ACM now uses 12 MOSFETs instead of 6

[John Eucist]

Other 6 is underneath the visible 6.

Other 6 is underneath the visible 6.

[dmethvin]

The thermal paste looks like it might be something like Arctic Silver rather than the cheap white stuff? That would be a step up too. If that connector is for the motor it looks like it is better than the cheap single-lead barrel connectors that most other EUCs use, it's gold plated for one thing. 

[TremF]

Do you happen to know how long they have been using these new boards for? I got mine last week after it was held for changes the week before - I hope this was it

 

I'll be interested to hear your thoughts on the trolley handle too please.

[OliverH]

55 minutes ago, John Eucist said:

Other 6 is underneath the visible 6.

Other 6 is underneath the visible 6.

Fix or work around?

[John Eucist]

38 minutes ago, dmethvin said:

The thermal paste looks like it might be something like Arctic Silver rather than the cheap white stuff? That would be a step up too. If that connector is for the motor it looks like it is better than the cheap single-lead barrel connectors that most other EUCs use, it's gold plated for one thing. 

Not sure about the thermal paste.  I also have no idea what that connector is for but boss Lin said something in Mandarin that I didn't quite understand but emphasized how good that connector is like he was very proud of it. Sorry I didn't investigate more but at least you guys have this photo.  

21 minutes ago, TremF said:

Do you happen to know how long they have been using these new boards for? I got mine last week after it was held for changes the week before - I hope this was it

 

I'll be interested to hear your thoughts on the trolley handle too please.

I'm not too sure about how long they've been using these boards for either but I suspect a couple of weeks.  Will update you on the trolley when I have more chance to use it.

[TremF]

3 minutes ago, John Eucist said:

I'm not too sure about how long they've been using these boards for either but I suspect a couple of weeks.  Will update you on the trolley when I have more chance to use it.

 

Brilliant! Cheers. If it's been a couple of weeks then mine will have the new board. Very happy  

It will be interesting to know how well the trolley copes in a crash too. I don't want anyone to crash but sometimes things happen out of our control - like mine flipping on me over the weekend.

[John Eucist]

54 minutes ago, OliverH said:

Fix or work around?

I didn't follow the ACM problems, if any, prior to getting mine because I was simply waiting for them to make a trolley first.  I was told the extra mosfets would better handle any current spikes. 

[TremF]

32 minutes ago, OliverH said:

Fix or work around?

As far as I am aware the ACM was having a problem where, if you travelled slow then hit a bump/pot hole and the ACM needed more power to get over, the board couldn't cope with the current and was frying so was modified.

 

I could be wrong but this is my guess from what was being reported before the fix. I was told I was getting an updated version but not the specifics - here we go.

[OliverH]

How does a EV Car motor driver look like to handle 650/ 700A?

[Chriull]

like https://endless-sphere.com/forums/viewtopic.php?f=30&t=65691 

[Hunka Hunka Burning Love]

1 hour ago, TremF said:

 

Brilliant! Cheers. If it's been a couple of weeks then mine will have the new board. Very happy  

It will be interesting to know how well the trolley copes in a crash too. I don't want anyone to crash but sometimes things happen out of our control - like mine flipping on me over the weekend.

What happened to cause the ACM to flip you over? 

[TremF]

30 minutes ago, HunkaHunkaBurningLove said:

What happened to cause the ACM to flip you over? 

It didn't flip me over. I got on the ACM outside a shop then as I leaned to set off (slight lean to set off slowly) it suddenly flipped from under me as if the wheel itself didn't move but the body spun maybe.

I remained standing and somehow the ACM caught my right shin as it flipped. When I looked at it on the ground I saw smoke and thought the board had fried but it still works perfect so not sure if smoke was from the ground as the pedal hit it maybe or from a stone stuck inside.

I really don't know anything other than I was stood on it, it flipped on me and I saw smoke but it works perfect still with no issue since.

[esaj]

3 hours ago, dmethvin said:

If that connector is for the motor it looks like it is better than the cheap single-lead barrel connectors that most other EUCs use, it's gold plated for one thing. 

