Activity Bot mechanical assembly step 7 & 8 heads up

Ray0665

After mounting the encoders on the activity bot remount the wheel and check to be sure the IR transceivers are visible between the encoder spokes if not, loosen the encoder mounting screws and adjust the position of the encoder relative to the servo so that they are visible and then tighten the screws and remove the wheel. Do this for both encoders. Missing this crucial step will require a lot of disassemble to correct the problem when the encoder does not output pulses as expected in later steps.

Ask me how I know this hee hee hee

erco

Thanks for the heads-up, Ray0665. My Activitybot arrived Friday and you may have just saved me some precious time.

Phil Pilgrim (PhiPi)

Ray0665,

Can you take a photo of a misaligned encoder, please, and post it? I'm more than a little surprised that misalignment to the point of failure is even possible.

Thanks,
-Phil

Ray0665

I am not anxious to disassemble again to photograph the encoder. At first i thought i missed the pin hole on the mounting plate but that was not the case. The misalignment was a combination of the encoder to mounting plate and mounting plate to servo screw hole sizes such that lens of the encoder was closer to the wheel hub. If you really need it I will. The thing did not fail completely but was missing pulses and I had difficulties calibrating. I discovered the misalignment when I wrote a test routine that followed the encoder pulse (out[26]:=in[14]) and rotated the wheel slowly by hand.

dgately

I had a similar problem on my right-side encoder. It caused the servo to surge when running the example code, after I had run the calibration code. I dumped the calibrated values and they sure looked weird, very unlike the left-side. The sequence of values was very irregular, not showing the ramping that the left-side data had shown. During the calibration I did not notice any surging, so I assumed that the poor data was later causing the problem as the bot tried to use those values with an encoder that was still "wonky" (yes, that's a very technical term). That told me that the encoder was not working correctly. So, I removed the wheel and reset the position of the encoder slightly. Now it works as it should!

I'll see if I can re-test by loosening and resetting the encoder until it fails as it did before. I'll let you know...

dgately

Ray0665

So I did some extensive analysis (ie took it apart and played with it). I am kinda right and kinda wrong in my earlier post. Here's the story: Referring to the pictures in step 7 and 8 the servo needs to be aligned in its slot as far away from the circuit board as possible and the encoder needs to be as close to the circuit board as possible. In other words separate them as much as possible. This places the IR transceiver in the center of the spokes and all is well. When they are not aligned this way the IR transceiver moves closer to the hub and out of center, in the extreme, the transceiver sees the solid bar between the small and large spokes and you get no pulses.

It is all about the relative distance between the transceiver and the servo hub.

@Phil I do have several pictures, but pretty much all you see is what I described above.
You could guarantee this could never happen by modifying the wheel to elongate the encoder
spokes towards the hub.

Will the new wheels be available for purchase, would love to use them on my BoeBot?


Ray

dgately

I was able to replicate the problem that I had with my original setup on the ActivityBot. I removed the wheel, loosened the encoder, pushed down towards the servo as I re-tightened the encoder screw. I re-ran the calibration code and then the display calibration code. And, again I got "wonky" figures for the right servo. I reran the process of removing the wheel, loosening the encoder and pushed up away from the servo with a re-tightening of the screw. There is a little play in the screw hole.

I've attached the data from a "good" calibration and a "bad" one. I also attached close-ups of the encoder diodes as seen through the windows on the wheels with "good" positioning and with "bad" positioning. The image with the diode visible slightly lower in the wheel's window depicts the "bad" case...

There's very little difference in the positioning, but I've repeated the test twice with these results.

dgately

Steph Lindsay

Ray,

Thanks so much for bringing this up, and thanks to you and Phil and Dennis for looking deeper into what is going on. I honestly never encountered this with the numerous sample robots built and tested during development of the assembly instructions (I think my crew built about 20). I have alerted Andy to this thread, and I'm sure he will appreciate the calibration text output.

Ray, yes, we will sell a 1 encoder wheel + 1 o-ring tire set on our website soon, I'm working on setting that up. You can order the parts individually now by contacting our sales team directly.