WARNING: Gotway Nikola production (assembly) issue [Solved]

[Smoother]

30 minutes ago, wheelr said:

In my opinion GlueWay has to:

[Marty Backe]

42 minutes ago, stephen said:

Be nice to see how it does on over heat hill with the new motor 👍

I think the faceplant will be always be in the back of his mind now though on the newer wheels , 

@Marty Backe are you getting the wheel still if so when ,,ish😊

 

I'm not buying one right now, if that's what you're asking. If I'm getting the demo unit I'll know when it arrives at my front door. Otherwise I have no idea.

[Phil McLaughlin]

I have some new data on the glue issue. As a reminder to those joining late my Nikola was fully operational at the time of teardown and this exercise was undertaken as a precautionary measure. I was able to get back to examining the board, pad and heatsink this morning and things were a little different than they first appeared. The first finding is that there was no detectable glue remaining on the metal contact surfaces of the MOSFETs or the MOSFET contact side of the thermal pad. I verified this with a microscope and also tested the surface with an ohmmeter. It appears that the adhesive substance was squeezed out when the mounting screws were tightened during assembly. This is kind of surprising. An additional finding was that some of the adhesive had squeezed out from under the MOSFET and ran over the edge of the thermal pad onto the aluminum heatsink and actually ran a bit back under the heatsink side of the pad. All of this is visible in the attached images. This seems to indicate that on this particular board the glue was of very low viscosity at the time of MOSFET attachment. This may or may not be hot glue but I believe it is some manner of adhesive.

An important takeaway from this additional inspection is that in my analysis this particular board was in no way compromised by the use of the adhesive since there was bare metal on both sides of an apparently clean thermal pad as seemed to be the design intent. I am not sure whether this is good news or bad news but it points to the likelihood that are a number of Nikolas that have been assembled in this manner that will not have a negative effect from this assembly method.

Additionally on reexamination of the pictures of the two burned boards there seems to be considerably more adhesive and it seems to have balled up more at the MOSFET edges in contrast to the adhesive on my board that appears to have been less liberally applied and more free flowing at the time of assembly. This may be a clue as to which boards are vulnerable to failure but I am not convinced at this point without additional data.

I am conducting some thermal tests later today and hope to have additional data to report by tomorrow.

Here is the middle thermal pad in position:

With the thermal pad removed showing the glue running under the other side of the pad:

Showing the MOSFET side of the thermal pad surface. Clearly shows the squeeze out of the glue from under the pad.

A macro photo of the MOSFET thermal contact surface showing the exposed metal:

 

Edited June 24, 2019 by Phil McLaughlin
I did not attach photos on the original post

[Phil McLaughlin]

It looks like the pictures did not attach so I will try again with help from Nils!

Here is the middle thermal pad in position:

 

With the thermal pad removed showing the glue running under the other side of the pad:

Showing the MOSFET side of the thermal pad surface. Clearly shows the squeeze out of the glue from under the pad.

A macro photo of the MOSFET thermal contact surface showing the exposed metal:

Edited June 24, 2019 by Phil McLaughlin

[Nils]

Just now, Phil McLaughlin said:

It looks like the pictures did not attach on my last post so I am trying again...

Third time's the charm! After uploading the pictures, I believe you need to click on them as well for them to appear in your post.

[Chriull]

1 hour ago, Kens said:

I'm not expert in this but how about non-conductive thermal paste like Thermal Grizzly Kryonaut? Their website explicitly state that this paste has no electrical conductivity.

If one mounts a mosfet to a heatsink one puts two metal plates together. The thermal paste just is used to fill up the small mikro irregularities of both surfaces, as the otherwise trapped air would be a great insulator.

If the thermal paste is electricly conducting or not does not matter - there will be metal to metal contact. Obe wants to put some pressure on the mosfet to get a good contact to the heatsink.

So one possibility is to use mica or ceramic plates, with thermal paste on both sides between the mosfet and the heatsink.

The other possibility is to use such thermal pads without anything else between the mosfet and the heatsink.

One wants as less as possible between the metal parts, because everything mentioned above has a worse thermal conductivity than the metal parts!

I would not recommend anyone here to mess with the heatsink or try to "improve" it obeselves, if one has not the knowledge and experience! It's very easy to mess things up...

Next detail would be how to isolate and fix the screws, which hold the mosfets down....

[Nils]

4 hours ago, meepmeepmayer said:

If I were in your position, I'd make a plan how to remove the glue, and find a nice non-conducting thermal paste. Then I'd disassemble the board, look if there's glue and remove that if there is, and put it back together with the original heat pad and the thermal paste on both sides of the heat pad.

