Xiaomi Pro 2 mosfet and battery upgrade

[Jan-Erik-86]

I have a Xiaomi Pro 2 that i currently run at 30A, watt meter in m365 dashboard app peaking at ~950W when fully charged on stock 36V battery.

I know the controller already have some trace reinforcement on the Pro 2, but I will reinforce the PCB trace a bit more, and i also want to upgrade the mosfets to be able to handle a bit higher current. Also, i see the temp change rather quickly up and down, so i'm thinking of removing the insulating kapton tape they originally use between mosfet and case, and install a thermal pad instead. I found a 1mm pad with 5W/mK at Mouser for an acceptable price, will this be OK, or will i need 0.5mm or less and/or higher W/mK to get good effect (i'm not familiar with pads, but they start getting expensive over 5W it seems)?

As for the mosfets, i'm not sure which ones to get. I'm thinking about either IRFB4110PBF or FDP036N10A based on a Xiaomi M365 mosfet chart i found. Are either of these recommended, or are there better alternatives? I know the 4110 are known good and commonly used fets, but the FDP036N10A appears to be better in nearly everything other then price?

I have a 13s battery that i was thinking of using for a bit higher power and speed without need to use use field weakening. Am i understanding correct that the controller on the M365/Pro/Pro 2 is able to handle a 48/52V battery without any modifications at all? I assume the display gets powerd by the buck converter in the controller and won't need any modifications?
Will it auto-detect the different battery voltage, or do i need to change voltage in firmware for it to work properly?

 

[amberwolf]

I don't know the Xiaomi specifically, but in all other display/controller systems I've seen so far, the display itself runs directly off battery power with it's own LVPS, so that it's MCU is powered on all the time the battery is connected, and it uses an internal transistor to switch battery power to the controller's LVPS (so the controller's FETs have power all the time, too, but the MCU / sensors / etc do not until the display switches that battery power on to it's LVPS.

You can confirm the way yours works by monitoring the voltages on the wiring between display and controller. If one of them has no voltage on it before you turn on the display's power button, but has battery voltage on it afterward, then yours works the same way.

In this design, the regulator used on the display may be less tolerant than that in the controller, so you might need to open it up to find out what it uses and it's specs, and try not to run it near the edge.

 

[Jan-Erik-86]

I see, that makes sense actually. will double check it. Thanks.

From what i understand people run up to 14s on it without modification, but i haven't found any specific tutorial on upgrading from stock 36v to 48/52v. The only modification i see people do is replace main 63v filter cap and isolate the controllers low voltage regulator from the battery terminal and run it on another buck converter if going beyond 60V. Never seen anyone modify the display board.

I also see people hot connect another battery in series with stock even while scooter is on, but I don't know if this might just work equally well as replacing stock 10s for a 13/14s. Personally i think I'd rather just have one BMS to kerp an eye on. Unless of course there is some communication between stock controller and bms that is required for it to work...

Anyway, i think i'll figure that out, but does anyone have any thoughts on the two FETs i was considering?

 

[amberwolf]

The 4110 (if you can get genuine ones from a reputable seller like Digikey, Mouser, Farnell, etc) is well known for many controller upgrades over the years; it's a fairly safe bet, though a bit older technology. A quick search on the forum for the other one shows it has also been used for upgrades/repairs and has a slightly better RDSon.


Regarding "series" connections. If you connect a battery in series with another battery a system won't actually be powered on while you're doing it because you are disconnecting the original battery from the system to put the new one in series with it, and will only be powered back on once you finish the series connection. So you can't actually make a hot connection this way.

You *can* hot-parallel packs with a system on, as long as they are all the same actual voltage at that moment, or you otherwise have protection on each against backflow of current into the lower voltage one(s) of the set. (in case the voltages are significantly different and current flow could be high enough to damage wiring and components--a bit of calculation with ohm's law using the internal resistances of each pack and interconnects and the expected voltage differences can show you if there is a potential for that or not).

