WARNING high mileage/old wheel? Check your capacitors..

[Planemo]

OK I was going to put this in the Gotway section but thought I would put it here in case other wheels use the same or similar capacitors.

I was going to wait a little longer before I made this post as ideally I wanted to strip the board first but rather than hang about I wanted to post what I do know, as I would hate to hear of any riders suffering an injury due to my delay.

The scenario is this: the wheel in question is a Gotway MSX 100v 1230Wh. It was purchased around 2.5 years ago and had faultlessly covered some 5500 miles. At the point of failure, the speed was around 20mph on flat ground at a constant speed with very little wind. This wheel has been used hard throughout it's life, and was being pushed hard only minutes before the failure (thankfully) as it could have been a lot worse.

The wheel suddenly lost balance. No warning, no beeps, nothing. No bad smells, smoke, pedal dip or anything whatsoever to suggest there was a problem. Quite nasty injuries were sustained.

After the crash, the wheel powered up (I don't know if it was still on or needed to be turned on) and balanced no problem. A tentative push around by hand seemed to show all was OK. It was then ridden around at walking speed no problem for about 5 minutes. Much head scratching ensued. A thorough look around the surface where the wheel lost balance showed no suspicious dips, bumps or anything. Smooth asphalt path.

The rider gingerly got back on (we can argue all day that he maybe shouldn't have done but there we go) and continued at around 10mph for 500 meters or so and then it shut down again, same scenario, and a few more additional scrapes.

At this point, the wheel was deemed totally unfit to ride and the rider was collected by car. the side panel was removed at scene and one of the caps was found to have blown (pics below) which at this time is believed to be the culprit. No other damage can be found on the board but as I say I cannot be 100% confident on this until I strip it down to inspect the FETs properly. The point of this post, and following reading up of caps in general, is that the capacitors used can fail at any time without any wheel warnings. Reading up, it seems that these type of caps can 'dry out' over time (greatly increased from heat during use) and the No1 tell-tale of a cap on it's way out is either bulging or splitting of the end seal, although they can discolour or bulge anywhere on the body. As you can see, the cap in question has 'burst' it's end seal, although the swelling very likely started gradually some time back, maybe months.

So, I ask anyone with a high mileage or old wheel, please take your side panel off and visually inspect the caps, with a magnifying glass if needed, and see if there is any swelling whatsoever of the caps. The ends should be absolutely flat, the burst seal totally intact, no signs of discolouration and no swelling anywhere.

For me, this will now form part of my periodic inspections and in any event, I feel that either the board or at least the caps should be considered 'lifed' and replaced every 3k miles or so to be on the safe side. This failure could have ended so much worse, and the rider is incredibly thankful for that.

The only possible indication that has been brought up that something wasn't right before the failure was that the wheel was losing charge - very slight - about 0.5% a day if that - but it was put down to the cells simply being old and very well used so nothing more was thought about it. Maybe the cap was leaching power prior...? A new board has solved the problem and there doesn't seem to be any more battery drainage issues.

Finally, I have since heard from a respected dealer that he was actually aware of this possible cap problem on older/well used wheels. I believe that as we start seeing the mileages rack up on the older wheels, we will undoubtedly start seeing more problems. Please keep your eyes on this issue. Note that the other big capacitor does have a worrying 'star burst' looking mark on the shell (I will inspect this more clearly when I get the board, although the end seal did look fine in other pics), and of further note is that the smaller capacitor does appear to have a slightly opening end cap as well..

[Jengajuice]

Thanks, Planemo.  This was very informative.  I'm riding a second hand MSX 100V with some good mileage on it right now. I'll be sure to watch for signs if increased charge loss and add periodic board inspections as well.

I do wonder the what the lifespan of the other components are. I know for caps it can be notoriously bad. What are the next most vulnerable components? I can manage replacing caps, and would definitely consider doing that, considering them a wear part. Replacing the whole control board seems like that would lead to supply chain issues, though. Not considering the caps, what is a reasonable expected lifespan of a control board? (Question for anyone. )

[Chriull]

32 minutes ago, Jengajuice said:

Not considering the caps, what is a reasonable expected lifespan of a control board? (Question for anyone. )

Fan (bearings) will die sometimes. The rest will quite survive until it's eventually sometimes fried (mosfets,...)

[Planemo]

I'm not sure that anything else on the board would 'wear' tbh. If everything else is up to spec (and it appears so given the hard life this particular wheel has had), I would think it's just the caps and that the board will last the realistic life of the wheel but as you say maybe someone else can tell us otherwise.

[Ben Kim]

a parts description with a link to the appropriate replacement caps on digikey would be very helpful. 

 

[Chriull]

17 hours ago, Planemo said:

I would think it's just the caps and that the board will last the realistic life

Maybe one more thing - if there are bad manual solder joints (mosfets, capacitors, wires,...), they will lead to premature failure.

8 hours ago, Ben Kim said:

a parts description with a link to the appropriate replacement caps on digikey would be very helpful. 

