-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Yes, and that applies to both ends of the cable. The fan end's .250 blade connectors are a lot more robust than the PEM end's tiny pins, but mine proved to be less than perfect too. I guess no connector ever is. Part of my work documented on the other thread suggests the two sets of wires should be paralleled (electrically connected together) upstream from the blower connectors, to increase robustness in the event of a partial connector failure.
Weirdly my PEM motor had almost seized and was probably pulling a lot more power for some time. When I went to pick it up from recharging my AC it threw codes and obviously the PEM was hot. We threw it on the Ramp and blew air threw the PEM pan to see if we could get it moving and put penetrant on the bearing. It blew the fuse and while we were there we pulled all the plugs expecting something burnt, but it was all perfect. I was going to assume the plug would be burnt but it was not. I should also give a shout out to the SC that let me work on it while it was on the ramps and this was a 7:30 pm and made sure it would get me home to swap in a new fan. And didn’t even charge me for the AC recharge.
Good data point! I've been measuring the current draw over the past month or so, and it's been very stable and sane. Peter at Gruber ran a quick amps vs volts test on a blower there, and my results match his, so the motor and bearings appear to be healthy. I opted to not replace the blower assembly, as that would tread into that magical "fixing something that's not broken" territory, where unforeseen consequences lay.
If one suspects that the excessive current draw from the blower is the cause of a connector contact failure, it's easy to wire up an in-line ammeter, as the connectors under the car are easily accessed "F" style (0.250 blade), once the connectors are separated. Note that some of the LCD meters available online don't start working at under about 7 volts, and a normal top-off charge at 240v, 24 amps, 25C ambient, runs at between 6.5 and 7.0 volts, with each side supplying about 3.5 amps (6-7 amps or so total), and lasts about 7 minutes. There are meters that can use an external supply for running the meter's display for situations like this. Unfortunately, I didn't realize this when I bought mine, so I had to splice in a different meter that could work at the lower voltage for my monitoring.
Finally retired from designing power electronics and battery systems for unmanned aircraft, I now have time to give attention to my Roadster. I have been driving and depending on Tesla factory service for 10 years now. The cooling fan issues and surrounding 1144 and 1146 error codes have been persistent in my car from time to time also. I have read most if not all of the related posts on the subject. Many thanks for those that post often for the benefit of all. I would like to contribute my knowledge in return.
This will be a rather long post so for those that don't want to absorb the details, I agree with BartJ on the points he has made. The primary fix, assuming nothing has already overheated or burned, is to take care of the 4 pin connector where the cooling fan motor connects to the PEM. Make sure it is clean, Removing and re-seating may help. Neither will be a permanent solution. Replacement of the cable side, assuming the proper crimp tools is used, every now and again will most likely keep these errors away longer but again, not a "once and done" solution. This connector is the weak point in the system as it has been configured to serve the single motor retrofit. It was however a good choice in the original design when the PEM fan and the traction motor cooling fan were separate. So I would not be too hard on the Tesla engineers. Retrofits are rarely ideal. The original two fan system was a better overall design but as most of us know the fan/motor that was used in that design was not up to the task. I would assume that Tesla could not find a high quality replacement without going to the single motor set up.
For those that want to read deeper here is my analysis. Note that this applies specifically to those cars that have been retrofitted with the single motor / dual fan and also use the Molex 19418 series connector at the PEM.
The 1144 and 1146 errors come up when less current is being supplied by the motor speed control than would be expected. There are two motor speed control circuits. Originally this was for two motors not one. So an error was created for each motor control and related motor. Although monitoring for low current may be counter intuitive, there are good reasons to do this, at least in the original set up. Whether intended or not, these errors still serve a useful purpose. If ignored too long significant damage may occur as noted in these posts. When the single motor was introduced the outputs from the two motor speed controls were connected together to provide enough total current to run the larger motor. This parallel operation of the two motor speed controls is where the problem starts. I don't want to say anything bad about the engineer or team that made this decision. Sometimes the lesser of two evils is accepted, especially if those in command don't fully understand issues and force the engineer in a non-ideal direction. I have been there.
Here is my analysis of the system. Sharing current between two different conductors and associated connectors is rarely done and for good reason. It is very difficult to get the same amount of current in each path. Very small differences in the resistance of the two paths are bound to exist. The current will follow the path of least resistance. When everything is new contact surfaces are clean and tight. The contact resistance will be as low as it will ever be and new connectors are likely to be similar. So when the parts are new this kluge works okay. As time progresses thermal cycles and the mechanical stress of de-mating / mating the connectors will cause the contact resistance to go up. Between the PEM connector and the connectors near the fan there are many contacts involved. They age differently. Careful measurement of my Roadster showed there was about .13 Ohms more resistance in the wiring path of motor controller 2 as compared to motor controller 1. As a result, while running the fan motor at a slow speed, the measured current coming from controller 1 was 3.1 Amps and the current coming from controller 2 was only .42 Amps. This is not the equal sharing that was expected by those who designed the retrofit. My car repeatedly displayed error 1144 due to the low current being delivered by the associated motor control. When the error first occurred it only happened at higher fan speeds needed on hot days or more intense driving. Really anything that caused the fan motor to run fast including charging. As time progressed the error became more prevalent indicating that the contact resistance was increasing. I would speculate that the firmware in the car is programmed to expect a certain minimum current for a given speed. When the actual current does not meet the programmed current expectation the error will be displayed.
