Henrik Olsen Posted May 3, 2016 Share Posted May 3, 2016 We have in our FB group seen quite a few burned mosfets over time, due to stressed EUC's like jumping, hitting curb, etc... I assume this is created by voltage or current transient from temporarily blocking the wheel that makes the mosfet fry? Does anybody have a circuit diagram over the output stage of a EUC...? Link to comment Share on other sites More sharing options...
Chriull Posted May 3, 2016 Share Posted May 3, 2016 You have an "generic EUC" schematics in https://github.com/generic-electric-unicycle/electronics/tree/master/generic_electric_unicycle (i assume - i have no prog to view the layout/schematics files on my computer... ) The driver part are just 3 Half Bridges: Link to comment Share on other sites More sharing options...
Henrik Olsen Posted May 3, 2016 Author Share Posted May 3, 2016 Do we know what mosfet type is normally used on high end wheels like 9B? Link to comment Share on other sites More sharing options...
Chriull Posted May 3, 2016 Share Posted May 3, 2016 8 minutes ago, Henrik Olsen said: Do we know what mosfet type is normally used on high end wheels like 9B? Ninbot uses 12 100N8F6 or 75NF75 (http://forum.electricunicycle.org/topic/1817-ninebot-one-e-broken-after-1-12-weeks-any-troubleshooters-here/?do=findComment&comment=20371) Gotways have 6 B3207 (http://forum.electricunicycle.org/topic/3711-ideas-on-how-to-resolve-kingsong-14-overheating/?do=findComment&comment=38302) Link to comment Share on other sites More sharing options...
esaj Posted May 3, 2016 Share Posted May 3, 2016 Firewheels have IRF7759's: http://www.irf.com/product-info/datasheets/data/irf7759l2pbf.pdf I think someone posted a picture of IPS Zero mainboard showin 75NF75's too. EDIT: forever nitpicking: The driver part are just 3 Half Bridges: Wrong schematic, the wheels don't use P-channel mosfets on high-sides, but instead drivers / voltage pumps & N-channels? Otherwise all the 6 (or 12) mosfets wouldn't be the same model? Link to comment Share on other sites More sharing options...
Henrik Olsen Posted May 3, 2016 Author Share Posted May 3, 2016 @esaj could you please elaborate voltage pumps & N-channels? I know the difference between N and P channel I would like to understand this... and what might cause the mosfets to fry.. Link to comment Share on other sites More sharing options...
esaj Posted May 3, 2016 Share Posted May 3, 2016 26 minutes ago, Henrik Olsen said: @esaj could you please elaborate voltage pumps & N-channels? I know the difference between N and P channel I would like to understand this... and what might cause the mosfets to fry.. As usual, I start off by saying I'm not that good with electronics , but, from what I've seen so far, the half-bridges on the wheel mainboards are built using 6 (or 12 in some cases?) similar mosfets (N-channels). To get an N-channel (power) mosfet into "fully conducting"/saturated-state ("active region", I believe), the gate voltage must be fairly above the voltage at the source (ie. Vgs > Vgs(th)). A mosfet will start conducting when the Vgs = Vgs(th) (Gate voltage threshold, typically 2...4V, drops as the mosfet heats up), but it's still in the "linear" (non-fully-conducting) -region, so it will actually partially still "resist", causing voltage-drop across the drain and the source and heating up the mosfet. So to be on the "safe" side, the voltage difference between the gate and the source should be a good 10V or so (depending on the mosfet characteristics) to make sure it's in the "active"-region. Another topic is the gate charge, which will affect how fast the gate voltage changes, but I won't get to it here. If the motor were running at full speed, there'd be a back-EMF that can be (nearly) as high as the battery voltage, and if the high-side mosfet gate would just use battery voltage, it could not go high enough above the source to reach that fully conducting region. So, the voltage has to be somehow made higher than what the battery can provide. I believe most wheels use "real" mosfet (gate) driver-ICs, that is, integrated circuit chips that have the necessary "gate drive"-electronics inside and can supply/sink ample current to make the gate voltage go up and down fast. I haven't researched them much, just know that they exist Another way is to build a "charge/voltage pump" -circuit that charges a capacitor and then discharges it to the gate, driving the voltage (temporarily) higher. That's what I used when I built my (first) 3-phase motor driver using discrete components & Arduino, it was mostly based on the half-bridges here: ) Two pretty sure-fire ways to kill a mosfet I know are exceeding the gate voltage limits (it's usually something like source-voltage +-20V for power mosfets) or dissipating too much heat in it without enough cooling. If the wheels use "proper" gate driver-ICs, my best guess is that insufficient cooling is more probably the reason for burnt mosfets in the wheels, and sometimes maybe a bridge shoot-through (ie. both the upper- and lower-side are conducting at the same time, and the current runs straight through the mosfets to ground, not through the motor, overheating them in an instant). The momentary power during braking can also get very high for a brief period, which could potentially overheat the mosfets... Even in full conduction, there's some (fairly small, but going up somewhat with temperature) ohmic resistance in the mosfets themselves, so they always dissipate some heat (the more current you drive through the mosfet, the more voltage it drops, and thus the dissipated power goes up with current). But, I'm just an amateur and "best guessing" here. Some real electronics/motor drive engineer can probably point out flaws in my logic and/or possible misunderstandings about mosfets and their characteristics Link to comment Share on other sites More sharing options...
