FWIW I remember something that I noticed during the PEM rebuild. The wires that go from the heatsinks to the 3 individual phase boards have a connector that you can separate in 2 parts. The upshot of this is you don't have to file the holes out in the boards any larger to get the connector through. I must admit I thought it a bit strange at the time. One less task if anyone attempts this job...
thefortunes
Active Member
Yep. We disassembled the PEM yesterday and didn't have the problem you described.
We also didn't find anything obviously blown - no smell, no debris, nothing swollen. We will order the insulation, put it together and cross our fingers.
Thanks again for the pictures and write-up.
DanielFriederich
Member
So... after driving whole summer, some thousand km with the overworked PEM with no issues what so ever two days ago I took little angry baby for a spin in heavy rain... cold, around 1-2 deg.C and the battery at about 14 C I got not much power (not more than 100 kw)... after driving 2-3 km and while accelerating a popping noise from the back like I went over something... seconds later two or three other "pops" and immediate powerloss on drive train... So got the car to a safe pullover and - been there, done it, got the t-shirt - was pulled back home by an old ICE again.... Turns out that on megapole C (the very outer one) at least two of the igbts had blown up, some debris in inside PEM (but dry)... no visible view of a broken insulation e.g.
I called the guys in Switzerland (GS technology) to take over from now on since I don't have the will and time taking a deeper look by myself....
Car will go to Switzerland coming Thursday and PEM overworked beginning next year... Looking forward to make my first trip home the having to use a HPC while traveling....
And the Visium Hardtop will probably arrive when the Roadster is not here... Kudos... I wonder how I can still be that calm....
I called the guys in Switzerland (GS technology) to take over from now on since I don't have the will and time taking a deeper look by myself....
Car will go to Switzerland coming Thursday and PEM overworked beginning next year... Looking forward to make my first trip home the having to use a HPC while traveling....
And the Visium Hardtop will probably arrive when the Roadster is not here... Kudos... I wonder how I can still be that calm....
BTW while you are at it check all the fuses. I found the little megapole fuse was blown - thats on the little board - if you have the PEM in front of you with the 3 phase boards on the left of the PEM - its on the back of the 1st phase board...
As long as you had the normal error messages - about desat fault etc - now I'd pretty well bet my house on your car being good to go after the install...
Man sorry to hear about that...does not sound related to the install of new insulation...would have been interested to see the temperature of the PEM when this happened...BTW you might want to pull the plug on the ESS...I wonder if IGBT's got slightly damaged/weakened last time due to short on heat sinks?
BTW Daniel what error messages did you get?
thefortunes
Active Member
Pics of the house will be required prior to an actual bet
Haha...Pics of the house will be required prior to an actual bet
Just spoke to Daniel - he got the same error message (Desat etc) so he thinks he may have broken his insulation. He thinks maybe the guy who did his repair in the spring did not torque correctly the IGBT's to the heat sink...we'll find out when he gets his car back..
If there are any geeks who want to know about the torquing of the IGBT's - there's a small report that I found prior to doing my repair...
Last edited:
DanielFriederich
Member
for those with an eye for finer details:
- no unusual temp in PEM
- no "massive" bumps from outside (just a little sliding on rainy road)
- no too hard acceleration (little below 100kw)
- no humidity or water inside PEM
only error: Phase C high side desat fault (from visual check only one megapole board dead cause of blown IGBTs)
I will keep you guys informed asa I get new info
- no unusual temp in PEM
- no "massive" bumps from outside (just a little sliding on rainy road
)- no too hard acceleration (little below 100kw)
- no humidity or water inside PEM
only error: Phase C high side desat fault (from visual check only one megapole board dead cause of blown IGBTs)
I will keep you guys informed asa I get new info
Daniel- so sorry to hear, but being a pioneer in this you are hashing out the issues to reach perfection.
It will be worth it in the end.
You truly hammered the car out well in the summer and all seemed to be fine. Besides possible coincidental failure of the
ceramic thermal conductors (or IGBT) as you noted, it seems this is the first time the car has gone through some quite cold storage to hot operation cycles (?)
A slight possibility is that the aluminum thermal sink substrate contracted with the cold slightly putting the ceramic in some tension (leading to failure) it has not seen before since installation or maybe creating a thermal gap. Coefficient of thermal expansion is 3X higher for aluminum than aluminum oxide. If it is a lateral shrinkage the ceramic should actually go into compression not tension, but maybe it interacts with bolts through the IGBT substrate in the perpendicular direction creating more stress on the Al2O3 somehow.
One should be able to model the stress loads via FEA (if one has a lot of free time...).
Anyway, you will know soon and find a pathway out. Thanks for giving us an update.
DanielFriederich
Member
my guess right now is that due to different torque on the clamps I have a broken insulation... but lets see what GS Technology finds out...
