% of the time, then off 88% of the time, until the next pulse comes around. It repeats 1000 times per second. It does this whether the car is running or not as long as the key is on.
When the electronics in a channel fails to sense the magnetic field, it defaults to a ~95% duty factor. I've seen this failure on several sensors and it seems to be the most common signature. When the TCU sees this it throws a P1731-P1734 fault depending on which channel failed. If a channel just burns out it's comparable to having no power applied to the sensor and the TCU reads this as a 0% duty factor (no pulse present). It throws a fault there too but I think it's a different one (can't remember for sure).
That's really about all I've ever seen, maybe someone else has seen other failure modes. I've never heard of a magnet sticking and have a hard time visualizing how that could happen. Most all the failures I've seen are some sort of an electrical failure in the DS that causes it to not sense the magnetic field on one channel (stuck at 95% duty factor). The other three channels continue to work fine.
The following users liked this post:
festina-lente (01-26-2023)
#114
Three Wheelin'
Thread Starter
As a PDK owner I'm thankful for all the efforts currently underway by PV and others; been following these threads for 3+ years. I agree with Wayne the actual causes of this problem aren't totally clear. We can assume temperature (extreme heat, or cold) as chief elements impacting the service life of electronic components. We also can assume there are quality control issues during manufacturing of this part at ZF; especially when low mileage, non-tracked PDK cars fail. And of course, older electronic components are prone to failure at any time. Separately, I have been wondering for a while what impact does Sport+(any kind of PDK aftermarket tune) have on the longevity of a distance sensor. Are most of the PDK cars failing equipped with Sport Chrono/Sport+ package? I think Softronic was against PDK tuning contrary to Cobb and others. I know this opens a completely different can of worms but in my opinion relevant. What do you think?
What also makes me suspect this is that Porsche added a gear oil cooler to the PDK in 2013. IN ZF's 2008 white paper on the PDK it specifically said gear oil cooling was passive and made no mention of active cooling (as opposed to cooling the clutch fluid which was present in 2008). So something made them change their minds, it may have been lessons learned from early PDK warranty repairs.
Jjrichar has a 981 PDK (with a gear oil cooler) and he monitored gear oil temp with a PIWIS during a grueling run. It never got over 100 C which is low for automotive grade electronics, which are usually rated to 150 deg C. Unfortunately the 997.2 doesn't have a gear oil temp sensor (unlike the 981) so we can't measure it directly. I have a 2010 Turbo so I'm going to take it out and try and get good readings off the case using an IR thermometer. It will likely have some big error bars on the measurements though.
Bottom line is that if it is heat then 2009-2012 tracked cars are likely to be the most failure prone. In addition, the PDK is rotated 180 deg in the Cayman/Boxster compared to the 911, so gear oil pooling at one end of the case or the other could also play a roll.
In the sensor we are building we used the highest temp grade parts we could find, which is a 170 deg C junction temp for the HE sensor and 175 deg C for the capacitors.
#115
I really don't know the failure mechanism but have long suspected it's heat related. We know that in most cases an electronic component in a single channel fails (the other three channels work fine as does the rest of the transmission). This seems to me to be a classic single-point failure where a semiconductor just craps out. Semiconductor life is inversely proportional to the device's junction temperature and it will almost always determine when it fails barring some sort of manufacturing defect.
What also makes me suspect this is that Porsche added a gear oil cooler to the PDK in 2013. IN ZF's 2008 white paper on the PDK it specifically said gear oil cooling was passive and made no mention of active cooling (as opposed to cooling the clutch fluid which was present in 2008). So something made them change their minds, it may have been lessons learned from early PDK warranty repairs.
Jjrichar has a 981 PDK (with a gear oil cooler) and he monitored gear oil temp with a PIWIS during a grueling run. It never got over 100 C which is low for automotive grade electronics, which are usually rated to 150 deg C. Unfortunately the 997.2 doesn't have a gear oil temp sensor (unlike the 981) so we can't measure it directly. I have a 2010 Turbo so I'm going to take it out and try and get good readings off the case using an IR thermometer. It will likely have some big error bars on the measurements though.
