Interesting, definitely much tougher. The fiberglass itself (meaning the glass) has low but somewhat decent thermal conductivity, rubber is really poor. However, all this depends on how thick the rubber coating is or whether it is made into a composite within the fiberglass pores. So in the end, it is tougher, but is the thermal conductivity good enough to sink out the IGBT heat? I would "assume" they tested it and all is good. Someone one posted on another thread that there are a few roadster owners with Gruber PEMs. No complaints have popped up yet.
Interesting, definitely much tougher. The fiberglass itself (meaning the glass) has low but somewhat decent thermal conductivity, rubber is really poor. However, all this depends on how thick the rubber coating is or whether it is made into a composite within the fiberglass pores. So in the end, it is tougher, but is the thermal conductivity good enough to sink out the IGBT heat? I would "assume" they tested it and all is good. Someone one posted on another thread that there are a few roadster owners with Gruber PEMs. No complaints have popped up yet.
The thermal conductivity of Gruber's insulating material is 1.0 W/m-K. The material is similar to the insulating/gap-filling material used in various modules in the Model Ss sans the second layer.
Out of interest I went for a drive yesterday...it was probably only 13º or 14º C outside gave the car quite a lot of hard accelerations - the PEM went has high as 39ºC and then in 1 minute or so was back to 37º - it seems to stay around that number most of the time when driving normally...Time will tell how long the Al Oxide insolation will last - but what I do know its very snuggly/tightly fitted 0.8N/M is quite a lot for a stud of just 3mm diameter. The good news about all of this - is dead easy to fix if there are any problems...The next time (if there is a problem) I will go for the aluminium nitate...
DanielFriederich
Member
Temps were the same to me... what I can tell is that you see the heat being better transfered from the IGBTs to the heat sinks (temps raise faster but decline faster too...)
On the other hand the old paste with as to be said 6W/mK works too... GS uses a paste slightly better but more heat resistent than the original one... so lets see how my car behaves when I hopefully soon have it back....
On the other hand the old paste with as to be said 6W/mK works too... GS uses a paste slightly better but more heat resistent than the original one... so lets see how my car behaves when I hopefully soon have it back....
supersnoop
Tesla Roadster #334
Why do you keep talking about Gruber in the third person? That seems a bit disingenuous.
How thick is the heat transfer material?
1 W/m-K is pretty low for thermal conductivity when it also has to be thick enough to act as an electrical insulator.
Note sure if the stuff we use at work for consumer electronics would be suitable for the outdoor-like environment of a car but it has a thermal conductivity of 12 W/m-K
1 W/m-K is pretty low for thermal conductivity when it also has to be thick enough to act as an electrical insulator.
Note sure if the stuff we use at work for consumer electronics would be suitable for the outdoor-like environment of a car but it has a thermal conductivity of 12 W/m-K
Its hard to know with the OEM insulation...they put a past down and screw the IGBT's into that...
The material that some people are using now is 0.63 mm thick
I found the spec sheet for the heat transfer material we use (different application)...
Thermal Conductivity = 12W/mK (+/-1.2)
Breakdown voltage = 12kV/mm (+/-1.2)
Working Temperature -45C to 200C
Deflection: 6% at 10psi, 30% at 30psi, 41% at 50psi
It is available in any thickness from 0.5mm to 2mm. We use 0.8mm and 2mm, pre-cut to the shape we need.
It is a bit like stiff clay/putty - you can dig your fingernail into it and take out a chunk.
For a Roadster it is fantastic for thermal conductivity but there is a risk that it may eventually crumble and need to be replaced like the factory material (haven't seen this, just speculation).
Thermal Conductivity = 12W/mK (+/-1.2)
Breakdown voltage = 12kV/mm (+/-1.2)
Working Temperature -45C to 200C
Deflection: 6% at 10psi, 30% at 30psi, 41% at 50psi
It is available in any thickness from 0.5mm to 2mm. We use 0.8mm and 2mm, pre-cut to the shape we need.
It is a bit like stiff clay/putty - you can dig your fingernail into it and take out a chunk.
For a Roadster it is fantastic for thermal conductivity but there is a risk that it may eventually crumble and need to be replaced like the factory material (haven't seen this, just speculation).
Last edited:
thefortunes
Active Member
You guys are all amazing.
I tore apart the PEM, replaced the crumbling insulating material with Gruber's, found no blown fuses/IGBTs/capacitors/etc..., reassembled (only ending up with (1) extra washer) and just took the Roadster for a short 5 mile ride.
No errors.
THANK YOU TO ALL WHO CONTRIBUTED TO THIS THREAD AND TO @petergrub FOR THE MATERIAL AND HELP!
P.S. This really is not too difficult and (now that I've done it) I would recommend proactive replacement before it becomes a problem. About $100 in material, ~8 hours for (2) people working very methodically, and a strong feeling of accomplishment. Oh, and about $11,000 saved - sorry Tesla.
Edited to add for anyone doing this in the future: I used a 1.5" plastic putty knife to clean off the old insulating material, as well as to spread a thin layer of thermal paste onto the IGBTs. Worked perfectly.
I tore apart the PEM, replaced the crumbling insulating material with Gruber's, found no blown fuses/IGBTs/capacitors/etc..., reassembled (only ending up with (1) extra washer) and just took the Roadster for a short 5 mile ride.