The 4 pin connector at the PEM is the worst offender but the connectors near the fan motor also contribute. There are also some wire length differences that make a small difference. In the case of my car just swapping the two connectors feeding the motor made the current sharing more equal. We are talking about very small resistance changes here, less than 10 milliohms. Because of this, fiddling with any of the connectors is likely to make a change. It might get better or worse depending. The only way to tell is to use a clamp on DC current probe. If you want to be informed about the state of your car I would suggest the EXTECH 380942 mini clamp meter. It is the least expensive clamp on current meter that I could find that is good enough for this measurement.
It does not take too much time to access the cooling fan motor and wiring from under the car. The motor is fed with a "Y" cable. This is where the current coming from the two motor speed controls is combined. Put the clamp meter around one of the wires passing through one of the connectors feeding the motor. Compare this to the other connector feeding the motor. A good time to look is when the battery has charged for a few minutes and the fan motor starts to run. In general if current sharing between conductors is included in a design, the lower current conductor should have no less than 80% of the current measured in the other conductor. So measure the two currents and compare. The fan speed does not matter as long as it is constant enough to measure both at roughly the same speed. If the lower current is less than 80% of that in the higher current path there is a problem that should be addressed and it is most likely the cable side of the 4 pin connector at the PEM.
I have some ideas for a "fix it forever" solution. If found I will post the details. If I don't find that I will be replacing the female connector that plugs into the PEM every other time I remove it for cleaning. This relatively easy regular replacement should eliminate the errors in most situations.
This will be a rather long post so for those that don't want to absorb the details, I agree with BartJ on the points he has made. The primary fix, assuming nothing has already overheated or burned, is to take care of the 4 pin connector where the cooling fan motor connects to the PEM. Make sure it is clean, Removing and re-seating may help. Neither will be a permanent solution. Replacement of the cable side, assuming the proper crimp tools is used, every now and again will most likely keep these errors away longer but again, not a "once and done" solution. This connector is the weak point in the system as it has been configured to serve the single motor retrofit. It was however a good choice in the original design when the PEM fan and the traction motor cooling fan were separate. So I would not be too hard on the Tesla engineers. Retrofits are rarely ideal. The original two fan system was a better overall design but as most of us know the fan/motor that was used in that design was not up to the task. I would assume that Tesla could not find a high quality replacement without going to the single motor set up.
For those that want to read deeper here is my analysis. Note that this applies specifically to those cars that have been retrofitted with the single motor / dual fan and also use the Molex 19418 series connector at the PEM.
The 1144 and 1146 errors come up when less current is being supplied by the motor speed control than would be expected. There are two motor speed control circuits. Originally this was for two motors not one. So an error was created for each motor control and related motor. Although monitoring for low current may be counter intuitive, there are good reasons to do this, at least in the original set up. Whether intended or not, these errors still serve a useful purpose. If ignored too long significant damage may occur as noted in these posts. When the single motor was introduced the outputs from the two motor speed controls were connected together to provide enough total current to run the larger motor. This parallel operation of the two motor speed controls is where the problem starts. I don't want to say anything bad about the engineer or team that made this decision. Sometimes the lesser of two evils is accepted, especially if those in command don't fully understand issues and force the engineer in a non-ideal direction. I have been there.
Here is my analysis of the system. Sharing current between two different conductors and associated connectors is rarely done and for good reason. It is very difficult to get the same amount of current in each path. Very small differences in the resistance of the two paths are bound to exist. The current will follow the path of least resistance. When everything is new contact surfaces are clean and tight. The contact resistance will be as low as it will ever be and new connectors are likely to be similar. So when the parts are new this kluge works okay. As time progresses thermal cycles and the mechanical stress of de-mating / mating the connectors will cause the contact resistance to go up. Between the PEM connector and the connectors near the fan there are many contacts involved. They age differently. Careful measurement of my Roadster showed there was about .13 Ohms more resistance in the wiring path of motor controller 2 as compared to motor controller 1. As a result, while running the fan motor at a slow speed, the measured current coming from controller 1 was 3.1 Amps and the current coming from controller 2 was only .42 Amps. This is not the equal sharing that was expected by those who designed the retrofit. My car repeatedly displayed error 1144 due to the low current being delivered by the associated motor control. When the error first occurred it only happened at higher fan speeds needed on hot days or more intense driving. Really anything that caused the fan motor to run fast including charging. As time progressed the error became more prevalent indicating that the contact resistance was increasing. I would speculate that the firmware in the car is programmed to expect a certain minimum current for a given speed. When the actual current does not meet the programmed current expectation the error will be displayed.