Then you should have a good board, right?

How does one remove such glue cleanly without damaging the mosfet plastic? And ideas?

From what I understand using both a thermal pad and thermal paste is not necessarily an improvement on only having a thermal pad, but could instead be the opposite as you'd want a minimum of hindrance for the heat dissipation. Also, re-using the same thermal pad seems like it might not be the best idea from what I've read so far, plus I believe that it should be possible to find a thermal pad of better quality and with better characteristics than what Gotway put on it for the stock heat sink thermal pad.

Good question with regards to removing the glue thing, especially as Marty suggests it's hard sticking. It wouldn't have to be complete removed though I gather as long you're getting the tabs cleanly exposed - extra glue or whatever it is shouldn't have much of an impact. For now I see it as one step at the time - when I get the material in place I can have a look at it. If my board has the same issue then perhaps replacing the MOSFETs would be the good option, and if not then it's "only" the heat pad to worry about (and the insulating sleeves for the fastening screws? Judging from the photos this far they're popping out upon disassembly - still have to do some research about that and would appreciate some insight from the resident experts). I've ordered replacement MOSFETs for now as well if needed, but they will take some to arrive.

Anyway, I'm going to see this as a learning opportunity to continue my education in electric components (glass is half-full!), but will not hurry things as it's kind of important to get it all right, and there's much learning to be done. We'll also see how everything unfolds, and I look forward to hearing/seeing more from @Phil McLaughlin's work on it before proceeding.

Edited June 24, 2019 by Nils

[meepmeepmayer]

So a better fix would be using new thermal pads? Where would one get those?

@Chriull The idea was never to only use paste. Only in addition to the pads as these alone apparently cannot be reused. Whether that makes sense or not I don not know.

[Chriull]

3 minutes ago, meepmeepmayer said:

So a better fix would be using new thermal pads? Where would one get those?

At any electronic supplier as conrad, pollin, mouser, etc ...

[Marty Backe]

25 minutes ago, Phil McLaughlin said:

It looks like the pictures did not attach so I will try again with help from Nils!

Here is the middle thermal pad in position:

 

 

With the thermal pad removed showing the glue running under the other side of the pad:

 

Showing the MOSFET side of the thermal pad surface. Clearly shows the squeeze out of the glue from under the pad.

 

A macro photo of the MOSFET thermal contact surface showing the exposed metal:

Great follow-up Phil. This story just keeps getting more interesting. I do take this a good news, but unfortunately it's still a bit of a roll-of-the-dice as to whether any particular Nikola owner has an issue. Looks like it depends on the assembler and how much glue/adhesive they squirt on the MOSFETs during assembly. Wonderful 

[Phil McLaughlin]

2 hours ago, Marty Backe said:

Exactly. I can tell you that it's not possible to peel it off. Perhaps @Phil McLaughlin will be able to say something since that was his plan (or was he going to replace the affected MOSFETs?).

Following the closer examination of my board I believe that it is not necessary to change out my MOSFETs. I have not tried removing the glue yet but as you stated it is not possible to just peel it off. I will try some things and report what seems to work best.

[Nils]

50 minutes ago, Phil McLaughlin said:

I have some new data on the glue issue. As a reminder to those joining late my Nikola was fully operational at the time of teardown and this exercise was undertaken as a precautionary measure. I was able to get back to examining the board, pad and heatsink this morning and things were a little different than they first appeared. The first finding is that there was no detectable glue remaining on the metal contact surfaces of the MOSFETs or the MOSFET contact side of the thermal pad. I verified this with a microscope and also tested the surface with an ohmmeter. It appears that the adhesive substance was squeezed out when the mounting screws were tightened during assembly. This is kind of surprising. An additional finding was that some of the adhesive had squeezed out from under the MOSFET and ran over the edge of the thermal pad onto the aluminum heatsink and actually ran a bit back under the heatsink side of the pad. All of this is visible in the attached images. This seems to indicate that on this particular board the glue was of very low viscosity at the time of MOSFET attachment. This may or may not be hot glue but I believe it is some manner of adhesive.

An important takeaway from this additional inspection is that in my analysis this particular board was in no way compromised by the use of the adhesive since there was bare metal on both sides of an apparently clean thermal pad as seemed to be the design intent. I am not sure whether this is good news or bad news but it points to the likelihood that are a number of Nikolas that have been assembled in this manner that will not have a negative effect from this assembly method.

