Since several of us on the GuzziTech site seem to have fallen victim to the Speedo sensor gremlin (mostly on Grisos), I thought I'd have a crack at finding a longer-term solution. I've not written this as a specific 'How-to', though it'd be easy enough to replicate the process.
To start with I've worked with the premise that the Hall effect sensor electronics are basically sound, more often than not (In my experience), failures of this type are a result of a process failure (Assembly). The store I bought the bike from was kind enough to let me experiment with a couple of defective sensors they'd amassed, plus a brand new replacement for my second failed unit. So I began looking at the body of the sensor under a microscope, probing where moisture might be getting in.
Firstly the cable entry point seems (As several have said), a likely candidate and as you can see from the 2nd picture there is a gap into which water could pass through. What the shot doesn't show is there doesn't seem to be any form of protection in this area other than the glue/conformal coating around the wire sleeve exit on the inside of the sensor body. You would have thought they'd have filled the bore of the hole with something to seal it?
Secondly the rubberized epoxy the manufacturers have used to cover/seal the PCB into the sensor case has not bonded sufficiently to prevent moisture ingress.
As you can see from the following two shots I used a jeweler's screwdriver to test the bond between the two surfaces, could open a gap between the two materials with very little pressure applied. What I found disturbing with both test samples was that approximately a third of the overall circumference had not bonded the surfaces adequately. Viewing the gap under high magnification I found moisture droplets in-between the two surfaces, though they evaporated quickly under the warmth of the lights.
Next I gently prized out the rubber epoxy, without damaging the PCB, was surprised at how cleanly it came away from the sensor case, without any tearing. As you'll see in the following pics, note the smooth internal walls of the sensor case (second pic below). This is what I suspect is the cause of all the trouble, the fact that the walls of the sensor case have NOT been scored to enable a good bonding surface.
I debated whether to pull the PCB right out of the case, decided against it since the risk of damaging the circuit board/magnet/cable was quite high., bearing in mind the sensor is about the same size as a postage stamp. So at that point I decided to dry it out thoroughly in an environmental chamber @ 50C for 4 hours. I used a chamber rather than a conventional household oven, or a hairdryer simply to maintain a consistent temperature, to be pretty sure all of the moisture had been expelled.
Applying the fix
Since the walls of the sensor case were essentially smooth they needed a good roughing up to allow a good bond, so I broke out a needle file and scored the internal walls thoroughly, also lightly scoring around the entry point of the cable sleeve. I decided on a second application of conformal coating, since, on pulling out the rubber epoxy a small amount of the CC came away on its removal. I also wanted to make sure that there wouldn't be an air gap around the edge of the PCB and the case wall + sealed the internal end of the wire sleeve, since in theory moisture could travel (Internally), down to the sensor if the wire sleeve got damaged. Once that had been applied with a small artists brush, had let that dry I could now mix up and pour the two part epoxy I'd chosen.
As you can see from the shot (Sorry I had to use my phone camera), having mixed up the two-part epoxy and poured it carefully into a syringe I could refill the sensors and lay a bead around the cable entry point.
The only thing to make sure of when filling them up is to do it in a way that will NOT allow for any air-gaps, or spaces. Once full it was just a case of laying down a strip of double-sided sticky tape on a flat surface, pressing the sensor firmly to the tape and let the curing process begin. I could have put them in the oven again to speed the process, but according to the instructions would have invited some shrinkage in the mixture. So I let them dry out naturally for several days.
You can see in the final shot the overall result.
Results
I've now fitted one of the sensors to the bike, have tested it in wet weather conditions, taken to the back roads for vibration tests and given the bike a good hosing down, during cleaning sessions and I'm happy to say no hint of a failure as yet.
*Note* This article is the result of some experiments by the author, will not accept any liability for faults/accidents/wrath of deities, etc incurred by anyone else attempting this fix, i.e. If you mess it up... Don't blame me!
I only say this as this is a public forum, we seem to sadly live in an age of litigation culture.
