My Sherman Trials, Tribulations, and Triumphs

[Alj]

3 hours ago, Marty Backe said:

Here are some additional pictures.

After further consideration, I think there are issues with their hardware design. The amount of board destruction that occurred is something I don't recall ever seeing on any Gotway wheels. The MOSFET's self-destructed long before the board could delaminate and melt down like this.

Am I wrong in concluding that they have not sized the components (including the circuit board traces) to support the amount of current that they are allowing to flow through the board?

I would be concerned that even a momentary (few seconds) high load could cause severe weakening of the board.

Basically, if the board fails it should be some components (MOSFETs) not the circuit board itself. Right?

What do any of you hardware guys think?

From my experience of smelling magic smoke it is usually that overloaded components go either without boom or they just have a small hole out of which "magic smoke" escapes. I havent seen the cases where the whole PCB overheats and separates into layers like that. This can happen by one of the 2 reasons (or combination of 2) 1. PCB wires got too hot due to current.... But usually when this happens PCB wires quickly evaporate and PCB itself stays intact.  2. Inadequate heat sinking. Which is most likely the case here, also, most often PCB goes smoked from hot component on it and not from current flowing thought it, which also makes it most likely the case. Also if heat transfer was adequate the temperature of mosfet would not be much different from the temperature of heatsink and heatsink was not hot enough to melt anything from what it looks like from the pictures.

IMO silicone pad (sil-pad (tm)) used for MOSFET insulation here (between MOSFET and heat sink) while being cheap go-to way of insulating MOSFETs from the heat sink, is inadequate way of insulation/heat transfer in this application because silicone is just "good enough" for transferring the heat but not very good for this particular heat dissipation levels. THe alternative can be ceramic insulation (which is more expensive) or, better, schematic solution that does not require MOSFETs to be insulated because direct mounting might be required for this power levels....

If schematic solution is not possible, for example, because 2 groups of MOSFETs are running in different signal polarity but they can be joined within the same group the solution can be to directly mount 2 groups of mosfets to 2 not connected copper or aluminum plates and then mount those plates on a heat sink using insulation of choice (sil-pad?), in this case area of heat transfer through the silicone will be much higher.  In any case it is not very hard or time consuming to test this setup with thermometer under load and figure out if solution is working properly.

 

there is also heat transfer happening from the silicon to the MOSFET enclosure pad but if that would be the cause of MOSFET burning then MOSFET legs (and PCB) would stay cold and it would be just a small hole in MOSFET with magic smoke.

 

for the reference here's random youtube i found about guy testing different types of insulation and silicone pads is about the worst of them (but the cheapest one of course).

 

Edited July 16, 2020 by Alj

[yuweng]

4 hours ago, Marty Backe said:

What do any of you hardware guys think?

Spoiler

 

First things first, for such a high current machine, they shouldn't be using this at all imo

[meepmeepmayer]

4 hours ago, Mr. Thompson said:

First I would like...

You're completely right. But this has been like this forever with EUCs, and I don't think it will ever change

You would think "Look at the maximum current/voltage/whatever this thing could produce, and let's use components that can take more" would be a no-brainer. But apparently it is not.

I don't think they (all of them!) are cheap, they simply don't understand (on some level) that things should be done differently. Right now we have "Usually it just overheats and stops the rider before anything actually fails" level of design, and I don't think the Chinese manufacturers see this as anything but a successful design.

[meepmeepmayer]

5 hours ago, Marty Backe said:

After further consideration, I think there are issues with their hardware design.

Definitely. It's a Gotway clone design that breaks down as they pump more power/current through it than ever before.

-

Could it be that capacitor suddenly discharged and burned that hole into the board?

Is there a timeline of what failed first? The wheel cut-off, and then you could hear the popcorn crackling which were the mosfets. So what failed in which order?

[Marty Backe]

4 hours ago, yuweng said:

  Hide contents

 

First things first, for such a high current machine, they shouldn't be using this at all imo

Those are the connectors that go to the charge ports. Current is much lower here.