Additionally on reexamination of the pictures of the two burned boards there seems to be considerably more adhesive and it seems to have balled up more at the MOSFET edges in contrast to the adhesive on my board that appears to have been less liberally applied and more free flowing at the time of assembly. This may be a clue as to which boards are vulnerable to failure but I am not convinced at this point without additional data.

I am conducting some thermal tests later today and hope to have additional data to report by tomorrow.

[..]

Very interesting Phil, thanks a lot, looking forward to your future observations!

So, assumedly then we're looking at strings of some adhesive substance being applied to the borders of the MOSFETs before fixing it to the heat sink, and the amount and placement of said substance will determine the heat dissipation properties of the motherboard and MOSFETs in question. That would explain some things, since it seems to me that different Nikolas are able to cope differently with load. It never felt reasonable to me that my Nikola would hold up so well while Marty's should fail in that manner (and same for @Lucas Alexander Oliver's first failing Nikola board that didn't appear to have much load on it to begin with). Still, it's very much a lottery for any given buyer, and even for people with well performing wheels you'd always have to wonder if it's a ticking bomb.

[Nils]

1 hour ago, Marty Backe said:

Great follow-up Phil. This story just keeps getting more interesting. I do take this a good news, but unfortunately it's still a bit of a roll-of-the-dice as to whether any particular Nikola owner has an issue. Looks like it depends on the assembler and how much glue/adhesive they squirt on the MOSFETs during assembly. Wonderful 

We'll it's actually a good case for bigger margins over all - typically I'd assume you'd worry about the electrical/physical properties of the design but the actual assembly plays as much a part as anything else so not only add some margins for the physical properties of the design, but add some for the Joe (or the overseer/manager) at the assembly line as well. Assume gross negligence across the board and plan for that and all might work out well!

Edited June 24, 2019 by Nils

[Marty Backe]

4 minutes ago, Nils said:

We'll it's actually a good case for bigger margins over all - typically I'd assume you'd worry about the electrical/physical properties of the design but the actual assembly plays as much a part as anything else so not only add some margins for the physical properties of the design, but add some for the Joe (or the overseer/manager) at the assembly line as well. Assume gross negligence across the board and plan for that and it might work out well!

Absolutely. I'll probably only recommend that people buy the 100-volt version because of this.

[Phil McLaughlin]

1 minute ago, Nils said:

Very interesting Phil, thanks a lot, looking forward to your future observations!

So, assumedly then we're looking at strings of some adhesive substance being applied to the borders of the MOSFETs before fixing it to the heat sink, and the amount and placement of said substance will determine the heat dissipation properties of the motherboard and MOSFETs in question. That would explain some things, since it seems to me that different Nikolas are able to cope differently with load. It never felt reasonable to me that my Nikola would hold up so well while Marty's should fail in that manner (and same for @Lucas Alexander Oliver's first failing Nikola board that didn't appear to have much load on it to begin with). Still, it's very much a lottery for any given buyer, and even for people with well performing wheels you'd always have to wonder if it's a ticking bomb.

Really good points! My thinking is that there is an adhesive that is applied in liquid state at the time of joining of the thermal pad to the inner row of MOSFETs. I do not think we know at this point whether the adhesive is melted, two-part or solvent based. I am also thinking that it is applied directly to the entire tops of the middle three MOSFETs that are facing the heatsinks. In the case of my board there was also application to one of the MOSFETS on the end. The attachment of the MOSFET's to the pad and heat sink ideally squeezes out the adhesive leaving the junction with the thermal pad virtually adhesive free. This seems to be the case on my board. I think that if there was a very, very thin layer of this adhesive that is not even visible on the pad it would probably not have much thermal impact. Themal glues have thermal conductivity that is poor but not terrible (maybe 0.2 to 0.3 W/M-K); maybe 3 to 5 times less thermally conductive than thermal pad material (typical .9 W/M-K). But a very, very thin, non-visible layer will probably have limited impact. Just a guess.

I think from current evidence that you are absolutely correct. I am thinking that some (or many, or the overwhelming majority?) of Nikolas may not be ticking time bombs. On the other hand a few of them have shown themselves to be exactly that. Doing more tests...

[Michael Tucker]

1 hour ago, Marty Backe said:

Absolutely. I'll probably only recommend that people buy the 100-volt version because of this.

Let's see how they do with the 100V version!

