V12 Cutout tracking

[Paul g]

I found an interesting material about dual redundancy BLDC motor:

https://www.researchgate.net/publication/347847348_Performance_analysis_of_dual-redundancy_brushless_DC_motor

[Gixxer]

2 hours ago, mrelwood said:

The mosfets’ voltage rating does sound like a plausible reason, but there are two aspects that make it hard for me to believe it as the root cause:

1) These failures are known to have happened in about a dozen units. Each of them has 12 mosfets, which would increase the odds for failure substantially. Yet they would’ve only failed in less than 0.1% of the units sold. An obvious issue like that should result in a much larger failure rate.

2) Inmotion has manufactured probably a hundred thousand mosfet driven BLDC motored vehicles. I’d like to lean towards thinking that they’d know how to design such a controller a bit better than any of the forum members, electronics engineers or not. I know that sometimes GW seems like they don’t have a single EE in their stables, but IM is not GW.

 We should be hearing from IM in a day or two. It’s a difficult wait, but I’d try to stay in my pants until they have something to report.

I think the the issue is we live in a world of approximation especially in manufacturing. Everything has a tolerance.  So the manufacturer of the MOSFET has some internal tolerance they have to hit.

 

The following section is completely hypothetical

 

Let's say the internal manufacturer rating is 115v with a +/- 5v so a MOSFET from this manufacturer could start leaking at 120v or at 110v depending on your luck. I'm not sure what the max bus voltage for the wheel is but if it exceeds that lower limit you start leaking voltage and slowly damage the MOSFET until it fails.

Back to reality ( oops there's still gravity)

 

Now the spread of these tolerances will not me statically even. As a cnc machinist I generally try to hit the tolerance at an area that gives me the highest probability of success. Regular old aluminum. Get it in that zone and go. Aluminum getting anodized or steel getting coated, low end of that zone. Steel in high volume. Low end knowing it will creep to the high end after wear. 

 

I can't speak for electronics manufacturing but I'd imagine there's similar considerations.

 

 

 

 

[Tawpie]

36 minutes ago, Gixxer said:

I'd imagine there's similar considerations.

Absolutely there are. With 'complications'. It's quite common for parts to be individually or lot tested and sorted for adherence to tolerance specifications—you select out the parts that are "best" and label them as close tolerance and add a few cents to their price. Then you select the "pretty good" parts, they're wider tolerance and priced accordingly. The left overs are relatively cheap "wide" tolerance parts (few are "best", but some pretty good ones are in there). In times of shortage, the known best and pretty good parts are spoken for by customers that are willing to pay and the rest... they tend to sit around in bins until someone is desperate enough to buy them. They're pretty common on the gray market (think Ali but a little dustier).

When a supplier can't deliver, every manufacturer eventually consults the gray market and some will retest before using. But production targets can increase the pressure to accept wider tolerance parts than specified, and often the design has sufficient margin to spare so things work out. But on occasion the odds stack up and your luck runs out. The finished item may work off the line, and even work for the customer, but it's not built as it was designed so it's at risk.

The current 'chip shortage' may not be on our side here.

Edited January 2, 2022 by Tawpie

[Chriull]

5 hours ago, mrelwood said:

The mosfets’ voltage rating does sound like a plausible reason, but there are two aspects that make it hard for me to believe it as the root cause:

With given facts it's quite impossible to get the root cause.

Just if 100V mosfets were used this is a definitive possible cause for "random" cutouts. No idea if any of the happened were caused by such a design fault or cutouts caused by this would have happened lateron.

As far as i read and understood here till now we do not know if such mosfets were used in any v12 besides @EcoDrift's teardown v12 - which could have been some prerelease? But still worse enough...

5 hours ago, mrelwood said:

1) These failures are known to have happened in about a dozen units. Each of them has 12 mosfets, which would increase the odds for failure substantially. Yet they would’ve only failed in less than 0.1% of the units sold. An obvious issue like that should result in a much larger failure rate.

