MSX 84V can't blow a mosfet? WRONG!

[mrelwood]

As it happened, I was able to fry my 84V MSX without even pushing a passenger car around (although I might have gained a bit of weight myself). I was thinking that since the electronics for this model has been constructed with Vibranium and Pixie Dust, that I can stress the wheel as much as I ever have the balls for. Therefore I didn't perhaps think through what I was about to do in much detail.

A mountain biker has calculated in his blog for the steepest side of a local hill to have 31m of elevation at an average of 12°, so about 150m of travel Having looked down the straight path a few times before, this time I decided that it's time to ride it down. I weigh 92kg.

The trip downhill went so well with my raised sidepads, extended pedal plates with shock absorbers and the H-666 tire, that I turned around and started riding up the hill right away. I soon learned that disregarding a few 1m slopes, this is the steepest incline I have ever ridden up. The MSX sounded like a geiger counter closing in on Chernobyl as I leaned on the raised sidepads, without which I wouldn't have gotten up the hill for more than the shallow start section despite the extra long extended pedal plates. I have measured going up a few 5m long 13° inclines, but the steep section of this hill was slightly steeper.

About 20 meters from the top I started hearing an additional noise from the wheel. I was still wondering wether it was pebbles in the wheel or something more serious, when 10m from the top the wheel said "clack" and turned off. The operational smoke came out from a previously cracked screw hole.

Opening up the wheel at home revealed that the motor wires doubled up under a retainer had melted shorting the leads. I was able to separate them though, after which the evenly stiffly rotating tire turned into steppedly stiff. Removing one of the motor leads from the control board makes the tire spin smoothly. Dismantling the control board, to my surprise the mosfets looked intact. (No glue, btw. Just one thin and large thermal conductor plate.) But the connector for one of the motor leads had molten it's solder, which was now resting at the legs of the first Mosfet. Two SMC resistors had also shifted from the heat. Removed the first mosfet, and measured that it had shorted all it's legs. Now the tire was spinning freely. Other mosfets seemed to measure fine without removing them.

The mosfet model is HY5012 (W), which I couldn't find available anywhere else but China (Aliexpress, a month to arrive). We'll see what choices I'll find for getting the wheel working sooner.

What can we learn from the event? 1) MSX is not made of Vibranium. 2) MSX uses rare mosfets. 3) Falling down at a steep uphill has the shortest fall. 4) Background notification "Connected" from Darknessbot does not mean it is recording data. 5) If you want to ride up stupid hills on your MSX, separate the motor cables first!

Edited July 15, 2019 by mrelwood

[meepmeepmayer]

Well, technically it didn't blow a mosfet Just the good old motor wiring melting, as seen in the car push test. Everthing else is just a consequence of that.

I'm surprised it happened so quick. How can you push a car for 6 minutes but a minute on a hill kills it? The way down will already have heated the wires heavily, but even that included it was only a short time, wasn't it?

[Gaz Bon]

So just out of interest how old and how many kms on yr MSX

[mrelwood]

3 hours ago, meepmeepmayer said:

Well, technically it didn't blow a mosfet Just the good old motor wiring melting, as seen in the car push test. Everthing else is just a consequence of that.

Most likely yes. But technically, the mosfet is blown!  Yeah... Should've gone with another title.

Quote

I'm surprised it happened so quick. How can you push a car for 6 minutes but a minute on a hill kills it? The way down will already have heated the wires heavily, but even that included it was only a short time, wasn't it?

That's what I've been starting to wonder as well. We had just had a break at the top of the hill, but I didn't turn the MSX off. Still, the stress level was zero for about 15 minutes before I started the descent.

3 hours ago, Gaz Bon said:

So just out of interest how old and how many kms on yr MSX

Darknessbot shows 6800km, but due to the bug in distance counter when lifting the wheel up after a fall, I'd say I'm very close to 7000km. The wheel had it's first birthday 3 days ago.

Edited July 15, 2019 by mrelwood

[mrelwood]

Isn't @Marty Backe's Overheat Hill around 13° at the steep spots? While I don't understand how he stays on the wheel without raised sidepads, the length of the trip is an order of magnitude longer than mine was.

Maybe the core issue was that I had to go so slow? Higher current peaks when switching, and a single mosfet pair / motor cable is active for a longer time?

[erk1024]

It's a drag that your wheel is out of commission. I hope you get it rolling again soon. Glad you didn't get injured.  

It's cool that you have the skills to fix it up. I'm sure when you're done it will be better than the original.

[EUC Custom Power-Pads]

3 hours ago, mrelwood said:

If you want to ride up stupid hills on your MSX, separate the motor cables first!

That with the bundled cables of the engine I know from my 700 electric model helicopter with 50 volts / 300 amps peak. I had the idea to put all 3 cables together in a shrink tube instead of leaving them separated and side by side. As protection from an edge of a carbon side plate and that only on a length of 3cm. After a harder use they looked exactly the same as yours. 

