Encoder repair - bulb replacement

What up?

The spindle encoder I spent ages fitting to the Bantam either never worked or has since demised, possibly due to rough handling on my part. I DID check the spindle encoder but I forget the specifics.

Certainly, trying to talk to it using Halshow is a fruitless effort. The circuit is getting power and there is some sort of voltage on most of the outputs. Main issue is that none of them actually change state. That's a bit of an issue in an encoder.

Let's take the bugger apart:

Can't be difficult to figure out WTF is happening, or not as the case may be.

Sure enough, the mechanical bits comprise an optically slotted glass wheel with a similar slotted (stationary) grid providing a sort of Vernier effect. 

It's lit from below through a simple lens with a 5V incandescent bulb the size of a matchstick head. Or was until the filament burnt out. There's also some sort of flat optical receiver for the A, B and Z outputs.

Beyond that, there's a quad comparator and a couple of RS422 differential line drivers. Looks to me as if replacing the light source would be "all" that's required.

Here's the light source.

It's just a simple incandescent bulb of 3mm diameter, straight across the 5V supply.

How to fix it?

I found some of those tape LEDs. The package included a 12V battery with a switch and a single segment (3 LEDS per segment) wired up for sales purposes. This showed me that the forward voltage is about 3V under the recommended(?) current. I also measured 2.5V across the 150 Ohm resistor, suggesting a current of 17mA or so. Given that I have around 5V in the encoder supply, simply reusing this 150 Ohm with a single LED would result in around 13mA or so, a cunning plan came to be.

I snipped out one of the LEDs that was neighbouring the resistor and cut back the flex PCB to fit in the cavity where the bulb had been. I also cut and filed the cavity to get the LED roughly where the filament had been, then glued it in place with Araldite Rapid. With a layer of Krapton tape underneath for electrical isolation, it fits back into the original location. This was a bit of an act of faith, as it was only when I powered it up that I could be certain it actually worked. Which it did.

Obvs I couldn't move on without fiddling with the 3 pots. only one channel worked, so I assume the circuit is sensitive to the brightness to some degree. There's one per channel and there seems to be a sweet zone where the comparators do their thing. With all 3 in the middle of their zones,  I seem to have a result.

Pleased with this, as the same encoders are used on the 3 axes of The Shiz. This one came with in the bag of spares and claimed (correctly) to be faulty. It's now going to serve as the spindle encoder on the Bantam.

There.

Let's call that job done. And yes, it works in LinuxCNC.