[Marty Backe]

3 hours ago, meepmeepmayer said:

Definitely. It's a Gotway clone design that breaks down as they pump more power/current through it than ever before.

-

Could it be that capacitor suddenly discharged and burned that hole into the board?

Is there a timeline of what failed first? The wheel cut-off, and then you could hear the popcorn crackling which were the mosfets. So what failed in which order?

It all failed rather suddenly. I really couldn't say what happened first.

[Marty Backe]

Failure Resolution?

A new control board is on its way to me. What did Veteran change?

• Apparently no hardware changes
• Firmware modified: "Now the current will be lessen once temperature goes up, it will stop you to ride on high temperature. After it cool down, it become normal again"
• They did some "climbing tests" with a 80kg rider (a real Chinese heavy weight apparently) and "it is fine"

Hmmm

So clearly the Sherman will never zoom up Overheat Hill like Jason had hoped and it will instead overheat like many previous wheels (I think the V10F being the last one).

Edited July 16, 2020 by Marty Backe

[.....]

So instead of attempting to make it a 'better' wheel and investing in a more robust design, they have programmed it to cripple  instead? Well, this IS a solution, but I'm sure I'm not the only one who reads into this a little bit. Hmm, at $3k and more expensive than my daily driver car, doubts have begun to creep into my head. I love the way they passively undermine Marty's findings by claiming its fine with a midget riding it. I'm sure the wheel can outride me TODAY, but it seems like eventually I will be outridding it in the end. Such choices.... Spend 3 months salary on a company that cripples (rather than strengthen), a wheel, in efforts to avoid failure. Or, save the $$ and maybe give some of these new designer drugs on the street, a try? decisions decisions...

Edited July 16, 2020 by ShanesPlanet

[Marty Backe]

9 minutes ago, ShanesPlanet said:

So instead of attempting to make it a 'better' wheel and investing in a more robust design, they have programmed it to cripple  instead? Well, this IS a solution, but I'm sure I'm not the only one who reads into this a little bit. Hmm, at $3k and more expensive than my daily driver car, doubts have begun to creep into my head. I'm sure the wheel can outride me TODAY, but it seems like eventually I will be outridding it in the end. Such choices.... Spend 3 months salary on a company that cripples (rather than strengthen), a wheel, in efforts to avoid failure. Or, save the $$ and maybe give some of these new designer drugs on the street, a try? decisions decisions...

They need to be shipping wheels now. There's no time for real changes.

I'd obviously prefer that a wheel overheat and safely stop me from riding (all wheels historically do this) instead of self-destructing. But on the other hand, the community has traditionally hated wheels that would overheat easily and prevent you have having fun (like climbing hills).

The reason my test hill is called "Overheat Hill" is that mostly it would detect which wheels had poor heat dissipation which would prevent a wheel from making it to the top. Now the Sherman will join the club of wheels that overheat easily. 

[HippoPig]

5 minutes ago, Marty Backe said:

They need to be shipping wheels now. There's no time for real changes.

I'd obviously prefer that a wheel overheat and safely stop me from riding (all wheels historically do this) instead of self-destructing. But on the other hand, the community has traditionally hated wheels that would overheat easily and prevent you have having fun (like climbing hills).

The reason my test hill is called "Overheat Hill" is that mostly it would detect which wheels had poor heat dissipation which would prevent a wheel from making it to the top. Now the Sherman will join the club of wheels that overheat easily. 

What sort of temp should I be keeping an eye out for?

 

In the U.K. so it’s not too hot here but she runs quite hot at a standstill.

[Marty Backe]

36 minutes ago, HippoPig said:

What sort of temp should I be keeping an eye out for?

 

In the U.K. so it’s not too hot here but she runs quite hot at a standstill.

I have no idea what their set point will be. The temperature is referring to the the control board temperature. Traditionally this has been 79-degrees but I suspect it's going to be much lower with the Sherman.

 

[Alj]

5 hours ago, meepmeepmayer said:

 

Could it be that capacitor suddenly discharged and burned that hole into the board?