 The unwanted way to earth was not as short as that of your MSX, but for me this experience was probably not as physically painful as it was for you.

Spoiler

 

 

[LucasD]

I think your misfit may be fine, simply the wire get so hot that unsolder stuff

[mrelwood]

1 hour ago, buell47 said:

The unwanted way to earth was not as short as that of your MSX, but for me this experience was probably not as physically painful as it was for you.

I think both you and me experienced the same physical pain, a broken heart! I was riding at walking speed and I wear motorcycle gear.

8 minutes ago, LucasD said:

I think your misfit may be fine, simply the wire get so hot that unsolder stuff

Unfortunately the first mosfet of the row is permanently shorted. The mosfet is only sold in China, so my break in riding the MSX seems to be a lot longer than I thought.

[Marty Backe]

Wow, and sorry for your loss. I've been there 

Looks like melted insulation that shorted the wires which blew the MOSFET.

Were you monitoring the current? When going up steep hills, I always monitor my motor current. If I have 90-amps going for more than a handful of seconds, I stop immediately. Because I know that even the MSX can fry, as you've just discovered.

Hopefully your hall sensor wires did not get melted too because then it's also a new motor.

Edited July 16, 2019 by Marty Backe

[Joker10]

Hope you get your wheel running again soon.  

[mrelwood]

 

4 hours ago, Marty Backe said:

Wow, and sorry for your loss. I've been there 

The thought of having to carry a dead 24kg wheel down from Overheat Hill or such, has given me nightmares for a year already. We had perhaps 1 km to the parking lot, and if I didn't have the guys to help me (thanks guys!), I would've tossed the damn wheel to the bushes. Instead:

https://imgur.com/QTnLXaJ

4 hours ago, Marty Backe said:

Looks like melted insulation that shorted the wires which blew the MOSFET.

Indeed. Although, I think the Mosfet actually blew due to the blob of solder that had molten and dropped from the motor wire connector directly to the legs of the first Mosfet.

4 hours ago, Marty Backe said:

Were you monitoring the current?

Nope. Vibranium, you know... I have never ridden a hill like this so I hadn't formed such habits. And although Darknessbot gives a "Connected" notification when restarting the wheel, it doesn't start recording data until the app is opened, so I can't even check the data afterwards.

4 hours ago, Marty Backe said:

Hopefully your hall sensor wires did not get melted too because then it's also a new motor.

Luckily they looked fine for the length that can be visually observed.

 

I'm contemplating my options regarding repair here:

https://forum.electricunicycle.org/topic/14594-msx-84v-mosfet-blown-options/

 

[MrRobot]

Shock absorbers? 

[neurokinetik]

FYI, that copper on the middle leg of the MOSFET is the feedthrough from the 2-sided circuit board.  Be aware that when you replace the MOSFET, you're going to want to make sure you've got a good, solid connection on both sides of the board.  Might want to use one of those replacement feedthroughs that they sell for this kind of repair.

[mrelwood]

1 hour ago, neurokinetik said:

FYI, that copper on the middle leg of the MOSFET is the feedthrough from the 2-sided circuit board.

Wow, good catch! I always solder mosfets and other crucial components on both sides anyway, but this is definitely something worth concentrating on.

 

[Azze80]

On 7/16/2019 at 3:12 AM, Marty Backe said:

 

Were you monitoring the current? When going up steep hills, I always monitor my motor current. If I have 90-amps going for more than a handful of seconds, I stop immediately. Because I know that even the MSX can fry, as you've just discovered.

Hopefully your hall sensor wires did not get melted too because then it's also a new motor.

Hi Marty,

How do you setup this monitoring of the current, via an app? Does it beep if reached a certain level that you've chosen ex at 88A ? .. Is the procedure then just to start, connect to the wheel before you start to ride? Thought of doing it to the Tesla V2 if needed?

Edited July 24, 2019 by Azze80

[Marty Backe]

2 hours ago, Azze80 said:

Hi Marty,

How do you setup this monitoring of the current, via an app? Does it beep if reached a certain level that you've chosen ex at 88A ? .. Is the procedure then just to start, connect to the wheel before you start to ride? Thought of doing it to the Tesla V2 if needed?

I use WheelLog and a Pebble watch. WheelLog has various alarms including one for current. I use 90-amps for the level that triggers the alarm

[Seba]

3 minutes ago, Marty Backe said:

I use WheelLog and a Pebble watch. WheelLog has various alarms including one for current. I use 90-amps for the level that triggers the alarm

Upcoming release of euc.world WheelLog will have two current alarm settings - one for peak (momentary) current and another for sustained (short-term averaged) current. I think it will be especially useful with Gotway wheels that use motor phase current instead of battery (DC link) current. There is a chance it will be released this week.