Also, if either of the seriesed batteries BMS is not designed for the total pack voltage and any regen/etc that could be created, then as soon as the BMS tries to turn offf it's output to protect the cells from any overlimit condition, the FETs may be damaged by the excessive voltage. Since the most common failure mode is to be stuck on, that can leave the pack still on even though a limit has been exceeded (LVC, etc) and so the pack will continue to operate, potentially being damaged (such as overdischarging cells if it was LVC). Fechter has posts here on the forum about using schottky diodes in parallel with each pack to shunt around a turned-off FET in this instance that should help to prevent such damage, but the best thing is to not use a BMS that can't handle the total seriesed packs' voltage. (or upgrade the FETs to ones that can).

 

[Jan-Erik-86]

Yes, you're right, i couldn't quite figure out how they did it either, since a series connection without all batteries or a jumper is only an open circuit. I thought they had a battery management device that allowed to quickly detect when a 2nd 2-4s battery was connected and switch over to series connection nearly instantly - but it looks like i might have been wrong there, and what they use is known as a "RITA" Adapter that's somehow able to instantly switch from one battery to another with a completely different voltage. Those few who run series only use a simply series connector cable and need battery connected at all times. I think i'll have to look more into what i wish to do as far as this goes.

The every so slightly lower RDS is what mainly made me think of that as a better option. But the Power Dissipation is quite different though, 227 vs 370W is a big difference for something that only have .1mOhm different RDS, btu i'm not familiar enough with FETs to know is this or other factors will really have anything to say? I just know i don't want to risk blowing the FETs by running say maybe ~40A (where's stock isn't recommended higher then 28-30A). I guess i might not need more A if i increase voltage, but i haven't quite figured it all out yet what i want to do.

 

[amberwolf]

The amps the FETs see is different from the amps you get on the battery side. They see Phase amps, directly to / from the motor windings. So they have to handle the actual motor currents, which can be much higher than the battery amps ( up to 1.5-3x battery current, more in some cases).

So, if you want more torque (faster acceleration, etc) out of the motor, you need higher phase currents. (if you want more speed, you need higher voltage, and however much total power it takes to reach that speed under your riding conditions, which may include needing higher phase currents).

If the controller is FOC, then it will have current sensors on at least two phases. If it does, then it won't give you more phase current even with different FETs or voltage, becuase the MCU will limit that the same no matter what hardware is on it, based on it's firmware and settings.




The lower the RDSon, the higher currents the FET can take while creating the same heat. Die bonding wire sizes/etc and the actual FET legs limit the practical currents in any particular package; the spec sheets for a device in a particular package (TO220, etc) may list those. (if not, there is a thread somewhere here from a decade or so ago discussing practical package limits in one of the controller design or "overclocking" threads, probably with Methods, Liveforphysics, Jeremy Harris, or other technical experts posting in it).

Getting the heat out of the FETs fast and keeping them cool lets them handle more current, especially more bursts of current as that is really the practical limitation of most controller upgrades--you don't usually need nearly that much continuous current as you do peak currents. For package-limited controllers (where you're stuck with a specific size/type of FET) you can improve cooling, such as by putting the heatsinking directly to the FETs instead of having them mount to a bar that mounts to a case, etc., or by cutting the heatsinks into six electrically-isolated sections, so each bridge segment has it's own heatsink--doing that allows you to skip the electrical insulators between the FETs and heatsink, which also removes the thermal insulation and lets them cool faster. Doing this means making an insulating mechanical support for the six heatsinks as you don't want the FET legs to have to bear the vibration/etc of those weights.

Other FET limitations are switching times and gate charge. If it take more gate charge to turn it on fully, it will take longer to turn on than another FET with a lower charge in the same controller hardware. If it takes longer to turn on or off then there is more chance for shoot-thru which is when the whole bridge gets turned on at the same time, directly connecting battery positive to battery negative, which is bad.

Power dissipation is usually package limited, but sometimes spec sheets cheat and use different test conditions for their ratings than others do, so you have to also compare the spec sheet test conditions for the specs you wish to compare. If the test conditions are different, the specs can't be directly compared.