 

After further looking at the more recent videos capacitor actually did not get exploded, it got soldered itself out due to PCB board going in smokes due to MOSFET overheating due to poor heat transfer through the silicone pad. IMHO.

[Boogieman]

14 hours ago, Marty Backe said:

Failure Postmortem

All the wiring external to the control board looks great and I see no evidence of heat stress 

 

Allways so positive 😁 

No heat stress except your cable ties melted from the temp of the motor cables, a DIY solution that might have prevented THAT.

Could it have been a case of solder melting, making contact with the aluminium hest sink causing the short and vaporizing the board at that point, i Saw a peep hole in the board at one spot where the capacitor was soldered dowm 🙈 (but other capacitor had the mosfet imsusltion between its solder points and the heat sink)

Maybe they should move from regular solder to silver "solder" (not sure of the word in English, the kind that need blue flame to melt) to prevent meltdowns 🥰 or at least add a insulating layer between heat sink and solder side of control board.

Anyway, if the solder melted, it was way to hot for the config.

Edited July 16, 2020 by Boogieman

[Mark Wilson]

3 hours ago, ShanesPlanet said:

So instead of attempting to make it a 'better' wheel and investing in a more robust design, they have programmed it to cripple  instead? Well, this IS a solution, but I'm sure I'm not the only one who reads into this a little bit. Hmm, at $3k and more expensive than my daily driver car, doubts have begun to creep into my head. I love the way they passively undermine Marty's findings by claiming its fine with a midget riding it. I'm sure the wheel can outride me TODAY, but it seems like eventually I will be outridding it in the end. Such choices.... Spend 3 months salary on a company that cripples (rather than strengthen), a wheel, in efforts to avoid failure. Or, save the $$ and maybe give some of these new designer drugs on the street, a try? decisions decisions...

I am not surprised that this will be the solution. The wheel was never designed for slow speed, high torque applications. It's a high speed, comfortable cruiser. I wasn't that disappointing in the wheel not being able to make it up overheat hill, I was disappointing in it self destructing while trying to do so. Assuming that the performance of the while isn't limited until shortly before what would be the failure point, I think limiting the power then is the right choice, given the time constraints to have them mass produced.

[travsformation]

2 hours ago, Boogieman said:

Could it have been a case of solder melting, making contact with the aluminium hest sink causing the short and vaporizing the board at that point, i Saw a peep hole in the board at one spot where the capacitor was soldered dowm 🙈 (but other capacitor had the mosfet imsusltion between its solder points and the heat sink)

That makes a lot of sense, looking at that re-fried, charcoal-grilled board. But why did some MOSFET connection points burn and others didn't? 

2 hours ago, Boogieman said:

Anyway, if the solder melted, it was way to hot for the config.

That's what all of the conclusions seem to boil down with, one way or another: too much current/heat for that hardware. Until a definitive explanation diagnosis is made, I'm inclined to believe that the weak points that aren't up to the task are the soldering points, motor cable, PCB wiring, MOSFET connectors, insulation (thin silicone pad), board...maybe it's easier just to call it the hardware  

If I owned this wheel I definitely wouldn't push it on any steep hills. But I can't help but wonder...even if one only uses it for street riding...high current spikes from aggressive riding and acceleration, over time, could easily cause premature degradation (and possibly failure?). Which wheels that have failed at overheat hill have had heat-related hardware issues further down the road for users who weren't taxing the wheel as much (all at once)?

Do you guys reckon the Sherman will fare well over time if high load is kept to a "reasonable" level? The question being, what is that reasonable level for the Sherman?

@Marty Backe you're going to have to expand the repertoire and find a close-to-overheat hill, a halfway-to-overheat hill and a-touch-of-overheat hill so wheel stress tolerance can be more accurately classified 

Edited July 16, 2020 by travsformation

[Marty Backe]

11 minutes ago, travsformation said:

That makes a lot of sense, looking at that re-fried, charcoal-grilled board. But why did some MOSFET connection points burn and others didn't? 

