S22 Pro downgraded with 125V mosfets

[mrelwood]

During WrongWay’s S22 Pro tear down, he found that the wheel now uses the familiar HY5012 mosfets (MSX etc), which has a break down voltage of 125V. But the wheel should charge up to 126.4V, and peaks during braking will even go a good bit higher than that.

WrongWay’s teardown, time stamped:

 

HY5012 datasheet:

https://datasheetspdf.com/PDF/1269423/HOOYI/HY5012W/1

 

Inmotion got an immeasurable amount of filth as an aftermath to the V12 board failures that they themselves  claimed to be caused by a bad batch of mosfets. They used 100V mosfets on a 100.8V V12. The whole saga left a large dent in people’s trust on the brand.

If this single voltage rating is indeed as black and white as it was claimed to be for the V12, shouldn’t we then expect even more board failures on the S22 Pro? And the same amount of filth directed at KS?

If I would’ve been vocal at blaming Inmotion of lousy design during the V12 cut-out after math, I would definitely rush to be vocal at blaming KS even before a peak in the cutout charts even start to rise.

 Though in reality, I’m still not completely sure that Inmotion and KingSong both really fail so embarrassedly in such a basic aspect of crucial component selection.

[mrelwood]

Let it be said that while I am extremely far from even knowledgeable when it comes to BLDC motors’ and their drivers’ electronics, to me it still seems that a single mosfet (or a parallel pair) never sees even half of the battery’s voltage. Though a bit more than that during regenerative braking.

Unless I’m gravely mistaken, at any given phase peak the current flows through a mosfet, the motor, and then another mosfet. For example SW1 -> Motor’s U-V coil -> SW4. The current is the same throughout any series circuit, but the voltage is shared between it’s members. And since the resistance of these two mosfets (SW1 and SW4) is expected to be roughly the same at any given time, they would both see roughly the same voltage, which must then be less than half of the total voltage.

 Please explain where I’m wrong! I want to learn.

[Chriull]

1 hour ago, mrelwood said:

Please explain where I’m wrong! I want to learn

Sw1 and sw2, sw3 and sw4 as sw5 and sw6 are never open at the same time. So if one of the half bridge mosfets is open the other one is closed and has to withstand the whole battery voltage minus the voltage drop off the open one. The drop is about 10mOhm x 100A = 1V. Maybe plus some eventual overshots produced from the wires and coils?

Between two closed mosfets the divided voltage could be in quite any range too.

During regenerative braking some of the lower mosfets are pwm'd. The upper act just as rectifier. So the voltage generated in the coils is transformed above battery level. The upper mosfet of this coil conducts "in reverse" through his body diode. The other upper mosfet were the coil is held on ground has to withstand the whole voltage difference again.

[Wolverine]

2 hours ago, mrelwood said:

to me it still seems that a single mosfet (or a parallel pair) never sees even half of the battery’s voltage.

I recommend to watch this video to get better understanding of three-phase BLDC motor. Every single MOSFET in that system must withstand the power supply voltage completely alone. This is due to the way the three-phase BLDC is designed to work. However, the power supply is not the only source that creates voltage in this system, there are other situations (regenerative braking etc.), but it is not worth going there now.
I don't understand how KingSong and Inmotion don't know basic things about electronic components.
So, how to choose a MOSFET? Choose a MOSFET whose drain-to-source voltage rating (VDS) is at least 20% higher than the supply voltage. In some cases - especially in systems with large torque steps, large currents, and poorly controlled power supplies - you may need a margin as much as twice the supply voltage. That´s it.

 

 

[mrelwood]

Oh right, I was so stuck at thinking about the travel of electrons, I didn’t think about the mosfets that are closed. Duh! Thanks guys.

Inmotion V12: 100/100.8 = 99.2%.

S22 Pro: 125/126.6 = 98.7%.

Yet something tells me that KingSongs move with the Pro won’t create nearly as much outcry than the V12 did.

[rS22]

On 7/10/2023 at 11:05 AM, mrelwood said:

Oh right, I was so stuck at thinking about the travel of electrons, I didn’t think about the mosfets that are closed. Duh! Thanks guys.

Inmotion V12: 100/100.8 = 99.2%.

S22 Pro: 125/126.6 = 98.7%.

Yet something tells me that KingSongs move with the Pro won’t create nearly as much outcry than the V12 did.

So, has anyone discussed the merits (and risks) of doing a component-level replacement of the mosfets with something more robust? Is there a mosfet in the same form factor that would be a good replacement? If you create more headroom in the mosfets, would it be smart to do the same somewhere else in the system?

When I think about a question like that, I usually wonder if the OEM just made a cost decision, or if there’s some other consideration, like it would just shift the failure point somewhere else. But since these FETs are critical to the function and safety of the wheel, I doubt the latter applies.

Also, I usually wonder if the hypothetical mod carries some risk, like you’ll brick the unit or ruin the PCB by doing it poorly. This would be a pretty straightforward mod with basic soldering skills. So that doesn’t really apply.

I waited until this Spring to buy the S22 Pro, and one of the marketing bits that helped me pull the trigger was “upgraded mosfets.” Was that just fiction?

[chanman]

It's not just the pro, the initial couple batch of S22's shipped with the JMSH1504AS with the 150V ratings, but apparently those became difficult to source during the pandemic and so they switched to the HY5012's soon after that. My board from the factory came with the JMSH but a replacement board from EWheels has the 5012's. According to Jason he hasn't seen a significant difference in failure rates between the JMSH boards and the HY5012 boards, so I don't know what to make of that. Possibly the 40T cells in his units have smaller regen voltage spikes? Or the motor issues with the first couple of batches clouded the data so we can't make a good comparison.

I did find it interesting that when I put the HY5012 board in, the original charger I had refused to charge it, and I got sent a new charger that worked but only charges to 124.5V (BMS reported). I have to wonder if this is intentional trying to mitigate the 125V mosfets (with a razor thin margin, but still) or just coincidence. One could also limit the charging using the app to 90% or so to give more margin, and rely on the smart BMS to check on the balance health.

1000 miles with this board charging to 100% with no problems so far. I did a spin test with it V12 style when I got it, but not sure if it was at 100% then, might have to try again.

Edited July 16, 2023 by chanman

[Milordas]

imo 125v mos does not mean will burn even on 150v, and 126V of the wheel you can have max for 5 min after removing from charger, as battery discharge curvature shows that recharge rate from top charged battery is a lot faster then in the middle of voltage

[Chriull]

15 hours ago, Milordas said:

imo 125v mos does not mean will burn even on 150v, and 126V of the wheel you can have max for 5 min after removing from charger, as battery discharge curvature shows that recharge rate from top charged battery is a lot faster then in the middle of voltage

How do you come to your opinion?

DS Breakdown voltage has no burden relation - it happens instandly once the specific value of the soecific mosfet is exceeded. The only thing why such mosfets survive often/mostly is because it's just a minimum value. Typical and max are above, but not specified. Also breakdown voltage value increases with higher temperature - which happens under normal operation. Just in cold climate (below 25°C) at startup breakdown voltage can be below such specified minimum values.