If for some reason you have to redo the sinks, you may also want to take a look at aluminum nitride instead of alumina (aluminum oxide). The thermal conductivity is much (3X or so) higher than alumina, I believe the thermal expansion match with aluminum is also a bit better. It has similar or slightly better strength but slightly lower fracture toughness. In the end, if the spacing between the IGBT and the aluminum sink allows, you could use a thicker AlN ceramic (2-3X thicker) which will be much stronger and still have better thermal conductivity than the thin aluminum oxide.
However, I'm not sure you can get aluminum nitride laser machined to the widths you want..
However, I'm not sure you can get aluminum nitride laser machined to the widths you want..
Hello
Maybe a dumb question, but would it nor be possible to use a thermal paste instead of those sinks that breaks? Is there so much difference in performance? It looks like the original stuff that did wear out in time was a paste no? So maybe it's easier to use new paste and replace it every 3-4 years instead of having a part that can break while you drive on a pothole? Just asking...
Kind regards
Maybe a dumb question, but would it nor be possible to use a thermal paste instead of those sinks that breaks? Is there so much difference in performance? It looks like the original stuff that did wear out in time was a paste no? So maybe it's easier to use new paste and replace it every 3-4 years instead of having a part that can break while you drive on a pothole? Just asking...
Kind regards
HI Rolf - good question....I feel pretty sure that once this is properly torqued - its really held tight in there and it cannot move. I'm sure that I am not the 1st car to use this material and - if this is Daniels problem? - then we would have heard more about this happening to other cars. I will be driving around a lot this holiday period and it will get a good test. None the less, I'm very interested in seeing if aluminium nitride is a viable option. The choice of these materials is a bit above my skill set but at the end of the day - if I have problems then I'm happy to try an alternative substrate - its a pretty easy fix...this is all part of our learning curve.
@ion_1 - you theoretically have space for a thicker substrate - the only limit is the length of the studs currently used. As is now, you could get another millimetre no prob maybe more. The substrates used now are 0.63mm. If you want to use something even thicker you'll have to get longer studs...not sure its a big problem...
@ion_1 - you theoretically have space for a thicker substrate - the only limit is the length of the studs currently used. As is now, you could get another millimetre no prob maybe more. The substrates used now are 0.63mm. If you want to use something even thicker you'll have to get longer studs...not sure its a big problem...
Some of you might find this interesting....
Aluminum Nitride (AlN) Substrates | Products | MARUWA CO., LTD.
Aluminum Nitride (AlN) Substrates | Products | MARUWA CO., LTD.
Good news, another mm could buy one a very significant amount of strength and still be ahead in absolute thermal conductivity relative to Alumina. If/when I do this in the future, I think I would very seriously consider aluminum nitride.
The thermal conductivity is massively superior and strength is just as good for the most part (just slightly lower fracture toughness).
There are a decent number of good suppliers including the one Jonathan notes. Check out Kyocera also as they are traditionally quite strong in the technical ceramics field. However, finding substrates machined to spec may be a challenge.
Probably the best thing would be to get an applications engineer from one of these companies on the phone and talk through the issue. They may actually be interested in this application.
I thought I read that Gruber is using a non stock solution for the thermal conductor in their PEM rebuilds, does anyone know what it is?
thefortunes
Active Member
I should be able to answer that question, as I have ordered their strips and hope to receive them shortly.
Well.. I have been speaking to several individuals lately about the insulating material in the PEMs. Gruber does use a non-stock solution because the OEM insulating material degrades over time causing cracks and splits from thermal dilations. However, they also optioned for a solution that adds an extra layer of protection in the form of a second layer of rubber coated fiberglass. The second layer provides tear resistance and electrical insulation should the thermally conductive polymer degrade as the OEM insulating material does. That in turn saves the electronics from catastrophic failure—e.g. blown IGBTs due to shorting, capacitors, fuses, etc.
Why doesn't Gruber use the Aluminum Oxide strips for thermal insulation as few have suggested? There are definitely pros to using it like excellent thermal conductivity at 24.0 W/m-K. The cons have me weary due to the high cost, fragility, and possible low lifetime, "> 1,000 cycles". I have a feeling @DanielFriederich is going to find the Aluminum Oxide strips cracked from shock/vibration while driving owing to the fragility of the Aluminum Oxide. A discussion with an applications engineer at CeramTec should clear that up.
DanielFriederich
Member
I think this will be most likely the cause since everything else still is online in the car... It was picked up by GS Technology today and taken to Switzerland... The engineer told me that they dont use the oxide stripes due to possible crack failure but a thermal paste similiar to the original one but with a better heat resistance and slightly better values in thermal conductivity... lets see what they say beginning january after taking the PEM apart... btw: still waiting for the visium hardtop.... tbc...
Well I spoke to my supplier about Aluminium Nitrate...he said it is less fragile - which is good - better thermal conductivity (also good) but he said to me it was 3 x the price of Aluminium Oxide. He has enough in his shop do do 1 car (12 strips) but said to me the lead time to get more would be 2 months. He indicated a price to me for 12 strips (0.63mm) would be around €500. Anyway FWIW....
Similar threads