No errors.
THANK YOU TO ALL WHO CONTRIBUTED TO THIS THREAD AND TO @petergrub FOR THE MATERIAL AND HELP!
P.S. This really is not too difficult and (now that I've done it) I would recommend proactive replacement before it becomes a problem. About $100 in material, ~8 hours for (2) people working very methodically, and a strong feeling of accomplishment. Oh, and about $11,000 saved - sorry Tesla.
Edited to add for anyone doing this in the future: I used a 1.5" plastic putty knife to clean off the old insulating material, as well as to spread a thin layer of thermal paste onto the IGBTs. Worked perfectly.
Last edited:
Thanks for your report! The only thing I question is this:
Those seem like two contradictory statements. Of course it depends on what you call "difficult." Would it take 1 person ~16 hrs?
Those seem like two contradictory statements. Of course it depends on what you call "difficult." Would it take 1 person ~16 hrs?
Something could be difficult and take 10 minutes, while something else could be easy and take 10 days. Level of difficulty does not equate to time required.
thefortunes
Active Member
Yep. What he said.
If you can do the annual service, you can do this.
It does take time, but we were working very methodically and carefully. Depending on your level of comfort working with electronics you could do it much more quickly.
Congratulations on your job....for sure its a great feeling of accomplishment...yesterday I had a long chat with Daniel re his PEM...When you do this job make sure the substates are placed flush with the edges of the board. The lines drawn on the heat sink are NOT good in terms of getting your IGBT's CENTRALLY mounted on the substrate. 2nd point 0.63mm may not be thick enough. I'm going to reopen mine and replace my substrates with 1mm thick material and realign the positioning.
DanielFriederich
Member
ok... here now the results from GS in Switzerland (just got informed my PEM is working again and (worst case) the battery fuse was not blown...)
the pictures they sent me of the heat sinks and megapole boards looked rather disencouraging....
- one of the ceramic insulation stripes broken in the corner, disruptive discharge burning a round hole into the ceramic... yeah plasma baby...
- several other igbts (all the boards) disruptive discharges on the corner sites
- many igbts blown or overheated - thermal paste between ceramic and igbts doesnt stay where it should
- two gate resistors on the driver board also destroyed
I talked to Jonathan yesterday (he also uses the Rubalit as insulation) and we are quite certain that my stripes weren't aligned properly enough to guaranty a long lasting insulation of the igbts. Also the use of (too much) thermal paste caused problems concerning heat transfer and isolation... If you consider using ceramics the torque of the clamps hast to be obeyed and the alignment to the igbts has to be precise...
All in all there are pros and cons for that material... I opted now for GS Technology (the are using an insulation material similar to the original gum paste but more heat resistant therefore not degrading over time again) because I knew that at least a few transistors got broken and unsoldering them is a pain in the a... Also GS offers warranty on their work which makes me feel more comfortable and sometimes you have to admit that you're at a dead end of the road being taken...
On the upside I get to travel to Switzerland (300km one way) and driving my Roadster back home maybe using 1st time a HPC on long travel
the pictures they sent me of the heat sinks and megapole boards looked rather disencouraging....
- one of the ceramic insulation stripes broken in the corner, disruptive discharge burning a round hole into the ceramic... yeah plasma baby...
- several other igbts (all the boards) disruptive discharges on the corner sites
- many igbts blown or overheated - thermal paste between ceramic and igbts doesnt stay where it should
- two gate resistors on the driver board also destroyed
I talked to Jonathan yesterday (he also uses the Rubalit as insulation) and we are quite certain that my stripes weren't aligned properly enough to guaranty a long lasting insulation of the igbts. Also the use of (too much) thermal paste caused problems concerning heat transfer and isolation... If you consider using ceramics the torque of the clamps hast to be obeyed and the alignment to the igbts has to be precise...
All in all there are pros and cons for that material... I opted now for GS Technology (the are using an insulation material similar to the original gum paste but more heat resistant therefore not degrading over time again) because I knew that at least a few transistors got broken and unsoldering them is a pain in the a... Also GS offers warranty on their work which makes me feel more comfortable and sometimes you have to admit that you're at a dead end of the road being taken...
On the upside I get to travel to Switzerland (300km one way) and driving my Roadster back home maybe using 1st time a HPC on long travel
Rotarypower
Member
@DanielFriederich........Wow, thanks for the update. Looks like this is a more difficult problem to fix than initially thought, but perhaps the latest insulating material you are going with will prove to be more reliable. Can you please post the images of the failed ceramic material?
thefortunes
Active Member
I will be interested to follow the reliability of material used between Gruber's (which I used) vs the ceramic alternatives. Hopefully we can determine the best option for everyone moving forward.
What are the dimensions of each strip?
I will order some of this 12W/mK material cut to size (thinking 1mm thick).
I will order some of this 12W/mK material cut to size (thinking 1mm thick).
thefortunes
Active Member
@Jonathanm spec'd 167mm x 32mm, I went slightly larger (2mm more) to ensure coverage.
I pulled the top cover to take a look.
Pair of heavy wires into the PEM on the left, and a pair of heavy wrires into the PEM on the right... wondering which is what...
(v1.5)
Pair of heavy wires into the PEM on the left, and a pair of heavy wrires into the PEM on the right... wondering which is what...
(v1.5)
Similar threads
