Another Wheel (ACM2) Succumbs To Overheat-Hill

[meepmeepmayer]

When I'm saying power, I mean the power usage of the wheel at every moment, not the motor nominal or peak power. This power usage is what heats the components up, and we would just have to know the nominal power of the wheel - the maximal possible power usage of the entire machine for an unlimited time that does not lead to a heat overrun (of course that's where the cooling system comes in). Which, for a fixed rider weight (and speed), is more or less proportional to the incline % a wheel can do without overheating or frying (because most power usage is lifting the weight, and that's proportional to the height and incline %).

Torque does not matter, for a bigger tire or different motor build, you may need more (or less) power to get the same torque. So it's power again, really the only thing that counts. And what I'm always proposing, overengineer everything so the wheel's power usage is limited by the battery. Easy to monitor, easy to implement conservative limits if wanted, etc.

You're right though, torque is important to judge what a wheel can do and how it will feel to ride (and incline % really isn't as it is rider dependent), and manufacturers should just give some nominal torque for the wheel with the specs, along with a nominal wheel power usage (who cares about the motor's power numbers if it isn't even the weakest link, the mosfets and cables seem to be). Problem is, for a fixed power budget, torque depends on the speed so you would have to state it dependent on the speed, so it gets confusing again (like inclines and ranges).

I'm no physicist (really wish we had one to clear up this stuff) and might be wrong, but in theory, everything should be very simple. Give nominal (forever) power numbers, along with 5min, 1min, 30s, 10s (for example) peak power numbers that don't lead to hardware problems. Everything else should follow from that.

[Jerome]

14 hours ago, meepmeepmayer said:

When I'm saying power, I mean the power usage of the wheel at every moment, not the motor nominal or peak power. This power usage is what heats the components up, and we would just have to know the nominal power of the wheel - the maximal possible power usage of the entire machine for an unlimited time that does not lead to a heat overrun (of course that's where the cooling system comes in). Which, for a fixed rider weight (and speed), is more or less proportional to the incline % a wheel can do without overheating or frying (because most power usage is lifting the weight, and that's proportional to the height and incline %).

Torque does not matter, for a bigger tire or different motor build, you may need more (or less) power to get the same torque. So it's power again, really the only thing that counts. And what I'm always proposing, overengineer everything so the wheel's power usage is limited by the battery. Easy to monitor, easy to implement conservative limits if wanted, etc.

You're right though, torque is important to judge what a wheel can do and how it will feel to ride (and incline % really isn't as it is rider dependent), and manufacturers should just give some nominal torque for the wheel with the specs, along with a nominal wheel power usage (who cares about the motor's power numbers if it isn't even the weakest link, the mosfets and cables seem to be). Problem is, for a fixed power budget, torque depends on the speed so you would have to state it dependent on the speed, so it gets confusing again (like inclines and ranges).

I'm no physicist (really wish we had one to clear up this stuff) and might be wrong, but in theory, everything should be very simple. Give nominal (forever) power numbers, along with 5min, 1min, 30s, 10s (for example) peak power numbers that don't lead to hardware problems. Everything else should follow from that.

I get what your after now. You are advocating that the components should be slected such that the maximum figures of performance can be handle with room to spare.

The battery can provide virtually it's entire capacity and would in a dead short situation. It is therefore the control circuits and wire's and connector's power handling capability that determine smoke or no smoke. The EUC Makers "should" over-engineer all critical components but that would cost more if they wanted to maintain the same performance specs., or they would have to lower the performance which has marketing ramifications. Gotway is putting in bigger mosfets and wire-connectors, but their prices are rising. The others are raising their performance, while trying to keep the safety features in place and their prices are rising. What will the enthusiast pay?

You are right in that we should not have to figure out a wheels limitations by trial and error. The makers could build calculators into the firmware and apps that could take inputs like rider's weight, gradient, and length and report whether said hill is within the capability of the wheel and other scenarios. This is just programming.