The 4 pin connector at the PEM is the worst offender but the connectors near the fan motor also contribute. There are also some wire length differences that make a small difference. In the case of my car just swapping the two connectors feeding the motor made the current sharing more equal. We are talking about very small resistance changes here, less than 10 milliohms. Because of this, fiddling with any of the connectors is likely to make a change. It might get better or worse depending. The only way to tell is to use a clamp on DC current probe. If you want to be informed about the state of your car I would suggest the EXTECH 380942 mini clamp meter. It is the least expensive clamp on current meter that I could find that is good enough for this measurement.
It does not take too much time to access the cooling fan motor and wiring from under the car. The motor is fed with a "Y" cable. This is where the current coming from the two motor speed controls is combined. Put the clamp meter around one of the wires passing through one of the connectors feeding the motor. Compare this to the other connector feeding the motor. A good time to look is when the battery has charged for a few minutes and the fan motor starts to run. In general if current sharing between conductors is included in a design, the lower current conductor should have no less than 80% of the current measured in the other conductor. So measure the two currents and compare. The fan speed does not matter as long as it is constant enough to measure both at roughly the same speed. If the lower current is less than 80% of that in the higher current path there is a problem that should be addressed and it is most likely the cable side of the 4 pin connector at the PEM.
I have some ideas for a "fix it forever" solution. If found I will post the details. If I don't find that I will be replacing the female connector that plugs into the PEM every other time I remove it for cleaning. This relatively easy regular replacement should eliminate the errors in most situations.
This is one of my all time favourite first forum posts. Welcome, and interested to hear how your fix works out.
Please do share! The only "forever" fixes I've seen are ones where both sides of the connection at the PEM are swapped for something capable of handling both the current and frequent (yearly) removal cycles.
But I think there are things we can do to minimize the chances for failure. #1 is that when you do get either error, don't ignore it. They shouldn't happen if everything is working properly, and if left unresolved, there will likely be permanent damage to the male (PEM-side) connector, which is very hard to replace. #2 is that the connectors don't take kindly to being angled, so pull straight up / down on them, instead of wiggling them from side to side. The "spring" in the female side is rather inelastic (aka, not a spring). So, finger pointed at Molex on this one too, not just Tesla for using it. +1 for considering it a single-use connector.
Please take a look at my post on this topic in Think I need a new PEM Fan... Anything else to check? where you can follow my analysis of what turned out to be a pair of problems (spoiler alert: it wasn't the fan!), and what I believe was the contributing cause. Fast forward to the last couple of posts if you aren't a fan of engineering detective mysteries.
Any comments on what I did, or could do better, will help us all.
Oh, and welcome to the forum!
Thanks to all for the welcoming responses.
Greg, I took a look at the other discussion "Think I need a new PEM fan...". Lots of interesting information. Kudos for sticking with the diagnostic process.
While reading, I had a few additional thoughts. There was some talk about gold pins. They would be an improvement but only if both sides of the connector are gold. There is some disagreement in the connector world regarding mixing. It ranges from mixing contact materials is bad and should not be done to it is okay to mix but performance will still be reflective of the lesser material. I am a big proponent in gold connectors for anything beyond disposable consumer goods where mating is usually done only once. Have not found the gold version of the PEM side in stock anywhere yet. I will only change to gold if I can find both sides.
The information that it is usually, if not always, the positive side pins that go bad is interesting and has good reason. It is a bit late so I hope this explanation is not too fuzzy. I can give it another go if there are questions. As has been discovered, the positive pins are not switched. They connect directly to the +12 power bus. The minus side pins connect to an FET switch that when closed connects to ground. There is one FET for each motor speed control. The FET switches are pulsed on (PWM) from 0 to 100% of the time to control motor speed.
Some background. One way to make current share more equally on a pair of conductors is to add resistance in series with each of the conductors. If the added resistance is significantly higher than the associated connectors and wire in the current path, the current sharing will primarily be dictated by the larger resistance. Earlier I noted that the current will take "the path of least resistance". Let's say for example there are two conductors and related connectors providing current to the same end point. One of these conductive paths has a resistance of .1 Ohm and the other has a resistance of .15 Ohms. From an Ohms law perspective there is a big difference. Most of the current will take the conductor that has the lower resistance. Now if we insert 1 Ohm in each conductor the math will be different. One current path will see 1.1 Ohms and the other will see 1.15 Ohms. The relative difference in resistance between the two current paths now becomes insignificant and current sharing will be close to equal. If someone wants the math, I can show it later.
Getting back to why the positive pins burn first. The positive side power is connected to the +12 power bus directly. Very little resistance between the supply and the pins on the PEM connector. So the current sharing is almost entirely dictated by the resistance of the harness wire and connectors. The elements that contribute to the total resistance in this path are very low and small differences cause large imbalances in current sharing as has been discussed. At some point the imbalance gets so bad that the connector contacts on the conductor carrying the bulk of the current become overloaded and burn. Once they burn the other path that was previously the lesser path becomes the only path and it burns also.