Having mosfets in parallel could double the chance. But normally/often electric compoments from one batch/reel tend to be similar.

So one can not assume a normal distribution and use general probability calculations.

5 hours ago, mrelwood said:

2) Inmotion has manufactured probably a hundred thousand mosfet driven BLDC motored vehicles. I’d like to lean towards thinking that they’d know how to design such a controller a bit better than any of the forum members, electronics engineers or not. I know that sometimes GW seems like they don’t have a single EE in their stables, but IM is not GW.

Using a 100V mosfet with 100.8V supply voltage with potential higher spikes and higher voltages during regen breaking is bs. That has nothing to do with experience - just having heard of mosfets drain source breakdown voltage and thinking of the simplified schematics in the other topic

https://forum.electricunicycle.org/topic/24258-inmotion-v12/page/27/?tab=comments#elControls_397671_menu

 is enough reasoning.

There is _no_ excuse for such a design choice, if done.

[supercurio]

Agree it is impossible to define a root cause for now (for us)

My best guess today is a combination of hardware and software factors:

• Somewhat off spec components on the new batch.
• Sensors data (gyro, motor, controller) coming into the boards with different values than accounted and tuned for during development.
• Unexpected self balancing and motor control algorithms input data leading to undefined behavior as output, resulting in cutout or cutout+MOSFET destruction.

If this guess proves right, all these issues might be fixable with a firmware update, handling the additional hardware characteristics introduced by components and production variations.

Root cause being only bad hardware was ruled out earlier by the technical Inmotion representative, but we'll see.

Edited January 2, 2022 by supercurio

[mhpr262]

8 hours ago, mrelwood said:

I’d like to lean towards thinking that they’d know how to design such a controller a bit better than any of the forum members, electronics engineers or not.

I dont doubt their engineers do, but I am not so sure about bean counters not forcing them to cut corners.

[goatman]

if you  get a cut out from hitting a bump, alot of times its a bad connection from my experience in the ebike world

 go look at Wrong Ways tear down video, motor wires cracking, pulling plugs off the control board, the top shell has a 13? pin wireless connector to the board, i was curious if theyre spring loaded or friction or magnetic

[Mike Roe]

Are all these recent problems on batch 2 units? Or people who have recently updated their unit? 
 

     I’m rolling on batch one and I haven’t updated since august. 
 

just curious 
  

[supercurio]

8 minutes ago, Mike Roe said:

Are all these recent problems on batch 2 units? Or people who have recently updated their unit? 
 

     I’m rolling on batch one and I haven’t updated since august. 
 

just curious 
  

Current data is all 1.5.x firmwares.

We can't tell if it's coincidental (most updated their wheel to get the newest features like split modes, everyone here is an early adopter after all) or significant.

Until confirmed, it seems prudent to not update.

[RArtem]

Hi there, from russian community. Here's some cutoff videos.

 

 

 

 

https://t.me/inmotionV12/34265

[supercurio]

Thanks @RArtem for sharing.

This video illustrates perfectly how a problematic V12 can trip and die over almost nothing like this tiny curb.

It makes it easy to see (literally) the pattern with this failure.

 

Concerning also in the comments are the multiple people reporting V10 and V8 board failing the same way, making it seem like an epidemic. This goes in the same direction as the report from a supplier describing how 30% of its summer V10F failed.

We know Inmotion uses some of the same components in the V10 and V12, like MOSFETs in early models and maybe more, so it's possible the failures are linked across models, either due to bad components or manufacturing issues with their boards supplier.

Edited January 3, 2022 by supercurio

[mkygod]

Someone have Marty take the V12 on overheat hill and then do a teardown.

[Richardo]

Does anyone have a video or other instructions for replacing the affected board? If possible I'd like to spend less than a full day tearing this thing down when I get my replacement parts. 