Like most consumers we want it all for a low price. Our need for speed in the USA particularly is the source of wheel disappointment. Selecting/designing motors built for maximum torque would solve all hill climbing problems, even with 220 kg riders. The problem is the wheels would have top speeds of 30 kph or less.   Harley Davidson Mc for years have had HP only slightly above some of the large gas scooters and medium size bikes. They have had, however, a lot of torque! I use to dog them on the road with my Honda Silverwing scooter until we got to a steep long grade. They could not only maintain their speed, but accelerate. I could not and if I tried I would the temperature rise on my water-cooled motor! If I live in a flat area I want speed, but if my riding terrain includes steep or many long climbs, I want torque. One day I will learn to make a short reply .. or not.

[meepmeepmayer]

@Jerome Exactly! Very well said

I'm no expert by any means, and it's probably harder to build good electronics than just using big enough components. Motor overheating also seems to be a possible serious obstacle. But having the battery as the primary limiting component both gives EUC their maximum ability (as battery tech seems to be a limit of how good EUCs can be at the time, not motor tech or mosfets or cables or whatever) and seems easy to implement (the BMS essentially does the work for you), compared to limiting some other wheel parameters (which still should be done for safety and longevity, like temperature sensors, current limits and whatnot).

I just know, one look at the 1300Wh 6P battery of my ACM (can easily put out 60A permanently), one look at the original (older style) motor cabling (16AWG = ~20A permanently), and I know Gotway either didn't do basic considerations or didn't care about how they build their wheel electronics. That pisses you off after you fried one. All I want is to know a wheel won't suddenly fry without warning. And good performance also can't hurt

Torque is unrelated to that in the narrower sense, but I agree with your thoughts on that too, I would take a high torque wheel over a high speed one. Though e.g. my ACM's torque is fine if I knew I couldn't kill it by using that for too long.

[esaj]

On 7/26/2018 at 1:00 AM, Mimir said:

so one interesting thing about this failure is the mode...

a MOSFET was missing a *lead*.

this doesn't sound like a standard short-circuit / overcurrent failure. it might be an explosive decap that caused it- is there a picture of the damaged MOSFET- but...

i can't find exact details for the packages of the FETs used on the ACM2 control board, but they're HY3712s, in what looks like some variant of a TO-220 package.

From the HY3712 datasheet, we find this rather concerning chart- remember, a HY3712 is allegedly rated for 170A...

 

 

The semiconductor may pass 170A.

But the package is only rated for 75A...

and this sort of failure, a missing lead, is exactly what I'd expect to see if this was a thermal issue with the *leads*- which aren't heatsinked except through the PCB traces. It's also possible they just mean the thermal resistance of the package prevents the dissipation of the heat generated by 75A, but then you'd expect to see it change with temperature. Not a flat line.

The 75A rating for TO-220 is some sort of (de facto?) industry standard, I've seen it quoted many times as limit for the TO-220's legs. But that's sustained (continuous) current, to actually "blow" (melt very fast) the leg, the momentary current is much, much higher. EUC Extreme blew off the legs of a mosfet in his MSuper V2 boards with RC-LiPo batteries, capable of very high currents. A short circuit with caps dumping their charge through can cause very high (momentary) current if the voltage is high and the resistance is small enough (kiloamperes, that is, 1000's of A).

 

Quote

 

tl;dr- are the new ACM2 transistors only actually rated for a sustained 75A (instead of the claimed 170)? apparently. is this because of the small leads in the TO-220 package? possibly.

Yeah, but the transistors can't even handle 170A continuous current in reality. The number in the datasheet is theoretical maximum value, assuming infinite heatsinking capability. If you work the numbers "backwards" from the actual thermal resistances of the entire system & ambient temperature, you get nowhere near the rated maximum current of the datasheet before the junction temperature has raised to critical levels and the component dies.

But, for switching-mode systems, like the wheels are for example, the momentary currents can get very high, it's the average power dissipation of the part that counts (assuming the momentary currents when switched on aren't anything completely "insane"). The absolute maximum rating of the HY3712 for pulsed current is given as 585A when the case is at 25C (again a theoretical number).