The negative side is somewhat different. There is not a direct connection to ground. The FET switch that connects it to ground when switched on does have internal resistance. So in effect each of the conductive paths in the negative side has series resistance that is significantly larger than that of the wires and connectors. Current sharing is then dictated by the larger internal resistance of the FET rather than the wire and connectors as in the example above.
Although not ideal (I am not recommending it at this time) one possible fix would be to deliberately add additional resistance in each of the +12 supply lines. This does work but I consider it a kluge and fixing the "Y" power feed kluge that Tesla provided with another kluge goes against my engineering values.
Still thinking about an acceptable solution while waiting for my new connectors to arrive. Has anyone found the gold version of the male pin part in the PEM in stock anywhere?
Greg, I took a look at the other discussion "Think I need a new PEM fan...". Lots of interesting information. Kudos for sticking with the diagnostic process.
While reading, I had a few additional thoughts. There was some talk about gold pins. They would be an improvement but only if both sides of the connector are gold. There is some disagreement in the connector world regarding mixing. It ranges from mixing contact materials is bad and should not be done to it is okay to mix but performance will still be reflective of the lesser material. I am a big proponent in gold connectors for anything beyond disposable consumer goods where mating is usually done only once. Have not found the gold version of the PEM side in stock anywhere yet. I will only change to gold if I can find both sides.
The information that it is usually, if not always, the positive side pins that go bad is interesting and has good reason. It is a bit late so I hope this explanation is not too fuzzy. I can give it another go if there are questions. As has been discovered, the positive pins are not switched. They connect directly to the +12 power bus. The minus side pins connect to an FET switch that when closed connects to ground. There is one FET for each motor speed control. The FET switches are pulsed on (PWM) from 0 to 100% of the time to control motor speed.
Some background. One way to make current share more equally on a pair of conductors is to add resistance in series with each of the conductors. If the added resistance is significantly higher than the associated connectors and wire in the current path, the current sharing will primarily be dictated by the larger resistance. Earlier I noted that the current will take "the path of least resistance". Let's say for example there are two conductors and related connectors providing current to the same end point. One of these conductive paths has a resistance of .1 Ohm and the other has a resistance of .15 Ohms. From an Ohms law perspective there is a big difference. Most of the current will take the conductor that has the lower resistance. Now if we insert 1 Ohm in each conductor the math will be different. One current path will see 1.1 Ohms and the other will see 1.15 Ohms. The relative difference in resistance between the two current paths now becomes insignificant and current sharing will be close to equal. If someone wants the math, I can show it later.
Getting back to why the positive pins burn first. The positive side power is connected to the +12 power bus directly. Very little resistance between the supply and the pins on the PEM connector. So the current sharing is almost entirely dictated by the resistance of the harness wire and connectors. The elements that contribute to the total resistance in this path are very low and small differences cause large imbalances in current sharing as has been discussed. At some point the imbalance gets so bad that the connector contacts on the conductor carrying the bulk of the current become overloaded and burn. Once they burn the other path that was previously the lesser path becomes the only path and it burns also.
The negative side is somewhat different. There is not a direct connection to ground. The FET switch that connects it to ground when switched on does have internal resistance. So in effect each of the conductive paths in the negative side has series resistance that is significantly larger than that of the wires and connectors. Current sharing is then dictated by the larger internal resistance of the FET rather than the wire and connectors as in the example above.
Although not ideal (I am not recommending it at this time) one possible fix would be to deliberately add additional resistance in each of the +12 supply lines. This does work but I consider it a kluge and fixing the "Y" power feed kluge that Tesla provided with another kluge goes against my engineering values.
Still thinking about an acceptable solution while waiting for my new connectors to arrive. Has anyone found the gold version of the male pin part in the PEM in stock anywhere?
I have been continuing to resolve the conditions surrounding the 1144 and 1146 errors all through the summer season and have gathered some information of interest. As of now I am no longer getting errors in my car but my last revisions were recent and the true test will be next summer when temperatures get above 90F.
First I have to admit to making a bad assumption based on bad information from the Tesla service department. When the conversion was made from dual fan motors to a single motor set up I questioned them about the "Y' cable that parallels the two fan controllers. I was told that there was a firmware update that was part of the modification. It was said that this firmware change forced the two fan controllers to operate at identical power levels so that the current required by the larger fan motor would be shared equally between the two controllers. My concerns about proper current sharing started the day I learned of the "Y" cable modification. Seems the concerns were not unfounded.
Based on the above information I started this project with the assumption that the two fan controllers were doing the same thing. In my car the controllers run quite independently, It would seem that the firmware change never existed as decrobed or my car does not have the correct firmware. Any information regarding the existence, or not, of a firmware change related the the fan conversion would be appreciated.
