Jump to content
Marty Backe

My 84-volt Nikola Triumphs, Tribulations, and Failures

Recommended Posts

6 hours ago, Jason McNeil said:

Apologies for the late attendance to the party. Since receiving a 2nd board failure (this case was different, it was powered on from a stand still & didn't respond), including Marty's, within two weeks of shipping out the first batch of Nikolas, I've reached out to Linnea on Thursday to try to come up with an action plan to see what options are available to greatly reduce/eliminate the risk of these MOSFET thermal overload events.

The evidence from both of these failures has marked similarity in outcome (see below), where this glue may be acting as a thermal barrier, not helped by the absence of thermal paste on the underside of the metal heat transfer surface. One option being explored, is to recall the boards. remove this glue, & apply the thermal paste. To date Gotway have shipped 600 units; they say they have not received reports from other regions of this failure—it's likely more will trickle in in time. Putting out an official recall is not something to be taken lightly; in their defense, they're probably looking for a few other reports, before taking this step.

Gotway have a lot of experience with building high power motor vehicles; they had gone through several rough patches, particularly with reliability of earlier generation boards, with the weaker MOSFETS, but at least then they were pretty consistent with applying liberal quantities of thermal paste to this underside. It was baffling to me why they would go back to an inferior TO22x FET package AND now we learn about the lack of thermal paste. From my experience, such oversights are usually not the result of penny-pinching, but execution on the production line, where the factory needs a ruthless eagle-eyed floor manager, to see that every step of assembly is done by the book. 

The silver-lining to all this, is that an early finding in the production cycle is a necessary facet to a small-scale non-ISO production facility, where the manufacturer's financial pain of having to deal with a crisis, means that they will be imbued with a sense that this situation must never occur again. 

For all current Nikola owners, please hang in there, we should have an action plan in the next few days. 

EDIT: To anticipate questions on the 100V Nikolas, they are using the TO247 MOSFET package. The first production of these special edition 1845Wh variants is still underway, so there is assurance that in light of this episode, they will getting the thermal paste treatment!

uc?export=view&id=1YkbgpMpESOtmhsYpqSFfb

Do you know if this issue extends to the newest Tesla (speaker and anti-lift version)?  

Share this post


Link to post
Share on other sites
Posted (edited)
1 hour ago, Topkek said:

Do you know if this issue extends to the newest Tesla (speaker and anti-lift version)?

At this stage I don't believe there have been any reports of this issue with the latest Tesla board. The board used for the Tesla V2, was already an upgraded board that actually began getting installed in the Tesla V1 quite some time ago and was distinguished by a change to the "ringlight" pattern options (sadly losing the brake light option) so it has been around for a decent period of time now.

However, as mentioned elsewhere and possibly of some concern, the construction method is the same as the Nikola having one row of hidden mosfets along the centre of the board (so who knows whether the hot glue assembly method may be a factor here too!)

New Tesla board & fan.jpg

New Tesla board (1).jpg

Edited by fbhb

Share this post


Link to post
Share on other sites

Or they could just use an easily de-mountable piggyback motherboard like the Z10 so that all the FETs can be reached and affixed easily without having to juggle around holding things in place with poor access.

Share this post


Link to post
Share on other sites
1 hour ago, Harold Farrenkopf said:

@Jason McNeil, no thermal paste when using the thermal conductive pad. When the thermal paste is used where no electrical isolation is required, only the minimal amount is required to fill any air gaps in the irregular surfaces even though they look flat and smooth. 

One simple way of placing the pad in place with the spacers for the screws is to use a made tool for the task that looks like a 6 pronged comb that goes through the holes of the pc board and the mosfet holes and the pad slid on it and then the comb put into the heatsink holes and then slid down to the board. Then place the screws in place with no need for any glue or paste to hole the pad aligned. 

And if they really want a easy and fast precision assembly, use an assembly that has 2 rows of 6 to assemble all mosfets at once. 

Exactly. It took me about 60-seconds to think of this. Why the Chinese can't is beyond me.