That's what all of the conclusions seem to boil down with, one way or another: too much current/heat for that hardware. Until a definitive explanation diagnosis is made, I'm inclined to believe that the weak points that aren't up to the task are the soldering points, motor cable, PCB wiring, MOSFET connectors, insulation (thin silicone pad), board...maybe it's easier just to call it the hardware  

If I owned this wheel I definitely wouldn't push it on any steep hills. But I can't help but wonder...even if one only uses it for street riding...high current spikes from aggressive riding and acceleration, over time, could easily cause premature degradation (and possibly failure?). Which wheels that have failed at overheat hill have had heat-related hardware issues further down the road for users who weren't taxing the wheel as much (all at once)?

Do you guys reckon the Sherman will fare well over time if high load is kept to a "reasonable" level? The question being, what is that reasonable level for the Sherman?

@Marty Backe you're going to have to expand the repertoire and find a close-to-overheat hill, a halfway-to-overheat hill and a-touch-of-overheat hill so wheel stress tolerance can be more accurately classified 

I now know what the motor sounds like under real stress. If it still sounds like then when tackling tough hills, I'm going to stop.

[Patton250]

I feel so bad for this company and this wheel. 

[Violinfun]

On 7/9/2020 at 5:36 PM, meepmeepmayer said:

One of the funniest EUC videos I have seen. You lifting the Sherman into the truck the first time with the unexpected awkward body bend was hilarious

-

Looks like they could easily fix this by extending/modifying the metal tubing along the top of the wheel to offer a grab handle.

Apparently nothing a cherry picker won't resolve 8-)

[MrRobot]

Sooo I have no plans on riding any ridiculously steep hills. 

 

But if the solder points are the points of failure it really reminds me of the whole Xbox 360 debacle. For those that don't know or remember they would get the dreaded red ring of death board failures--but they wouldn't happen or be apparent right away. Some would last a few days some would last a few years. The cause of this was due to the solder points on the board heating up and cooling down over 1000s of cycles to the point where eventually they would fail. 

 

This is what I'm worried about. It working fine for 1000s of miles until one day riding at 40mph and it dies due to the solder points slowly weakening from heating up and cooling down over time. 

 

Ughh I want this wheel so bad but I can't help but think we might have a similar situation on our hands. What do you guys think? 

Edited July 16, 2020 by MrRobot

[Shield]

[Alj]

2 hours ago, Mark Wilson said:

I am not surprised that this will be the solution. The wheel was never designed for slow speed, high torque applications. It's a high speed, comfortable cruiser. I wasn't that disappointing in the wheel not being able to make it up overheat hill, I was disappointing in it self destructing while trying to do so. Assuming that the performance of the while isn't limited until shortly before what would be the failure point, I think limiting the power then is the right choice, given the time constraints to have them mass produced.

Any design has a bottleneck somewhere. Ideally this bottleneck should be just user, but that is not something you can control, so once user is out of question you just want this bottleneck to be in a proper place. From what it seems the bottleneck appear to be in heat dissipation (very bad place) and they are moving it to a software (better place, but seems not ideal). What other choices they have? Lets see:

1. Battery - pretty bad place. If your battery is a bottleneck you lose power to your controller, that will cause wheel cutoff. Unless you run controller from a separate battery, then maybe battery is not so bad bottleneck if you can dissipate heat from it properly.

2. Motor - that is very good bottleneck but it is not stable and varies with RPM. On low RPM motor draws more current because of lower frequency (inductance is not an obstacle). So motor cannot be a managed bottleneck as well.

3. MOSFET.  If mosfet overheats then you get magic smoke. Bad bottleneck. However if you just control amount of power that mosfet delivers from the battery, it becomes maybe the only choice you have. But that will be not MOSFET bottleneck that is actually a ... drumroll....