Without getting too deep in the details here is how the project progressed. Current shunts were added in line with each of the 4 conductors at the 4 pin PEM connector that feeds power to the fans. The shunts were selected with a value, .02 Ohms, that would also enhance current sharing through the connector. See earlier post. I know I said I would not fix a kluge with another kluge but in this case the added resistance was needed anyway so that I could monitor current for each of the four connector pins so I allowed it. All other options required intrusive modifications to the car. I wanted to make the changes easily reversible. Preliminary results were great. Current was shared equally between the two +12 pins and similarly between the power return pins that are switched to ground inside the PEM. Thought that success was near.
When I started driving the car the 1144 error started to show up. No Joy! The errors were less than prior to my changes so for the most part I just ignored them and enjoyed the summer. While driving I started to notice a pattern in the errors. To further investigate I installed meters in the cockpit so the currents could be seen while driving. This is when I discovered that the two fan controls operate independently. In some situations the two controllers are in concert, as they were in my initial testing, but other times the one originally dedicated to the motor draws more current and other times the controller originally for the PEM draws more current. The sequence of these changes looks logical if the PEM and Motor cooling fans were still on separate motors.
It was also a bit of a surprise to see that the total current to the fan motor never goes above about 13 Amps, much less than the fan is rated for. Not a problem, just a data point. It actually makes me feel better about the current rating on the connector pins at the PEM.
So the revised circuit topology resulted in very good current sharing when the two controllers were doing the same thing but the 1144 errors continued. The 1144 error is indicating low current on the controller that was originally dedicated to the PEM. This is pin 4 on the connector. I had previously given some consideration to the fact that the wire length in the harness is different. The harness was cut with shorter wires going to the PEM fan. If the connector pins all have close to the same resistance, as when new, the current would not share equally because of the unequal wire lengths. It is impossible to know for sure but my guess is that the actual firmware change must have been an adjustment of error thresholds to account for the unequal sharing caused by the wire length difference. I may be giving them too much credit here but my guess seems to be somewhat confirmed by my next step.
Extra resistance was added in the wire connecting to pin 3, originally the controller for the motor fan. This forced more current to the PEM controller. Values were tested that distributed the current between the two controllers in roughly the same proportion that the wire length difference would have dictated. The first value tried resulted in 60% of motor current going to the PEM controller and 40% going to the motor controller. This reduced the number of 1144 errors by a factor of 10, most cleared in seconds. Almost error free but still not happy. Next, the current share was shifted a bit further to direct 67% to the PEM controller and 33% to the motor controller. Having only driven a few hundred miles and in cooler weather, it is hard to say this is a permanent fix but I am optimistic. No errors yet. In any case the current sharing is now determined by resistors of constant value rather than the contact resistance of connector pins, which change over time and use. As a result the current sharing proportions should remain much more constant than the original Tesla configuration.
Looks like this took a lot of words without really getting into the details. Sorry for the long posts but don't know how to cover it any other way.
First I have to admit to making a bad assumption based on bad information from the Tesla service department. When the conversion was made from dual fan motors to a single motor set up I questioned them about the "Y' cable that parallels the two fan controllers. I was told that there was a firmware update that was part of the modification. It was said that this firmware change forced the two fan controllers to operate at identical power levels so that the current required by the larger fan motor would be shared equally between the two controllers. My concerns about proper current sharing started the day I learned of the "Y" cable modification. Seems the concerns were not unfounded.
Based on the above information I started this project with the assumption that the two fan controllers were doing the same thing. In my car the controllers run quite independently, It would seem that the firmware change never existed as decrobed or my car does not have the correct firmware. Any information regarding the existence, or not, of a firmware change related the the fan conversion would be appreciated.
Without getting too deep in the details here is how the project progressed. Current shunts were added in line with each of the 4 conductors at the 4 pin PEM connector that feeds power to the fans. The shunts were selected with a value, .02 Ohms, that would also enhance current sharing through the connector. See earlier post. I know I said I would not fix a kluge with another kluge but in this case the added resistance was needed anyway so that I could monitor current for each of the four connector pins so I allowed it. All other options required intrusive modifications to the car. I wanted to make the changes easily reversible. Preliminary results were great. Current was shared equally between the two +12 pins and similarly between the power return pins that are switched to ground inside the PEM. Thought that success was near.
When I started driving the car the 1144 error started to show up. No Joy! The errors were less than prior to my changes so for the most part I just ignored them and enjoyed the summer. While driving I started to notice a pattern in the errors. To further investigate I installed meters in the cockpit so the currents could be seen while driving. This is when I discovered that the two fan controls operate independently. In some situations the two controllers are in concert, as they were in my initial testing, but other times the one originally dedicated to the motor draws more current and other times the controller originally for the PEM draws more current. The sequence of these changes looks logical if the PEM and Motor cooling fans were still on separate motors.
It was also a bit of a surprise to see that the total current to the fan motor never goes above about 13 Amps, much less than the fan is rated for. Not a problem, just a data point. It actually makes me feel better about the current rating on the connector pins at the PEM.