 

2 hours ago, John Eucist said:

I also have no idea what that connector is for but boss Lin said something in Mandarin that I didn't quite understand but emphasized how good that connector is like he was very proud of it. Sorry I didn't investigate more but at least you guys have this photo.  

The connector's MT60, kind of like a 3-pin version of XT60:

http://www.aliexpress.com/item/MT60-3-5mm-Motor-Plug-Connector-Set-for-RC-lipo-battery-FPV-Multicopter/32686641236.html

Just ordered a couple of those few days back

[HEC]

11 minutes ago, esaj said:

 

The connector's MT60, kind of like a 3-pin version of XT60:

http://www.aliexpress.com/item/MT60-3-5mm-Motor-Plug-Connector-Set-for-RC-lipo-battery-FPV-Multicopter/32686641236.html

Just ordered a couple of those few days back 

Yeah I have bunch of those here as I use those to connect motors to ESCs on multirotors ("drones"). They're clearly used for mottor connection in the picture. I prefer those on large models to individual bullet plugs despite higher weight and larger size a it's much faster and easier to swap faulty motor or ESC in field and also avoids possibilit of swapping the phases around by accident.

[Keith]

8 minutes ago, esaj said:

The connector's MT60, kind of like a 3-pin version of XT60:

http://www.aliexpress.com/item/MT60-3-5mm-Motor-Plug-Connector-Set-for-RC-lipo-battery-FPV-Multicopter/32686641236.html

@esaj thank you for finding those, when I eventually have a puncture or otherwise need to drop the wheel out my EUC, I think I'll fit one of those rather an trying to get the wires off of the control board. I might just be tempted to put heat shrink over it to ensure it cannot come apart in use.

15 minutes ago, esaj said:

Just ordered a couple of those few days back

Don't hold your breath - "delivery 30-50 days"!

P.S. I see you are still singlehandedly keeping Aliexpress in profit :-)

[esaj]

1 minute ago, Keith said:

@esaj thank you for finding those, when I eventually have a puncture or otherwise need to drop the wheel out my EUC, I think I'll fit one of those rather an trying to get the wires off of the control board. I might just be tempted to put heat shrink over it to ensure it cannot come apart in use.

Don't hold your breath - "delivery 30-50 days"!

Well, it's pretty normal that the deliveries take between 1 and 8 weeks to here, no matter what the page says   Of course I'm skimpy and always use the free shipping option  

1 minute ago, Keith said:

P.S. I see you are still singlehandedly keeping Aliexpress in profit :-)

Ahem, my order count has just broken 400 this week...   Whoopsie again

Total reward points: 1871

A3 level member

Your level is higher than 99% of other users.

 

 

 

Even though I only order the "cheap stuff" and always with free shipping, it all starts to add up...   But still, the price difference for basic components is so huge compared to here, that I've probably "saved" a four figure sum by now.

[HEC]

5 minutes ago, Keith said:

@esaj thank you for finding those, when I eventually have a puncture or otherwise need to drop the wheel out my EUC, I think I'll fit one of those rather an trying to get the wires off of the control board. I might just be tempted to put heat shrink over it to ensure it cannot come apart in use.

They don't come apart for sure. They're quite stiff and as being used in RC flying models proved fairly secure. You can though always use small cable tie or two winds of electrical tape for the peace of mind but I'd not recommended encasing those into heatshrink tubes.

[Hunka Hunka Burning Love]

Don't forget to get some Ebates cashback to your Paypal account when buying stuff from AliExpress!  My total cashback from eBay, Walmart, etc is about $272 over the past year or so.

BTW is there less chance of MOSFET burnout if 12 are used instead of 6?  Does that divide up the heat dissipation so it's halved for each MOSFET?

[esaj]

10 minutes ago, HunkaHunkaBurningLove said:

Don't forget to get some Ebates cashback to your Paypal account when buying stuff from AliExpress!  My total cashback from eBay, Walmart, etc is about $272 over the past year or so.