Also! In the three weeks or so it has been broken, my battery seems to have drained to 0%, so if yours is broken, I recommend you check on it and detach the batteries from the motherboard. (I can still connect to the inmotion app if the wheel is plugged in to a charger)

[Paradox]

Everyone with a bad wheel should send it back.  This is InMotion's problem not the user. 

[Khazik]

Is it possible these cut-outs are caused by any increased power allocation values in "Fancier" mode (43mph limit) vs. "Normal" mode (37mph limit)?

It would seem to me the Fancier mode allows too much spike voltage than the hardware can support. I'm no electronics engineer, but I wouldn't be surprised if this was all caused by over-adjusted values in the firmware update (although 2 incidents on the spreadsheet list 1.5.0, all others are 1.5.4/Latest version). 

I wish we had some technical patch notes to review on what was changed in the update to help us all narrow it down.

[Gixxer]

1 hour ago, Khazik said:

Is it possible these cut-outs are caused by any increased power allocation values in "Fancier" mode (43mph limit) vs. "Normal" mode (37mph limit)?

It would seem to me the Fancier mode allows too much spike voltage than the hardware can support. I'm no electronics engineer, but I wouldn't be surprised if this was all caused by over-adjusted values in the firmware update (although 2 incidents on the spreadsheet list 1.5.0, all others are 1.5.4/Latest version). 

I wish we had some technical patch notes to review on what was changed in the update to help us all narrow it down.

I was in normal mode. We should add that to the spreadsheet though.

 

Good idea checking for that!

[Gixxer]

On 12/31/2021 at 1:23 PM, Richardo said:

Added two more events to the sheet this morning (Om moped, petranodon) and changed the speed column to include the actual speed when known

 

Also, Inmotion made an official announcement:

 
 

One question I have is why isn't this posted in the "Inmotion Official Announcements and Information" page?

[mrelwood]

10 hours ago, Gixxer said:

One question I have is why isn't this posted in the "Inmotion Official Announcements and Information" page?

Cecily and Fiona are new forum members, I’m not sure if they have access to that part of the forum yet. It is in the works though.

[MexicanBatman]

Any failures from the initial run of 150 wheels?

 

mines from that first batch

[MegaObi]

Just came back from a 60km ride at least 15km were done in real bumpy trails going specifically 20-35km no issues.

ive done over 1000kms no issue so far 

i also asked the guy on YouTube (omped) he did confirm he was in fancier mode.

i believe this may be an issue with fancier mode unless someone else who had a cutoff can chime in and let us know. 🤙🏽 Ride safe.

[spitfire1337]

12 hours ago, Gixxer said:

I was in normal mode. We should add that to the spreadsheet though.

 

Good idea checking for that!

I can add this information if those who have reported incidents can provide the data.

[Chester Copperpot]

No fancier mode on mine. 

[SiViG]

On 1/3/2022 at 9:34 AM, RArtem said:

Hi there, from russian community. Here's some cutoff videos.

 

https://t.me/inmotionV12/34265

In a followup video

he states that two burned mosfets had to be replaced, and the wheel is operable

[Wolverine]

Here is the new Inmotion announcement regarding to the V12 cut out cases.

[Richardo]

and here is InMotion's video on how to do the test: 

 

Here is a video on how to replace the board- maybe save some time compared to disassembling everything following a full-wheel tear down guide that's easy to find on Youtube. (Shared to me by ewheels)
https://drive.google.com/file/d/1kGFPG586w72mJ_x6L_oneaS7eJEdmism/view (original is on google drive, reuploaded to YouTube below)

 

Here is a video on how to check/replace individual mosfets, if you're feeling adventurous. Note that lead solder is less brittle than lead-free, and can make a more reliable connection- if you completely remove all other solder (don't mix types, thanks @ShanesPlanet). Also, tawpie had a bunch of good info

 

Edited January 5, 2022 by Richardo
more videos