[Smoother]

I can't say I've read all the posts on this subject but there doesn't seem to be much factual (with references) based discussion about what actually thermal past is and how many different verities there are.  So I looked it up

Edited to remove unnecessary conjecture

https://www.techpowerup.com/articles/overclocking/134

 

EDIT I had this thought a wile ago when Marty was describing the specific location of the substance and someone might have mentioned it).  It would appear that this dab of hot glue or whatever,  might be placed where it is to hold the thermal strip in place on this inaccessible row of mosfets long enough for the assembly to be completed. It's placement (only in the middle) suggests it is there merely to aid the assembly process and not to enhance the finished product.  I often apply a dab of something sticky to the end of a screwdriver, etc,  when assembling things in inaccessible places, like deep in the dashboard of a car, or fairing of a motorcycle.  Of course my applications are never thermal or electrically sensitive, with face plant consequences

 

Edited June 25, 2019 by Smoother

[Lukas83]

3 hours ago, Chriull said:

I would not recommend anyone here to mess with the heatsink or try to "improve" it obeselves, if one has not the knowledge and experience! It's very easy to mess things up...

Next detail would be how to isolate and fix the screws, which hold the mosfets down....

You can use loctite for this... holds pretty good. 

The electrical conductivity is a real problem...in my gaming pc I have liquid metal as a cooling paste, but in this case even non electrical cooling paste wouldn't help bc its to dangerous to get a short. I would try to change the mosfet to  to247 or even bigger ones and seal the little legs with clear coat - like nail polish. 

then build up the cooler wit some good quality thermal pad and fix everything tight together... get rid of the cheap fan and buy one from Noctua. Thermals should get a lot better and the reliability as well... 

 

[Phil McLaughlin]

2 hours ago, Phil McLaughlin said:

Following the closer examination of my board I believe that it is not necessary to change out my MOSFETs. I have not tried removing the glue yet but as you stated it is not possible to just peel it off. I will try some things and report what seems to work best.

Small nozzle heat gun quickly softened the glue enough that I was easily able to peel the layer off with a bit of patience. Surface comes clean. I have saved the peeled paste for further inspection.

[Jim Martin]

Following while my Nikola is grounded

[mrelwood]

6 hours ago, Marty Backe said:

Great follow-up Phil. This story just keeps getting more interesting. I do take this a good news, but unfortunately it's still a bit of a roll-of-the-dice as to whether any particular Nikola owner has an issue. Looks like it depends on the assembler and how much glue/adhesive they squirt on the MOSFETs during assembly. Wonderful 

I see it the other way round. If the blob of glue was clearly prohibiting cooling on all inspected devices, I think it would be easier to get proper call-back and repair systems in place. Currently one can still shrug at the issue like Gotway themselves seem to, thinking that it may not cause every Nikola to fail. Which is of course true, but nobody wants to buy a $2000 ticket to participate in russian roulette.

Every current Nikola owner must hear about the issue!

I also wouldn’t advise to purchase the 100V Nikola (or any other Gotway for that matter) until the specific model and version is checked for this issue.

[Dzlchef]

I feel like maybe my Nikola is going to be okay but who really knows, and that’s got my inner voice on edge but my body continues to trust. I would like a verified new control board installed but I pushed a bit harder today then I’ve done so far with success. This was a 21 mile ride from downtown up the Hollywood hills to Mulholland, riding along the hills then back down/across the hills again home. I chose 50C as my ceiling for heat and had to stop 2x to bring it back to 41C before resuming. The ride is mapped and the link listed below.  I was running some significant continuous current draws through a few sections.  I had my app current alarm ringing during the ascent, 100A, but the wheel only beeped at me once. 

I’m in love with this magnificent supermodel that has a couple weird visual flaws but is absolutely the best girl I’ve ever met. I only wish that I could trust her, but it hasn’t stopped me for going along for the ride!

https://euc.world/tour/576157899435318

Cheers!

 

[mrelwood]

18 minutes ago, Dzlchef said:

so far with success

Unfortunately that doesn’t mean that it can take the same stress the next time. Or after a week. Or month.

This girl is a back-stabber, don’t give her your bank credentials!

[stephen]

3 hours ago, Dzlchef said:

I’m in love with this magnificent supermodel that has a couple weird visual flaws but is absolutely the best girl I’ve ever met. I only wish that I could trust her, but it hasn’t stopped me for going along for the ride!

I hope that girl doesn't bitch slap 👋 you one day and send you to the ground 😁

[wheelr]

9 hours ago, Dzlchef said:

 

I feel like maybe my Nikola is going to be okay but who really knows

 

Why don't you take her to over heat hill to watch the sunset?