I stopped reading the terms when it said " Member must have a valid address in the United States or Canada only. "  Plus, I don't pay with Paypal or credit card...

 

10 minutes ago, HunkaHunkaBurningLove said:

BTW is there less chance of MOSFET burnout if 12 are used instead of 6?  Does that divide up the heat dissipation so it's halved for each MOSFET?

At least in theory, it should halve the heat dissipation per mosfet (as 2 mosfets "share the load"). In practice it should sort-of "auto-compensate" for slight differences in the mosfets, if the mosfets are mounted near enough each other to stay around the same temperature, if they're unequally loaded, the internal resistance should go up with temperature on the one that's taking a bigger share of the load, thus reducing it and passing more of it to the other one. That's what I've at least read about, one of my (unfinished) projects is an electronic load based on paralleled mosfets, but it's not exactly the same principle as with the wheels (for load usage, the mosfets are deliberately made to dissipate a lot of power, whereas in the motor drive you want them to dissipate as little power as possible). Still the principles hold, at least for the most part

[OliverH]

4 hours ago, Chriull said:

like https://endless-sphere.com/forums/viewtopic.php?f=30&t=65691 

For one motor or for all motors? So how much per phase? And how much current does the Mosfet handle in a Tesla?

[esaj]

21 minutes ago, OliverH said:

For one motor or for all motors? So how much per phase? And how much current does the Mosfet handle in a Tesla?

Teslas' don't use mosfets but IGBTs:

 

Fig.1: 2×14 IGBTs in parallel is one leg of inverter—All packaged in discrete TO247 – From Tesla Roadster and Model S inverter

http://www.pointthepower.com/on-tesla-electric-vehicles-semiconductor-packaging/

Also, I recall reading somewhere that the motors use something like 400V for voltage... thus the inverter? But not that familiar with the Tesla technology or IGBTs/inverters in general

EDIT: Oh, right, about the current:

• “That design allows Tesla to ‘insanely’ run 1500A through the inverter IGBTs and then the motor”

 

[Hunka Hunka Burning Love]

2 hours ago, esaj said:

I stopped reading the terms when it said " Member must have a valid address in the United States or Canada only. "  Plus, I don't pay with Paypal or credit card...

At least in theory, it should halve the heat dissipation per mosfet (as 2 mosfets "share the load"). In practice it should sort-of "auto-compensate" for slight differences in the mosfets, if the mosfets are mounted near enough each other to stay around the same temperature, if they're unequally loaded, the internal resistance should go up with temperature on the one that's taking a bigger share of the load, thus reducing it and passing more of it to the other one. That's what I've at least read about, one of my (unfinished) projects is an electronic load based on paralleled mosfets, but it's not exactly the same principle as with the wheels (for load usage, the mosfets are deliberately made to dissipate a lot of power, whereas in the motor drive you want them to dissipate as little power as possible). Still the principles hold, at least for the most part

Aw that's too bad.  I didn't think there was a location limitation on the Ebates site.  I usually use the .ca one, but the .com one has better cashback rates so I've switched to that instead.  I'm not sure if they check the address as I have mine set on sending the cashback to my Paypal account.  I could have registered with any address in Canada or the US I think.  They do send cheques as an option so you have to set it to offer Paypal deposits.  In any case, it's still a handy way to make some money back from purchases I find.  I spend so much that every bit adds up.

So is it mainly a temperature issue with these MOSFETs burning up or are there problems with voltages or current passed onto them going too high and ending up frying them?  I would think if the voltage/current is kept in spec, and if the heatsink is appropriately sized, there shouldn't be a problem?  But why do we see so many reports of control boards getting fried?  What's the common root cause?

[esaj]

1 hour ago, HunkaHunkaBurningLove said:

So is it mainly a temperature issue with these MOSFETs burning up or are there problems with voltages or current passed onto them going too high and ending up frying them?  I would think if the voltage/current is kept in spec, and if the heatsink is appropriately sized, there shouldn't be a problem?  But why do we see so many reports of control boards getting fried?  What's the common root cause?