Here is a recommendation: 

Imho mouser, digikey, etc should be nice sources to find an appropriate cap (size, voltage, capacitance).

[Planemo]

I only ever use Mouser, Farnell or RS for my components. Far too many knock-offs floating around. I have even had fake fets before which were utterly useless. Not knocking digikey I think they are fine.

High quality caps appear to be:

Nichicon (stock MSX), Rubycon and Chemicon. Around £3.00 each. I will definately swop a pair out at around the 3k mile mark, if not before. I will also do the smaller cap if it looks ropey.

[Chriull]

2 hours ago, Planemo said:

Chemicon

Afair they had the caps with the lowest ESR and highest ripple currents at the digikey search result. Just about half of the found caps had no ESR specified, like the ones from

2 hours ago, Planemo said:

Nichicon

Maybe interessted ones find something on their home page. Or one trusts in a quality product?

[Planemo]

Yeah I will see what I can dig up 

[Lilvodka]

My 84v msx is on 2717 miles and looks okay.  Was hoping to get to 10,000 miles with just tire changes.  

[Chriull]

6 minutes ago, Topkek said:

My 84v msx is on 2717 miles and looks okay.  Was hoping to get to 10,000 miles with just tire changes.  

The smaller cap (top right) seems to be a bit bulged?

[Lilvodka]

2 hours ago, Chriull said:

The smaller cap (top right) seems to be a bit bulged?

Thanks I’ll keep an eye on it.  

[Alj]

13 hours ago, Topkek said:

Thanks I’ll keep an eye on it.  

It should not be sitting on top of hot MOSFETs. 105C is the temperature limit for those caps.

[Lilvodka]

1 hour ago, Alj said:

It should not be sitting on top of hot MOSFETs. 105C is the temperature limit for those caps.

I think they’re all like this aren’t they - placement on board?  We don’t have proper hot weather here anyway, surely would only be an issue in Nevada or similar climate right?  
 

Hopefully these big battery high power wheels will lead to better boards. 

[Chriull]

26 minutes ago, Topkek said:

I think they’re all like this aren’t they - placement on board?  We don’t have proper hot weather here anyway, surely would only be an issue in Nevada or similar climate right?  

No! Some roughly 70-100 °C case temperatures are just normal working temperatures for many ICs/Mosfets/semiconductors...

Case temperature shifts with ambient temperature - so about the 70 °C or lower in wintertime, 100°C and higher in the desert...

With some temperature peaks for (hopefully) short burden peaks...

Of course designing the circuit to have case temperatures of some max 50°C would be great for (extreme) longlivety, but often just not possible. Especially not in EUCs...

Edit: The main point here is, that placing the capacitor on top of some 50-70°C part gives the capacitor an ambient temperature of these 50-70°C! So if by his own power dissipation he creates a temperature delta of 50°C he would work perfectly up to a "normal" ambient temp of 55°C. But on top of a 70°C Mosfet he'll just die...

Edited July 16, 2020 by Chriull

[Seba]

On 7/14/2020 at 1:10 PM, Planemo said:

side panel was removed at scene and one of the caps was found to have blown (pics below) which at this time is believed to be the culprit. No other damage can be found on the board

From what you described, wheel stops balancing at random times. For me it doesn't have to be caused just by one failed cap. But I noticed strange white residues on some parts of PCB, especially on MCU (microcontroler)  and IMU ("gyro") legs and/or pads. This may be a electrolyte that leaked from the cap and then dried. There's a chance that this residue in wet conditions may become conductive enough to cause shorts and intermittent EUC operation problems, including momentary loss of balance. But it's just a one of many possible causes. For sure both caps needs to be replaced and board should be cleaned from any dust and residues (make sure not to wash out conformal coating, if any).

[Alj]

7 hours ago, Chriull said:

 

Edit: The main point here is, that placing the capacitor on top of some 50-70°C part gives the capacitor an ambient temperature of these 50-70°C! So if by his own power dissipation he creates a temperature delta of 50°C he would work perfectly up to a "normal" ambient temp of 55°C. But on top of a 70°C Mosfet he'll just die...

That is  correct, though if mosfet was transferring heat to heatsink properly it (in theory) should not get that hot as well. It is always combination of things. Caps need to be moved from the hot spot, also mosfet cooling will need to be fixed.

[esaj]

On 7/16/2020 at 4:27 PM, Seba said:

make sure not to wash out conformal coating, if any

Rehab1 sent me a broken ACM mainboard some years ago, if they still use the same stuff, it's pretty much impossible to wash off by accident. I soaked it in IPA (Isopropanol/isopropyl alcohol) that I use to clean flux off boards, took an overnight soaking and a stiff brush (repeatedly soaking & brushing), and I still couldn't get all of it off...  

Edited July 17, 2020 by esaj

[Denny Paul]

Can someone post a picture of what exactly we should be looking at? I dunno how to tell a MOS from a FET from the green and silver bits of computer-y stuff. 
 

(or link to a thread with good a good breakdown explanation?)