Short/arc fault in cold weather on XT-90

bombadero

10 mW
Joined
Jun 19, 2019
Messages
32
Hey, I had a similar problem to one seen in a few other threads here on ES, but mine is slightly different, so wanted to get everyone's take on it. I have a dual-motor, dual-battery, wired in parallel setup with components purchased from Grin. The wiring setup is basically two Y-cables plugged into each to form an 'x' between the batteries and two controllers to achieve a parallel wiring harness; works great. Initially, everything was stock as ordered from Grin, with APP connectors. I had issues with connectors out of one of the batteries melting and fusing one day while hauling a heavy load on a warm day. I discovered this is sometimes an issue with APP's (from ES, general web search and confirmed by a Grin employee), so I resoldered all 10 connectors with XT-90's, and have been riding it that way for several months. A few days ago I was taking the dog to the vet and then to the dog park; on the way to the vet, I experienced a sudden loss of power heading up a hill. The bike is 1.1 kW nominal power from both motors combined, so it usually crushes this hill, even towing a 55 lb. dog. I pulled over and power cycled and it went away, so I thought nothing of it. Later in the evening as I was coming home from the dog park, after cresting another hill and resting at a light for a minute or two, when I proceeded through the intersection I heard a brief 'pzzt' noise and the bike died shortly after starting out across. I has some momentum, so I was able to coast to the other side and pull over. After another power cycle, the bike seemed fine again and I finished the trip home without incident, although the bike smelled strongly of an electrical fire. It wasn't actually on fire or smoking, just had that smell.

After some troubleshooting, I narrowed it down to one of the Y-cables, one leg of which only registered ~8.5V on a multimeter, verified with a LunaCycle Power Meter. All other connections read ~58V, which is normal for this bike. The conditions on that night were humid and cold, but it didn't rain that day. Moreover, I covered all of my connections with cable repair tape, which if you're not familiar, sticks to itself like glue once you wrap it on; it's pretty much impossible to unwrap it, you have to cut it off. I have that wrapped from one half of the XT-90 on either side until it meets shrink wrap that had been applied by Grin. I call it 'poor-man's heat shrink', because the seal is easily as tight as shrink wrap. It's also extremely thick. As expected, when I cut that off, there was no moisture inside. What I did not do is apply any dielectric grease or liquid tape. The cables were burnt for a couple inches on the short cable leading to the Baserunner in the rear of the bike; on the Y-cable end, it was much worse, with gold-colored flecks on the outside of the cable insulation from the electrodes in the connector getting so hot they melted, and the carbonization from what appears to be a very small, very brief fire, extended several inches. Fortunately the cable repair tape and the Y-cable contained it. Back on the controller side, the connector had been damaged a little and the little nylon apron that clicks in to protect the solder points was going in at an angle. I was reminded on inspection that I had stripped my cables too much and about 5mm of exposed cable was sticking out of the apron for both hot and neutral. I repaired the connector for now as it reads correctly, supergluing it back into the right orientation with Loctite gel superglue clamped together, and reapplied two layers of tape this time, both high temperature electrical tape and a fresh layer of repair tape on top of that to seal it. I also wedged a piece of rubber between the two wires and taped them separately with electrical tape to insulate them from each other.

After some research, I will probably open that controller connector again and coat that exposed wire with Plastidip Liquid Tape, and grease them liberally with Permatex dieclectric as well for good measure, and do the same with all other connections on both sides. I ordered some new Y-cables to convert to XT-90 (a replacement and some back-up inventory) to replace the fried one. Going forward, I will never strip my cables too long and plan to always make liberal use of liquid tape and dielectric grease.

So here's my question though: does it seem like the 5mm of exposed wire on both hot and neutral was definitely the culprit, or does it seem like moisture may have played a factor? As I said, everything I examined was dry inside, but given how hot that tiny fire burned from the arc/short, if there was any moisture it was probably vaporized instantly. I don't like to use shrink wrap for the end of these XT-90 cables because the connector is so much wider than the cable, and it's not convenient to take apart and inspect when things like this happen. If you do that, to re-apply heat shrink you now need to de-solder and re-solder the whole thing, and then re-shrink it. So that second question is, does Plastidip + Permatex dielectric grease + high temp electrical tape + cable insulation repair tape seem like sufficient insulation against moisture? It seems like overkill to me, but I'd like to hear other folks' opinion. I live in San Francisco, so extreme winter weather is mostly not a concern, apart from the record-breaking bomb cyclone we had recently, but obviously I wouldn't be riding in that! Just light amounts of rain and extremely high humidity on some days.

I've attached a photo of the connection in question. BTW, in that photo there is a layer of automotive cable loom on top of the cable repair tape. I'm going to stop using that on cables because it holds moisture, but in this case there was the thick repair tape underneath it. Unlike the repair tape, the automotive loom can unravel a little, although usually it also has very good adhesion to itself once applied, which is why I use it on bikes for various things.