So the revised circuit topology resulted in very good current sharing when the two controllers were doing the same thing but the 1144 errors continued. The 1144 error is indicating low current on the controller that was originally dedicated to the PEM. This is pin 4 on the connector. I had previously given some consideration to the fact that the wire length in the harness is different. The harness was cut with shorter wires going to the PEM fan. If the connector pins all have close to the same resistance, as when new, the current would not share equally because of the unequal wire lengths. It is impossible to know for sure but my guess is that the actual firmware change must have been an adjustment of error thresholds to account for the unequal sharing caused by the wire length difference. I may be giving them too much credit here but my guess seems to be somewhat confirmed by my next step.
Extra resistance was added in the wire connecting to pin 3, originally the controller for the motor fan. This forced more current to the PEM controller. Values were tested that distributed the current between the two controllers in roughly the same proportion that the wire length difference would have dictated. The first value tried resulted in 60% of motor current going to the PEM controller and 40% going to the motor controller. This reduced the number of 1144 errors by a factor of 10, most cleared in seconds. Almost error free but still not happy. Next, the current share was shifted a bit further to direct 67% to the PEM controller and 33% to the motor controller. Having only driven a few hundred miles and in cooler weather, it is hard to say this is a permanent fix but I am optimistic. No errors yet. In any case the current sharing is now determined by resistors of constant value rather than the contact resistance of connector pins, which change over time and use. As a result the current sharing proportions should remain much more constant than the original Tesla configuration.
Looks like this took a lot of words without really getting into the details. Sorry for the long posts but don't know how to cover it any other way.
Interesting analysis, thanks for sharing. In my car at least, the firmware upgrade did get applied, and the two drivers do appear to work in concert. Unfortunately, changing the flag that enables the "move together" mode is not one we can access. It's Tesla Service Center territory.
In addition, I think you're giving the contacts on the PEM-end of the wiring harness far too much credit. They are very much undersized, and worse, their construction almost guarantees that they are only good for a couple of insertions, if even that. Wiggling the connector from side to side in order to pull it from the PEM can permanently deform the nearly spring-less contacts. Worse, the connectors on the other end of the cable - the "Y" part - while far better, can be the source of problems back at the PEM. I'm on my 3rd PEM-end connector because of them. Please see Think I need a new PEM Fan... Anything else to check? for my analysis and (hopefully final) repair. It's a long thread which took a number of false turns, so you might just skip the otherwise entertainment-value stuff by flipping to the last page.
I got quite familiar with your post Think I need a new PEM fan while researching my problem. Thanks for that information. It saved me a lot of time running down connector part numbers.
Specifying connectors has been a large part of my work in power electronics design. I am not a fan of the Molex designs or the quality control within the company. I don't believe the 18 Amp rating they give the pins in this connector is realistic. Worst case a pin will see about 9 Amps, I will trust them that far. I will also be able to easily check the current flowing through each pin from time to time. If the contact resistance in any of the pins changes significantly it will be apparent.
I am curious how you were able to determine that the two drivers in your car are working in concert. Can you explain?
Specifying connectors has been a large part of my work in power electronics design. I am not a fan of the Molex designs or the quality control within the company. I don't believe the 18 Amp rating they give the pins in this connector is realistic. Worst case a pin will see about 9 Amps, I will trust them that far. I will also be able to easily check the current flowing through each pin from time to time. If the contact resistance in any of the pins changes significantly it will be apparent.
I am curious how you were able to determine that the two drivers in your car are working in concert. Can you explain?
As you, partly by just observing the behavior of the two circuits with a meter. The definitive analysis came from Peter at Gruber, who has a log viewer that showed both sides driving together. Post #23 and #24 in the thread; also #99. Peter might be a wee bit distracted with other things right now, but he's a great resource.
Thanks for your post references. #24 seems to be making the assumption that the controller is on the +12 side of the motor. It is not. The only devices between the +12 power source and the pins that supply +12 to the motor are fuses. One for each motor in the original two motor design. With the "Y" cable in place both 20 Amp fuses are feeding the (+) side of the motor. The PWM switching to control fan speed is done on the (-) side of the motor. The Hall sensors are also on the low side of the motor.
The really bad thing about this arrangement is that if a fuse blows or a connector contact on the +12 side things goes bad, all of the fan current comes from the other +12 feed. Fused at 20 Amps there is plenty of current available to do damage. This was demonstrated in my car some time back when a Tesla service tech forgot to connect one side of the "Y' cable to the PEM harness. All was good until things warmed up and full fan speed was commanded. At that point the sole controller that was connected to the fan failed and a new PEM had to be installed.
I have been monitoring the current on pins 3 and 4 of the 4 pin Molex connector. These connect to the (-) side of the motor. It is here that the current load on the individual controllers can be seen. Measuring voltage is not helpful since the "Y" cable connects these two pins, forcing them to be very close to the same voltage.