Henrik Olsen Posted May 3, 2016 Author Share Posted May 3, 2016 @esaj Thank you for taking the time to explain... I will run this by one of the electronics geeks here at the office for his opinion... Link to comment Share on other sites More sharing options...
Chriull Posted May 4, 2016 Share Posted May 4, 2016 Edit: deleted this paragraphs -was nonsense Quote Two pretty sure-fire ways to kill a mosfet I know are exceeding the gate voltage limits (it's usually something like source-voltage +-20V for power mosfets) or dissipating too much heat in it without enough cooling. If the wheels use "proper" gate driver-ICs, my best guess is that insufficient cooling is more probably the reason for burnt mosfets in the wheels, Maximum current i could achieve with my ninebot one e+ (very shortly) were about 48A! *) With an resistance of about 9,5mOhm/2*1,5~7 mOhm (/2 because there are two in parallel and times 1,5 for the increases resistance at higher temperatures). With P=R*I²=7mOhm*48*48=16W in one Mosfet (edit: pair) - for driving the motor a second mosfet (edit: pair) has to be active. So ~32W could be "burned" with this "small" 500W wheel. Just for a short period of time - but if happening frequently while e.g. going steeply uphill this could exceed the heatsink capabilities easily... Quote and sometimes maybe a bridge shoot-through (ie. both the upper- and lower-side are conducting at the same time, and the current runs straight through the mosfets to ground, not through the motor, overheating them in an instant). The momentary power during braking can also get very high for a brief period, which could potentially overheat the mosfets... While (regenerative) braking these high currents could go through the invers body diodes which have a forward voltage of a little bit above 1V - these result in real high numbers for power dissipation in a single mosfet. This can/should be avoided by the firmware which hopefully switches this one mosfet on, so that just the low on resistance of the mosfet dissipates power. If not implemented right/neat in the firmware this also could be a real problem. I fear that these examples are just a very incomplete enumeration of the ways how mosfets could be killed whith an EUC... To find the real cause(s), which mainly distribute to dead mosfets, quite some empirical measurements would have to be taken - or someone asked with profound experience in this field... Edit: *) This value was delivered from the ninebot itself - so no real idea how accurate this number is, but on the other side the values are "only" sent ~30 times a second! So the real peak (inbetween) the sample points could be even higher... Maybe something for @Jason McNeil- you once posted logs from a logging hardware (Eagle Tree Systems?) installed once in a ninebot? If you still have this with a ninebot you could once compare this with a parellel log by 9BMetrics (http://forum.electricunicycle.org/topic/3589-9b-metrics/?do=findComment&comment=39334) - if you have an iphone and are willing to install that logging system again in a ninebot... So the accuracy of the values sent by ninebot could be measured - and maybe used as calibration factor for 9BMetrics? Link to comment Share on other sites More sharing options...
esaj Posted May 4, 2016 Share Posted May 4, 2016 8 hours ago, Chriull said: Maximum current i could achieve with my ninebot one e+ (very shortly) were about 48A! *) With an resistance of about 9,5mOhm/2*1,5~7 mOhm (/2 because there are two in parallel and times 1,5 for the increases resistance at higher temperatures). With P=R*I²=7mOhm*48*48=16W in one Mosfet (edit: pair) - for driving the motor a second mosfet (edit: pair) has to be active. So ~32W could be "burned" with this "small" 500W wheel. Just for a short period of time - but if happening frequently while e.g. going steeply uphill this could exceed the heatsink capabilities easily... Umm, aren't the mosfets in series (high-side in one half bridge, low-side in another half bridge, two motor-phases in-between)? Link to comment Share on other sites More sharing options...
Chriull Posted May 4, 2016 Share Posted May 4, 2016 17 minutes ago, esaj said: Umm, aren't the mosfets in series (high-side in one half bridge, low-side in another half bridge, two motor-phases in-between)? This was for the ninebot with its 12 mosfets (if all 12 are mosfets...) Link to comment Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.