As usual, this is pretty much guesswork and combining the puzzle pieces I've picked up when reading about this stuff, so take it with bagful of salt:

 Don't know about the wheels exactly, but based on what I've read and still remember, there are multiple ways to destroy a mosfet (these are probably not even all of them):

 

Someone mentioned that the "easiest" way to destroy a mosfet is applying too much voltage (difference) at the gate (going beyond the absolute maximum Vgs). The gate is what is used to control the mosfet (not conducting / "partially conducting" ie. linear region / fully conducting ie. active region). As the above picture shows, there's a layer of metal oxide between the gate and the channel that that the actual current flows through the mosfet from drain to source (for N-channel mosfet, or source to drain for P-channel). A high enough voltage difference between the source and the gate can cause a strike-through from the gate to the channel (or vice versa), shattering the microscopically thin insulator layer (that picture's not to scale). Yet for the mosfet to conduct, there must be enough voltage difference between the gate and the source...

I don't know how the actual gate drive chips in the wheels work, or if they will under any circumstance allow the gate voltage to go "out of bounds". Another possibility (although I think unlikely) could be careless handling during assembly, an electro-static discharge (ESD) from a finger could damage the insulator. Short-lived static discharges are "evil" in the sense that in case of many (active) components, they can cause internal damage that's not obvious immediately. The device can keep operating normally for a long time before finally failing:

 

Also exceeding the maximum voltage between the drain and the source can damage the mosfet, although I'm not familiar with the actual effect it has (maybe something to do with the channel structure?)

However, my guess would still be that the mosfets simply overheat under high stress situations. "Too much current for long enough time" (which, depending on current could be from very small fractions of a second to minutes) will overheat the mosfet. The are also (at least) two types of losses that occur in the mosfet: conduction losses (the "normal" voltage drop that occurs when the mosfet is fully conducting) and the "switching losses", that are due to the mosfet passing through the "linear/ohmic" region, where it actually acts more like a resistor, more "ohmic" resistance = more voltage drop over the mosfet = more power dissipation.

If the gate voltage hasn't reached high enough value (or for some reason doesn't at all), it won't be in full conduction (active region / saturation), and as long as it isn't in the active region, it's actually dropping more voltage, thus causing more power dissipation in the mosfet itself and it heats up. When the mosfet gate is being pulsed with the PWM-signal, it actually opens (starts to conduct) and closes (stops conducting) as the signal goes up and down. Each time it does that, it takes some time for the voltage to raise and drop (due to gate charge, kind of like a capacitor, it needs to be "charged and discharged" for the voltage to change), and the mosfet "travels" through the linear region:

On the latter picture in the upper graph it can be seen that while the Vds (voltage difference between drain and source) is going down (the mosfet is beginning to conduct), the Ids (current flowing from drain to source) is already starting to raise. The lower graph shows the power dissipation (losses) occurring in the mosfet, and they spike up during the transition period from off-to-on and vice versa. My guess is that that's the reason why especially high-power wheels tend to make that high-pitched noise, probably they needed to drop the PWM frequency, so the transitions aren't done as often, and the frequency is now in the audible frequency range (on the wheels you can't hear it, it's probably still there, but so high frequency that you can't hear it?).

The situation is even "worse" when the current is flowing in reverse through the body-diodes, as those have a high(ish) voltage drop, something like 0.5-1V, and it can even go up with higher currents. Unlike when the mosfet is "normally conducting", the power dissipation won't drop at any point during which the diode is conducting. If the braking isn't done in a manner that they can open the mosfets to conduct "normally" (I believe I've read that a mosfet can allow current to flow even in reverse through the channel, as long as the gate-voltage is sufficiently different from source), the regeneration happens through the body diodes, and that could at least explain why some wheels seem more prone to blow them during power braking.

All in all, I doubt the gate voltage actually would be the problem, more likely it's simply a matter of overheating.

 

[Linnea Veteran]

Everytime when I saw esaj's post I will be reminded that I am just a 3 years old child. Too professional

[John Eucist]

Photo from @Jane Mo