The really bad thing about this arrangement is that if a fuse blows or a connector contact on the +12 side things goes bad, all of the fan current comes from the other +12 feed. Fused at 20 Amps there is plenty of current available to do damage. This was demonstrated in my car some time back when a Tesla service tech forgot to connect one side of the "Y' cable to the PEM harness. All was good until things warmed up and full fan speed was commanded. At that point the sole controller that was connected to the fan failed and a new PEM had to be installed.
I have been monitoring the current on pins 3 and 4 of the 4 pin Molex connector. These connect to the (-) side of the motor. It is here that the current load on the individual controllers can be seen. Measuring voltage is not helpful since the "Y" cable connects these two pins, forcing them to be very close to the same voltage.
Right. But recall that the thread from beginning to end was a journey from knowing nothing about that part of the car to working out how it works, understanding the failures (plural) at hand, and deciding on a fix. At #24, I was still under the assumption that, as most power supplies, the fan was regulated on the positive side, with the negative side grounded. And a lot of stuff not making sense. Reality was revealed a page later, and everything came together shortly after that.
You bring up a good point about the fuses being another source of failure that could unbalance the draw on the fragile PEM connector pins. Is there a separate alert that would notify the driver that it had gone, so that it could be addressed before the connector fails? I'm guessing no, that they are depending on the lack of current flowing on the negative pins (and the existing 1144 / 46 alerts) which will not be effective. If one fuse goes, the other side's positive pin will be carrying all the current (and frying as a result), with nothing to alert the driver.
What is the right fix for this, assuming that we can't change the connector under the PEM? Smaller fuses (mine are 20's)? Cross-connect the positive side inside the PEM and go to a single fuse? Crossed fingers and leave bad enough alone?
Last edited:
I understand your journey; I found it quite a challenge to fully understand the topology and pitfalls of the fan system in the revised single motor version. The original system using two controllers, two fuses, two motors is straight forward. That circuit topology is according to standard design norms and would generally pass muster in the industrial / avionics world. The only issue I see in the original layout is the selection of 20 Amp fuses, assuming this is the original value. Having a connector pin with an 18 Amp rating downstream of the fuse is not acceptable.
For my needs the solution should be one that does not require any modification of the PEM. I don't want to do anything to my car that would prevent swapping PEMs or any other module in the car. I also want to keep any modifications to a form that is easily reversible. Unfortunately meeting these requirements makes it much more difficult to prescribe the "right fix".
At present I believe I have a good fix that meets the criteria. I do want to resolve all of my questions before releasing a drawing and explanation of my modifications. The question that remains is with regard to the firmware in my car. I would not want to release something that works with the wrong firmware. At this point I thing the likelihood of wrong firmware or a wrong setting is low. The car ran without errors for over 8 years after the installation of the single motor fan set up. The question still must be resolved.
Once the above question is answered and I have more test time on my modification I will publish a physical build description and schematic. At this time I can offer a discussion of the most straightforward part of the modification. It applies to the fuses and pins 1 and 2 on the problematic Molex connector on the PEM. These are the two pins that connect to the +12 VDC supply by way of the fuses. In my earlier post I described how series resistance can be used to promote equal current sharing when there is a necessity to have parallel current paths. Also noted in that post that the two pins serving the (-) side of the motor already have some resistance in series with pins 3 and 4 on the PEM connector. This resistance is contributed by the FET that is used as a switch the motor to ground according to a PWM command. Although I would like to see a bit more resistance in series with pins 3 and 4, the contribution from the FETs is better than none at all as is the case on the +12 side.
Here is the +12 side of the modifications. I should offer a sketch but don't have time at present. Note that this is for discussion at this point, there are additional details necessary to implement this in terms of safe mechanical mounting of the resistors, power ratings etc. Proceed at your own risk if you want to give it a go with this limited information.
Starting with pins 1 and 2 on the cable mounted connector that plugs into the PEM. Remove the wires inserted into pins 1 and 2. Cut the pins off as close as possible to the pins, saving as much wire as possible. Connect these two wires together. Now make up two short lengths of wire with new female Molex pins one one end. Install these pins in positions 1 and 2 in the Molex housing. Attach one end of a .02 Ohm resistor to the free end of each of the short wires that have been installed in the Molex connector. At this point the modified connector should have two short wires with resistors attached. Each resistor should have an unconnected lead. Now be join the unconnected resistor leads together along with the two wires feeding the (+) side of the Motor. Replace fuses with 15 Amp parts.
Sorry no drawing. To further explain here is the current path for one side. Current flows from from the car's +12 VDC power supply through 15 a Amp fuse then through the associated Molex connector pin then through the added .02 Ohm resistor. At this point the current is summed with the other side to create a single point source providing current to the motor. This modification assures that current sharing between pins 1 and 2 will remain close to equal even as the contact resistance of the pins changes over time and use.
Since I have seen burned pins on the + side in many posts, implementing this part of the modification is a good step forward. Assuming all of the connector pins, including those on the "Y" cable are in good condition, this alone could solve many issues. Worst case wasted power in the resistors is only a few Watts and is proportional to fan speed. I don't like to waste power but it seems like a small price to pay in this case.