20220110_161809.jpg
 
That should take quite a lot of force. Do you think it was caused in the original event? Or is this the replacement unit?
It was the first replacement unit. I think what may possibly have happened is that I inserted the driveside bowlware and spindle into the BB simultaneously and perhaps the spindle was slightly misaligned to the sensor, and didn't seat quite straight. The alternative, which seems more likely, is that it came bent from Grin, but they didn't want to even consider that and in fact it occassioned some subtle, polite Canadian snark from Justin... basically implying I might be a hoser. For the latest replacement unit ordered from Aliexpress, I made sure to remove the spindle from the driveside bowlware completely, and mount both bowlware pieces fully first, before inserting the spindle, which is what the instructions indicate, and ensures that the spindle inserts perfectly flush.
An arc fault on a battery line can generate a lot of high-energy RF, which could directly affect and even damage things that are directly connected to the same battery line. But it shouldn't affect things that go thru a regulator or two to get their power, like the torque sensor (which is powered from the CA...but if the CA's regulator was damaged or destroyed and passed battery voltage to the sensor, it could destroy that..but the CA would also be destroyed, and very likely everything powered from it, like throttles, etc).
Agreed, but keep in mind this is after I had replaced the sensor and CA3 with entirely new units. The only old components were the spindle and PAS cable (I suppose the latter could be a culprit, I can try swapping that out). The throttle seems to be working fine.
If it were the sensor causing the problem by having too low an output, you would see the readings themselves be much much lower than before, for the same physical input.
If this were the case, I would be exceptionally unlucky, as this is the 2nd replacement sensor I have tried. I ordered yet another from Grin, which arrived yesterday but I have yet to pick up or try installing. I definitely plan to do another comprehensive/deep dive into the config, but there is nothing in it currently, accessible directly through the CA3 interface, that I can think that might be causing it.
I would guess that the last number is the length of the spindle (or BB shell), with 100 being a fatbike version. If it is not, and they all have the same spindle/BB length, then the number may designate the type / sensitivity of the sensor inside, so they may operate differently for the same inputs.
Yes, T9 is the model and the last digits are the BB width. The characters I am concerned with are the first 3: ERS or KYD. ERS sensors seem to be manufactured by Erider; KYD sensors seem to be manufactured either by a licensed 3rd party, or an unlicensed company outside of Hong Kong. Given PRC's treatment of HK, it wouldn't surprise me if a company outside HK was just straight up stealing ERider's IP, but I can't seem to find any info about who 'KYD' represents and if there are any differences between the two versions. As I said, even Grin's supply chain has had these KYD units in it, whether knowingly or not.
My first suggestion is to mount the fender to the bottom of the rack itself, rather than the fork, if that's possible; then it won't move around.
I though about doing that a while ago, but then the tire would move relative to the fender, which is not very ideal. I researched how motorcycles with USD forks resolve it, and they do it by mounting the fender at the very bottom of the fork so that the fender hardware does not interfere with fork movement and the fender is fixed relative to the tire, and both move relative to the fork. The only wrinkle here is my front racks, which obviously motorcycles don't possess. There are only about 5mm of clearance on either side, and forks, racks and fenders being wobbly the way they are, the metal fender bangs a little against the metal racks on bigger bumps. I ordered some u-channel rubber edging like some people use on automotive chassis around door frames, and I hope that does the trick if I line both sides of the fender with it.
 
As requested, here are some photos of the front fender install:
20220404_180325.jpg
Overall view.
20220404_180335.jpg
Close-up detail from the back. A discerning eye will notice that my mounting hardware fixing the stays to the fender are for fender stays external to the fender. I don't like the stays that came with the kit, so I have some hexagonal cross section SS stock that fits perfectly, but I haven't worked out how to affix the other end yet, so those stays will work for now. My first thought was to flatten that end out with a hammer and anvil after heating, but SS is very hard and my initial attempt at that proved very difficult. I would have to acquire a more powerful torch like a propane torch rather than my dinky Dremel butane torch and heat it to molten levels first to do it that way. My current idea is to use more of the same fender hardware attached to a small galvanized furniture strap where the current stays attach at the side of the bash guard with the stock passing through, which would have the additional benefit of making the stays adjustable.
20220404_180403.jpg
Side detail: Front stays attached to a 90-degree Cateye adapter from T-cycle from the inside front bolt of the fork bash guard; rear stays are attached on the side of the bash guard, between the bash guard and fork. The ensures minimum interference between the stays and fork/racks.

Another problem I have is the tendency of the fender to want to rotate backwards. Most fenders handle this by attaching the fender to the crown of the fork, but this being a USD fork, I can't do that. My solution is not the most elegant, but it works: I attach the front stays to the racks with Rovaflex Soft-ties. This would lock out the fork if these ties were not rubber, because when the rack moves relative to the fender, the distance between the two components increases. When the fork has no load, the soft-ties are 'tuned' so maintain an ideal position between fender and rack. When the fork takes a load, this position changes and the soft-ties stretch. Basically, I'm using fancy reusable rubber bands. A bit wonky, like many other things on my gonzo bike, but it works 8).
 
If you can swap out the fender mounting "stay" for something made of round tubing instead of the stamped-metal, so that it has the same stiffness at it's mounting ends and at the "U" where the fender attaches to it, it will reduce (perhaps nearly eliminate) the sideways movement of the fender.

One of the biggest problems of the stamped "stay" is that the point at which it bends doesnt' have any triangulation or other material to prevent easy bending of that angle. Same with the flat portion that bolts the fender to the fork.

Even just "solidifying" or triangulating those sections would laterally stiffen the fender.

If there is no rear mounting, or it's similar to the front, then that end is still going to flop around; you'd need something stiff at that end to fix that.
 
Back
Top