Share this post


Link to post
Share on other sites

There is of course chance that new guy come to line and used hot glue to help him. As we don't know how many guys assemble and how many uses hot gun, it is really hard to judge. But given that we know of 2 cases, gateway need to investigate it...

Share this post


Link to post
Share on other sites
1 hour ago, Jason McNeil said:

Is there now max current regulation in the firmware? If so, what is this value, 100A max?

Afair some times ago (~?msuper v3?) the official statement from the GW representative was that the firmware limits the current to 120A.

Imo this can only be survived for short power peaks - such currents leads to dissipated powers in the range of ~140W (~120A² * 10mOhm), with paralleled mosfets to a fourth of this - but with no real guaranteed of equal distribution.

So measuring the compartment temperature instead of the heatsink temperatures near the mosfets can easily lead to too much time delay, overheating and destroying the mosfets.

Share this post


Link to post
Share on other sites

I would say it depends on board complexity and individual assembler. Also to give some peace of mind, remember this hot glue is still better than air, but worse than thermal pad/ paste.

Air> hot glue>thermal pad>thermal paste > toothpaste (for short time, as overclockers knows ;)

Share this post


Link to post
Share on other sites
Posted (edited)
2 hours ago, Jason McNeil said:

MOSFETs: What type of MOSFETs are used in the Nikola, are they HY brand, what model? Why are they not the TO-247 type found in the MSX? Is there now max current regulation in the firmware? If so, what is this value, 100A max?

Did he answer? If not:

The 84V Nikola (for now) uses HY3712P mosfets in a TO-220 package. Source: EcoDrift in the comments of this video. If you look at the data sheet, you see 170A continuous/585A(?) max. For comparison, the MSX has mosfets (I don't know which) with around 200A cont/600A max. So the mosfet change is no big downgrade itself. But I guess the smaller package instead of TO-247 will have a huge influence on cooling.

But given that wheels like the Tesla have survived with smaller mosfets than the Nikola (right?), I'm not convinced the mosfets itself are a problem. But of course it would be better if they just switched to the MSX setup for the 84V Nikola, which is apparently pretty much what they have done for the 100V Nikola. So if you could tell them that (again):)

Edited by meepmeepmayer

Share this post


Link to post
Share on other sites
51 minutes ago, scotchtape said:

So every gotway wheel had this issue? Including msx? :(

We don't know. Gotway need to figure out how widespread this problem is.

But there haven't been any failure reports from other 12-mosfet wheels like the MSX or Monster (the upgrade with the MSX board). So either the problem isn't there, or the stronger mosfets survived despite glue. Either way, no need to worry about existing non-Nikolas for now:) The MSX board has proven to be exceptionally strong and problem-free. Just look at @EUC GUY's car push test - the cabling died before the mosfets! So even if there is glue there, it's still no problem. But of course you don't want glue there, even if it does nothing.

Share this post


Link to post
Share on other sites
6 hours ago, Jon Stern said:

Well that wouldn't be a good sign either. In manufacturing it's vital that each operator is trained on, and follows, the Standard Operating Procedure (SOP).

This should be a document that provides precise, step by step instructions on how to perform the build. There is no room for interpretation. 

It's the job of QC to make sure that the SOP is followed. If there's an operator who is deviating from it, they should be taken off the line and retrained.

If the QC doesn't catch these things, that a huge red flag too. What else are they missing?

Jon, I admire your rattling off of standards, etc., in multiple posts, but these are Ma & Pop Chinese companies :D

Probably at least a couple of years ago I gave up any allusion to believing that they would ever operate or behave like Western companies.

Share this post


Link to post
Share on other sites
Posted (edited)
4 hours ago, meepmeepmayer said:

... So the mosfet change is no big downgrade itself. But I guess the smaller package instead of TO-247 will have a huge influence on cooling.

The troll again :D... As I said befor 7w for 1minute... Maybe this wasn't so wrong ;) ... As Jason said, the second was not burned on a long steep hill. 