Bottom line is that if it is heat then 2009-2012 tracked cars are likely to be the most failure prone. In addition, the PDK is rotated 180 deg in the Cayman/Boxster compared to the 911, so gear oil pooling at one end of the case or the other could also play a roll.
In the sensor we are building we used the highest temp grade parts we could find, which is a 170 deg C junction temp for the HE sensor and 175 deg C for the capacitors.
What also makes me suspect this is that Porsche added a gear oil cooler to the PDK in 2013. IN ZF's 2008 white paper on the PDK it specifically said gear oil cooling was passive and made no mention of active cooling (as opposed to cooling the clutch fluid which was present in 2008). So something made them change their minds, it may have been lessons learned from early PDK warranty repairs.
Jjrichar has a 981 PDK (with a gear oil cooler) and he monitored gear oil temp with a PIWIS during a grueling run. It never got over 100 C which is low for automotive grade electronics, which are usually rated to 150 deg C. Unfortunately the 997.2 doesn't have a gear oil temp sensor (unlike the 981) so we can't measure it directly. I have a 2010 Turbo so I'm going to take it out and try and get good readings off the case using an IR thermometer. It will likely have some big error bars on the measurements though.
Bottom line is that if it is heat then 2009-2012 tracked cars are likely to be the most failure prone. In addition, the PDK is rotated 180 deg in the Cayman/Boxster compared to the 911, so gear oil pooling at one end of the case or the other could also play a roll.
In the sensor we are building we used the highest temp grade parts we could find, which is a 170 deg C junction temp for the HE sensor and 175 deg C for the capacitors.
#116
Three Wheelin'
Thread Starter
Thanks for your explanation, very insightful. If in fact heat is the main culprit behind failures, can we assume adding a deeper aluminum transmission pan (better heat transfer than the plastic one?) with a higher oil capacity, is good insurance against higher transmission temps? Will lower transmission temps extend the life of the distance sensor?
There are aftermarket gear oil coolers than can be retrofitted to the 987 and 997:
https://lnengineering.com/bilt-racin...il-cooler.html
They are pricey and I wouldn't do it unless I regularly tracked my car, but that's just me.
An update on the prototype #2 in that my new PWB is due today and our new housings (printed but using the aluminum CNC profile) are due Monday. We've learned a huge amount of info from prototype #1, primarily regarding the importance of the sensor angle on the circuit board and gain/clamping limits in the programming.. By this time next week we should have them on their way for testing.
The following users liked this post:
festina-lente (01-26-2023)
#117
Thanks for your explanation, very insightful. If in fact heat is the main culprit behind failures, can we assume adding a deeper aluminum transmission pan (better heat transfer than the plastic one?) with a higher oil capacity, is good insurance against higher transmission temps? Will lower transmission temps extend the life of the distance sensor?
In the 9X7s, the outside casing will allow some cooling, and maybe some heat is conducted into the clutch/valve body chamber and then dissipated via clutch fluid cooling.
The results when I did the hard runs the other day were a bit counter-intuitive. Here's a synopsis of what I did and the results from a 981 BS that has the gear oil cooler.
Something to understand is that regardless of whether sport/sport plus mode is selected, when you jump on the gas the DME changes the thermostat 'set point' temp to 85 deg (it's actually 84.75, but 85 is good enough) from the 105 deg that is normal. It will stay this way for a few minutes after you take it easy and it thinks 85 deg isn't required anymore. The set point is the temp the DME is aiming to achieve at the coolant temp sensor, which is after the coolant has passed through the engine. The temp of the coolant as it leaves the thermostat/pump and enters the engine is going to be cooler. How much cooler? It depends on how much you are wailing on the engine. If you are thrashing the car the coolant entering the engine is far cooler than when you are cruising, as it's just trying to keep the temp leaving the engine 85 deg. The same coolant that is leaving the pump and going to the engine is also going to the engine oil and transmission coolers. So when you are wailing on it the temp going to the coolers is going to be quite low.
Additionally, at high speed the gear oil flow rate through the cooler is high, and at high engine RPM the coolant flow rate through the cooler is high. So if you are driving hard/fast the ability of the system to cool is far better.