4. A software bottleneck. Which seems like the only choice you have in this situation. You have to tweak it the way so software is always next bottleneck after user. When you fix heat dissipation issue you may open up software bottleneck a bit but it still has to be a bottleneck because thats the only place you have control of over range of other variables. It does not matter what wheel is designed for, no user action should cause magic smoke.

 

After all those thoughts we can arrive to conclusion that wheel reliability is really how far away you are between software bottleneck and all other bottlenecks.

Wheel performance depends on  enlargement of all possible bottlenecks and bringing of software bottleneck as close as possible to other bottleneck.

When you push for performance way too much you create overlap where you bottleneck migrates away fro your software to.... your heat dissipation, for example and then you have magic smoke as result... or to your battery then you have wheel cut off. Or you cross max current or max temperature of some component and so on.  You now can guess what are the most software updates are for and why they often hurt wheel performance on some area.

[Rehab1]

22 hours ago, Marty Backe said:

Failure Postmortem

Today I disassembled the Sherman so that I could remove the control board and take a closer look at the damage.

Sadly, total disassembly is required to pull the control board. I think Veteran assumed that they would have no control board failures. Considering how easy it is to do all other work on the Sherman, the work required to replace the control board is disappointing.

All the MOSFET's are destroyed, but it looks like the massive amount of current coming in from the batteries literally vaporized that section of the control board. It could be that the MOSFET destruction was secondary to the melt-down that occurred at the battery connections.

All the wiring external to the control board looks great and I see no evidence of heat stress 

 

Marty you are becoming revered throughout the scientific community. . 

 

[Marty Backe]

27 minutes ago, Rehab1 said:

Marty you are becoming revered throughout the scientific community. . 

 

You crack me up Dan, as usual 

[ExplodeMilk]

19 hours ago, Alj said:

From my experience of smelling magic smoke it is usually that overloaded components go either without boom or they just have a small hole out of which "magic smoke" escapes. I havent seen the cases where the whole PCB overheats and separates into layers like that. This can happen by one of the 2 reasons (or combination of 2) 1. PCB wires got too hot due to current.... But usually when this happens PCB wires quickly evaporate and PCB itself stays intact.  2. Inadequate heat sinking. Which is most likely the case here, also, most often PCB goes smoked from hot component on it and not from current flowing thought it, which also makes it most likely the case. Also if heat transfer was adequate the temperature of mosfet would not be much different from the temperature of heatsink and heatsink was not hot enough to melt anything from what it looks like from the pictures.

IMO silicone pad (sil-pad (tm)) used for MOSFET insulation here (between MOSFET and heat sink) while being cheap go-to way of insulating MOSFETs from the heat sink, is inadequate way of insulation/heat transfer in this application because silicone is just "good enough" for transferring the heat but not very good for this particular heat dissipation levels. THe alternative can be ceramic insulation (which is more expensive) or, better, schematic solution that does not require MOSFETs to be insulated because direct mounting might be required for this power levels....

If schematic solution is not possible, for example, because 2 groups of MOSFETs are running in different signal polarity but they can be joined within the same group the solution can be to directly mount 2 groups of mosfets to 2 not connected copper or aluminum plates and then mount those plates on a heat sink using insulation of choice (sil-pad?), in this case area of heat transfer through the silicone will be much higher.  In any case it is not very hard or time consuming to test this setup with thermometer under load and figure out if solution is working properly.

 

there is also heat transfer happening from the silicon to the MOSFET enclosure pad but if that would be the cause of MOSFET burning then MOSFET legs (and PCB) would stay cold and it would be just a small hole in MOSFET with magic smoke.

 

for the reference here's random youtube i found about guy testing different types of insulation and silicone pads is about the worst of them (but the cheapest one of course).

 

Do you think replacing the pad with thermal paste will help the situation?

[Chriull]

1 hour ago, ExplodeMilk said:

Do you think replacing the pad with thermal paste will help the situation?

As  @Alj wrote insulation will be needed, too! Some ceramic/mica plates. And don't forget insulation for the screws.

If this is made in a proper way it will increase the burden the wheel can take, but for sure not make it safe for all "possible burdens"