MSX has an open shell... This could be named "cooling". 

Edited by TuN3M@N

Share this post


Link to post
Share on other sites
On 6/20/2019 at 11:56 PM, Marty Backe said:

The Great Hot-Glue Fiasco

OK folks, here's the new theory. The "stuff" that was stuck to my burned MOSFETs was thought to be a thin plastic backing from the thermal transfer material. But it never quite seemed like plastic film to me, or melted plastic film. A bit rubbery actually as I tried to peel it off.

I believe it's hot glue. It melts easily when I touch it with my hot-glue gun.

So I did a little experiment. I put a dollop of hot glue on one of the MOSFETs and pressed the thermal transfer sheet onto. A little bit later I pulled it off and it peeled off just like the other one.

Here are some pictures that show the sequence of my experiment

IMG_20190620_204054508

IMG_20190620_204106689

IMG_20190620_204133823

IMG_20190620_204231918

 

IMG_20190620_204248013

Not a perfect match, but who knows what type of hot-glue or applicator is being used. After poking and melting the original substance, I can say it's some type of hot-glue.

Why?

To assemble the circuit board to the heat-sink plate, the board has to be flipped over and the thermal transfer sheet must be between the MOSFETs and the heat-sink. After flipping the board, it's easy for the assembler to lift up the board a little and slide the sheet under the MOSFETs. But the inner row of MOSFETs (the one with the blown MOSFETs shown above) can't be easily accessed to slide the thermal sheet under the MOSFETs. Imagine yourself trying to do this.

So instead of developing a fixture to align everything properly, I'm theorizing that the assembly line folks pulled out the glue-gun (the same one that they use to stick the pedal magnets on the shell and a bunch of other components - us Gotway owners know that Gotway loves hot-glue) and glued the thermal sheet to the MOSFETs so the sheet would stay in place when they flipped the board.

Seems to make sense to me, knowing Gotway and their tight oversight of QC :facepalm:.

Still waiting for feedback from EWheels. I'm pretty sure Jason is in communication with Gotway so hopefully we'll learn the real scoop eventually.

But if my theory holds, seems to me that every Gotway wheel that has an inner row of MOSFETs (not just the Nikola) is a time-bomb waiting to blow. Now I think the other wheels (MSX, Monster, etc) are using the bigger MOSFETs and the hot-glue thermal barrier might not be an issue for those. But when they switched to the smaller MOSFETs, the assembly process became the weak link.

BTW, hot-glue melts at ~120-degrees Celsius. That's probably hotter than these MOSFETs are supposed to get, so the hot-glue likely stays a solid and acts as a nice thermal break.

This is my theory, and I'm sticking to it :)

I miss my Nikola :crying: 

Can't be hot glue.  First of all what hot glue do you know looks like a perfect thin sheet of plastic.  Any thin sheet of plastic would melt with heating tip.  

If it happen like theorized, then it would be globs, not a perfectly flat sheet inbetween the mosfets.

Share this post


Link to post
Share on other sites

I think you're looking at the wrong bit...

Share this post


Link to post
Share on other sites
4 hours ago, eddiemoy said:

Can't be hot glue.  First of all what hot glue do you know looks like a perfect thin sheet of plastic.  Any thin sheet of plastic would melt with heating tip.  

If it happen like theorized, then it would be globs, not a perfectly flat sheet inbetween the mosfets.

Did you see Jason's picture of the other failure? Looks like a sloppier application of hot glue, to me anyway. I guess you'd have to see it up close and touch it. Feels and acts just like hot glue.

uc?export=view&id=1YkbgpMpESOtmhsYpqSFfb

Share this post


Link to post
Share on other sites
14 minutes ago, Marty Backe said:

Did you see Jason's picture of the other failure? Looks like a sloppier application of hot glue, to me anyway. I guess you'd have to see it up close and touch it. Feels and acts just like hot glue.

uc?export=view&id=1YkbgpMpESOtmhsYpqSFfb

Nearly all the mosfets are burnt and some are incinerated 😬😬

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...