The hard runs are on a local freeway that has U turn bays 400 m apart in a particular location. WOT accel to 110 kph, which is the speed limit. U turn, same again over and over until temps fully stabilise. Clearly not on the track but as hard as I've driven the car. Whilst doing the runs I monitored coolant temp, radiator outlet temp, set point temp, engine oil temp, clutch and gear oil temps.
The hottest the gear oil got to when I was beating on it was 86 deg. Being only a bit over 2L of gear oil, it seems the cooler flow is very effective at keeping the temp low when the flow rate of coolant and oil is high and the coolant temp entering the cooler is low.
Ironically the gear oil temp got to 91 deg when I went back to cruising and the set point temp returned to 105. 91 deg on the gear oil was the highest I saw at any time. When I thought about it, this make sense. The set point temp is high, and the coolant temp entering the coolers is high because you aren't driving hard. So the gear oil doesn't get the cooling that it does like when you are wailing.
Highest the clutch fluid got to was 98 and same for the engine oil. When I had reached these maximum temps the instrument cluster was reading 104. No idea why this is. The instrument cluster temp seems to be a mix of all three oil temps from the testing I've done previously, but the calcs for the final number have always been a bit of a mystery.
So what do I get from all of this? The system is very good at cooling the gear oil when driving hard. There's only a bit over 2L of fluid and combined with a low set point temp, coolant with a low temp entering the cooler, high flow rate of oil and coolant through the cooler it keeps the temp surprisingly low.
If there are failures of the distance sensor in the 9X1 series of cars, especially street cars, I would be very surprised it was due to excessive temp.
Last edited by jjrichar; 01-27-2023 at 01:23 AM.
The following users liked this post:
Noah Fect (01-28-2023)
#118
The pan is for the clutch fluid, which is cooled. I suspect changing this to aluminium will have little effect on gear oil temp.
In the 9X7s, the outside casing will allow some cooling, and maybe some heat is conducted into the clutch/valve body chamber and then dissipated via clutch fluid cooling.
The results when I did the hard runs the other day were a bit counter-intuitive. Here's a synopsis of what I did and the results from a 981 BS that has the gear oil cooler.
Something to understand is that regardless of whether sport/sport plus mode is selected, when you jump on the gas the DME changes the thermostat 'set point' temp to 85 deg (it's actually 84.75, but 85 is good enough) from the 105 deg that is normal. It will stay this way for a few minutes after you take it easy and it thinks 85 deg isn't required anymore. The set point is the temp the DME is aiming to achieve at the coolant temp sensor, which is after the coolant has passed through the engine. The temp of the coolant as it leaves the thermostat/pump and enters the engine is going to be cooler. How much cooler? It depends on how much you are wailing on the engine. If you are thrashing the car the coolant entering the engine is far cooler than when you are cruising, as it's just trying to keep the temp leaving the engine 85 deg. The same coolant that is leaving the pump and going to the engine is also going to the engine oil and transmission coolers. So when you are wailing on it the temp going to the coolers is going to be quite low.
Additionally, at high speed the gear oil flow rate through the cooler is high, and at high engine RPM the coolant flow rate through the cooler is high. So if you are driving hard/fast the ability of the system to cool is far better.
The hard runs are on a local freeway that has U turn bays 400 m apart in a particular location. WOT accel to 110 kph, which is the speed limit. U turn, same again over and over until temps fully stabilise. Clearly not on the track but as hard as I've driven the car. Whilst doing the runs I monitored coolant temp, radiator outlet temp, set point temp, engine oil temp, clutch and gear oil temps.
The hottest the gear oil got to when I was beating on it was 86 deg. Being only a bit over 2L of gear oil, it seems the cooler flow is very effective at keeping the temp low when the flow rate of coolant and oil is high and the coolant temp entering the cooler is low.
Ironically the gear oil temp got to 91 deg when I went back to cruising and the set point temp returned to 105. 91 deg on the gear oil was the highest I saw at any time. When I thought about it, this make sense. The set point temp is high, and the coolant temp entering the coolers is high because you aren't driving hard. So the gear oil doesn't get the cooling that it does like when you are wailing.