For my needs the solution should be one that does not require any modification of the PEM. I don't want to do anything to my car that would prevent swapping PEMs or any other module in the car. I also want to keep any modifications to a form that is easily reversible. Unfortunately meeting these requirements makes it much more difficult to prescribe the "right fix".
At present I believe I have a good fix that meets the criteria. I do want to resolve all of my questions before releasing a drawing and explanation of my modifications. The question that remains is with regard to the firmware in my car. I would not want to release something that works with the wrong firmware. At this point I thing the likelihood of wrong firmware or a wrong setting is low. The car ran without errors for over 8 years after the installation of the single motor fan set up. The question still must be resolved.
Once the above question is answered and I have more test time on my modification I will publish a physical build description and schematic. At this time I can offer a discussion of the most straightforward part of the modification. It applies to the fuses and pins 1 and 2 on the problematic Molex connector on the PEM. These are the two pins that connect to the +12 VDC supply by way of the fuses. In my earlier post I described how series resistance can be used to promote equal current sharing when there is a necessity to have parallel current paths. Also noted in that post that the two pins serving the (-) side of the motor already have some resistance in series with pins 3 and 4 on the PEM connector. This resistance is contributed by the FET that is used as a switch the motor to ground according to a PWM command. Although I would like to see a bit more resistance in series with pins 3 and 4, the contribution from the FETs is better than none at all as is the case on the +12 side.
Here is the +12 side of the modifications. I should offer a sketch but don't have time at present. Note that this is for discussion at this point, there are additional details necessary to implement this in terms of safe mechanical mounting of the resistors, power ratings etc. Proceed at your own risk if you want to give it a go with this limited information.
Starting with pins 1 and 2 on the cable mounted connector that plugs into the PEM. Remove the wires inserted into pins 1 and 2. Cut the pins off as close as possible to the pins, saving as much wire as possible. Connect these two wires together. Now make up two short lengths of wire with new female Molex pins one one end. Install these pins in positions 1 and 2 in the Molex housing. Attach one end of a .02 Ohm resistor to the free end of each of the short wires that have been installed in the Molex connector. At this point the modified connector should have two short wires with resistors attached. Each resistor should have an unconnected lead. Now be join the unconnected resistor leads together along with the two wires feeding the (+) side of the Motor. Replace fuses with 15 Amp parts.
Sorry no drawing. To further explain here is the current path for one side. Current flows from from the car's +12 VDC power supply through 15 a Amp fuse then through the associated Molex connector pin then through the added .02 Ohm resistor. At this point the current is summed with the other side to create a single point source providing current to the motor. This modification assures that current sharing between pins 1 and 2 will remain close to equal even as the contact resistance of the pins changes over time and use.
Since I have seen burned pins on the + side in many posts, implementing this part of the modification is a good step forward. Assuming all of the connector pins, including those on the "Y" cable are in good condition, this alone could solve many issues. Worst case wasted power in the resistors is only a few Watts and is proportional to fan speed. I don't like to waste power but it seems like a small price to pay in this case.
Thanks for the description; perfectly clear. Basically, it's my solution of the cross-connect of the positive wires, with the addition of the 0.02 ohm resistors in series for current balance, and a smaller fuse value (though that makes me a bit nervous).
Question: Is there a reason to not cross-connect the two negative lines as well, as I have done? My logic: They are cross connected at the fan, on the other side of the two F-type connectors, but adding a cross-connect upstream from there should protect against one of the blades failing. Without the negative side cross-connect, a negative side F-connector failure would drop the fan current (and presumably the fan speed) in half. The lower fan speed would force the remaining side to increase its power output, potentially exceeding the PEM connector pin rating and thus precipitating a second failure or a blown fuse (leaving you stranded). An argument for not cross-connecting the negative side is that an F connector failure on the negative side would at least result in a 1144/1146 alert; the cross-connect prevents that warning. However, given that the one of the 1144 / 1146 alerts (I forget which) is never displayed in the cabin, and neither are passed on to the OVMS, it could be quite a while before a hidden alert is discovered by the driver. I'm not sure which scenario is worse... With the F-connector failure being less likely than the Molex, perhaps it's not worth worrying about either way?
When I cross connected the negative side to assure equal current in both controllers I got a lot of 1144 errors. If you have done the cross connect and are not getting any errors it would seem that you are good to go. Cross connecting both the positive and negative sides at the PEM connector is a lot better layout in terms of equal current sharing than the Tesla plan of doing the cross connect in the "Y" cable.
Ok, interesting. Never saw any errors (1144 or 1146) after redoing the cable, which does both cross connects. I'll keep it as-is, then.
I'm guessing your errors have something to do with the mis-set firmware flag involved with the dual fan upgrade. It's probably seeing current from one side hitting the other (assumed independent) side. Taking the car to a Service Center to get the flag updated would probably be a good idea in any event.
Similar threads