[Marty Backe]

10 minutes ago, Seba said:

Upcoming release of euc.world WheelLog will have two current alarm settings - one for peak (momentary) current and another for sustained (short-term averaged) current. I think it will be especially useful with Gotway wheels that use motor phase current instead of battery (DC link) current. There is a chance it will be released this week.

Very cool feature. I still need to spend some quality time with your version of WheelLog. Here's a couple immediate comments since I've been using it for a few weeks:

• It disconnects a lot from my various wheels whereas the Playstore version stays connected. Any ideas why this may be the case? Did you change any related code?
• The Playstore version would play an audio tone when my current alarm was triggered. Your version doesn't appear to do this. Or am I missing a setting?  I don't think I like the voice prompts for how I use the alarms, but I need to experiment more.

[Azze80]

22 minutes ago, Marty Backe said:

I use WheelLog and a Pebble watch. 

Is that for easier monitoring? I mean instead of monitoring via the phone? 

If I don’t have the watch I’d have to pick up the phone I guess to check the status?

[Marty Backe]

1 hour ago, Azze80 said:

Is that for easier monitoring? I mean instead of monitoring via the phone? 

If I don’t have the watch I’d have to pick up the phone I guess to check the status?

Yes. Nothing beats the convenience of glancing at your wrist for riding stats. And the current alarm buzzes the watch so I don't even have to look at it. My wrist buzzes whenever I exceed my alarms.

[Azze80]

10 hours ago, Marty Backe said:

Yes. Nothing beats the convenience of glancing at your wrist for riding stats. And the current alarm buzzes the watch so I don't even have to look at it. My wrist buzzes whenever I exceed my alarms.

Perfect! Great advice, will have to google them and pick one that won’t cost me more than the wheel🙂.

[Seba]

11 hours ago, Marty Backe said:

It disconnects a lot from my various wheels whereas the Playstore version stays connected. Any ideas why this may be the case? Did you change any related code?

Euc.world WheelLog is often used to "ride a beep" with BT speakers or earphones. This is very useful as EUC beeps often are hard to hear in wind noise. With WL you are always notified by three distinct beeps (each per speed alarm) and "Slow down" message. I consider this a safety feature, so it needs to work reliably. If not, rider must be aware that there is problem with data feed from EUC.

Play store WheelLog only determine if BT connection is open or not. But open BT connection doesn't mean that any data is sent over it or the data is actual. I've changed this logic, so it is more sensitive to any problems or even delays in data received from EUC. If your phone is under heavier load, this may cause problems timely processing the data arriving. This in turn may cause intermittent disconnected/connected messages. You can always turn this message off in settings.

12 hours ago, Marty Backe said:

The Playstore version would play an audio tone when my current alarm was triggered. Your version doesn't appear to do this. Or am I missing a setting?  I don't think I like the voice prompts for how I use the alarms, but I need to experiment more.

My version currenlty provide repeated alarm signal and voice alert "Current - too high!" when current is above alarm setting. But I'm thinking to also add variable pitch signal (without voice alert or with just value, eg. "<beep> 95 <beep> 90...") and pitch will be proportional to excess of alarm value. This would be very useful for you when performing load tests.

[Marty Backe]

7 hours ago, Seba said:

Euc.world WheelLog is often used to "ride a beep" with BT speakers or earphones. This is very useful as EUC beeps often are hard to hear in wind noise. With WL you are always notified by three distinct beeps (each per speed alarm) and "Slow down" message. I consider this a safety feature, so it needs to work reliably. If not, rider must be aware that there is problem with data feed from EUC.

Play store WheelLog only determine if BT connection is open or not. But open BT connection doesn't mean that any data is sent over it or the data is actual. I've changed this logic, so it is more sensitive to any problems or even delays in data received from EUC. If your phone is under heavier load, this may cause problems timely processing the data arriving. This in turn may cause intermittent disconnected/connected messages. You can always turn this message off in settings.

My version currenlty provide repeated alarm signal and voice alert "Current - too high!" when current is above alarm setting. But I'm thinking to also add variable pitch signal (without voice alert or with just value, eg. "<beep> 95 <beep> 90...") and pitch will be proportional to excess of alarm value. This would be very useful for you when performing load tests.

Thanks for the replies.

Unfortunately the disconnects are so common for me that I'll probably have to go back to the previous version. The Playstore version just works better with my phone. And the voice messages for the current alarms are annoying to me and not practical.

Everyone is different. Looks like other people are really enjoying your version so I'm the outlier 

[Marty Backe]

Battery Percentage Not Correct

While performing my formal Nikola range tests I confirmed that WheelLog does not report accurate battery percentages. It appears to be 10 to 15 percent too low. For instance, under load WheelLog was reporting ~0-percente battery yet the wheel had over 15-percent battery remaining.

FYI.