Highest the clutch fluid got to was 98 and same for the engine oil. When I had reached these maximum temps the instrument cluster was reading 104. No idea why this is. The instrument cluster temp seems to be a mix of all three oil temps from the testing I've done previously, but the calcs for the final number have always been a bit of a mystery.
So what do I get from all of this? The system is very good at cooling the gear oil when driving hard. There's only a bit over 2L of fluid and combined with a low set point temp, coolant with a low temp entering the cooler, high flow rate of oil and coolant through the cooler it keeps the temp surprisingly low.
If there are failures of the distance sensor in the 9X1 series of cars, especially street cars, I would be very surprised it was due to excessive temp.
In the 9X7s, the outside casing will allow some cooling, and maybe some heat is conducted into the clutch/valve body chamber and then dissipated via clutch fluid cooling.
The results when I did the hard runs the other day were a bit counter-intuitive. Here's a synopsis of what I did and the results from a 981 BS that has the gear oil cooler.
Something to understand is that regardless of whether sport/sport plus mode is selected, when you jump on the gas the DME changes the thermostat 'set point' temp to 85 deg (it's actually 84.75, but 85 is good enough) from the 105 deg that is normal. It will stay this way for a few minutes after you take it easy and it thinks 85 deg isn't required anymore. The set point is the temp the DME is aiming to achieve at the coolant temp sensor, which is after the coolant has passed through the engine. The temp of the coolant as it leaves the thermostat/pump and enters the engine is going to be cooler. How much cooler? It depends on how much you are wailing on the engine. If you are thrashing the car the coolant entering the engine is far cooler than when you are cruising, as it's just trying to keep the temp leaving the engine 85 deg. The same coolant that is leaving the pump and going to the engine is also going to the engine oil and transmission coolers. So when you are wailing on it the temp going to the coolers is going to be quite low.
Additionally, at high speed the gear oil flow rate through the cooler is high, and at high engine RPM the coolant flow rate through the cooler is high. So if you are driving hard/fast the ability of the system to cool is far better.
The hard runs are on a local freeway that has U turn bays 400 m apart in a particular location. WOT accel to 110 kph, which is the speed limit. U turn, same again over and over until temps fully stabilise. Clearly not on the track but as hard as I've driven the car. Whilst doing the runs I monitored coolant temp, radiator outlet temp, set point temp, engine oil temp, clutch and gear oil temps.
The hottest the gear oil got to when I was beating on it was 86 deg. Being only a bit over 2L of gear oil, it seems the cooler flow is very effective at keeping the temp low when the flow rate of coolant and oil is high and the coolant temp entering the cooler is low.
Ironically the gear oil temp got to 91 deg when I went back to cruising and the set point temp returned to 105. 91 deg on the gear oil was the highest I saw at any time. When I thought about it, this make sense. The set point temp is high, and the coolant temp entering the coolers is high because you aren't driving hard. So the gear oil doesn't get the cooling that it does like when you are wailing.
Highest the clutch fluid got to was 98 and same for the engine oil. When I had reached these maximum temps the instrument cluster was reading 104. No idea why this is. The instrument cluster temp seems to be a mix of all three oil temps from the testing I've done previously, but the calcs for the final number have always been a bit of a mystery.
So what do I get from all of this? The system is very good at cooling the gear oil when driving hard. There's only a bit over 2L of fluid and combined with a low set point temp, coolant with a low temp entering the cooler, high flow rate of oil and coolant through the cooler it keeps the temp surprisingly low.
If there are failures of the distance sensor in the 9X1 series of cars, especially street cars, I would be very surprised it was due to excessive temp.
#119
Nordschleife Master
Wondering how they came to this conclusion. Most failures that start as intermittent, leading to complete inoperation, are one of four things ...
1. Component to output failure to conduct.
2. Connections at either end fail to conduct.
3. Contamination shorting out signals.
4. Mechanical device being sensed is sticking.
But looking at a sealed unit, it would be nearly impossible to know one from the other.
1. Component to output failure to conduct.
2. Connections at either end fail to conduct.
3. Contamination shorting out signals.
4. Mechanical device being sensed is sticking.
But looking at a sealed unit, it would be nearly impossible to know one from the other.
#120
Rennlist Member
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.