#robotics Logs

May 18 2018

08:48 AM Helle: I have a really boring mechatronics question (not directly in a full robot, but it may end up being in one) that has been frustrating me
08:52 AM Helle: I was looking at linear actuators, but the ones I want (in terms of stroke and force) do not come (anywhere near affordably) with position encoding of any time (potmeter or hall effect sensor)
08:52 AM Helle: now I was pondering adding my own, as I need to keep two of them in nearly sync (play is 1-2mm), is a linear optical encoder with a greycode of some description the best approach and can that, with just DIY (no CNC) approaches give a reasonable accuracy
08:52 AM durrf: what is moving the actuators?
08:52 AM Helle: these are the "ready to go" style actuators with a built in DC motor
08:52 AM Helle: I looked at adding an encoder internally, but the gearbox ends up being a nightmare on them
08:55 AM Helle: (I can increase play to 3-5mm with some effort if that helps)
10:38 AM z64555: How does the gearbox affect the encoder?
10:39 AM Helle: z64555: that is where the encoder would need to sit to be a direct read out
10:39 AM Helle: else I'd have to add a totally external read out
10:39 AM Helle: https://www.amazon.com/Encoder-Linear-Sensor-Roland-XJ-740/dp/B077GRYK7F reusing one of these may be an option, with some mounting trickiness
10:40 AM z64555: you've lost me. Where are you attaching the encoder relative to the motor, gearbox, and movement arm
10:40 AM Helle: z64555: I am not attaching it anywhere, but if you buy a linear actuator with feedback potentiometer or hall effect sensor, said sensor is mounted in the gearbox of the actuator
10:41 AM z64555: presumably on the output shaft of the gearbox?
10:41 AM Helle: probably, given that that is likely the slowest turning bit
10:42 AM z64555: Probably, because that's closes to the movement arm while remaining rotary, and if you put it on the input side of the gearbox you'll have to deal with backlash hell
10:42 AM Helle: eeeeh, these devices aren't super accurate ever
10:42 AM Helle: but probably
10:44 AM z64555: That's likely your best option, aside from going with a linear pot attached to the movement arm
10:45 AM Helle: yeah, it's not an option though, I cannot practically modify that part
10:45 AM Helle: the external optical linear sensor is probably safer
10:47 AM z64555: can't replace the output shaft with something longer, so that you can attach a simple pulley?
10:48 AM Helle: nope, not practical at these forces
10:48 AM z64555: er, what
10:49 AM z64555: do you have any pictures of the actuator
10:57 AM Helle: https://imgur.com/a/BMpoeAy this style of actuator
12:56 PM polprog: heh. playing with my dremel
12:56 PM polprog: i found a brass rod, snipped off half an inch off one end
12:57 PM polprog: used a file to smooth the end into a ball like shape
12:57 PM polprog: round tip etc.
12:57 PM polprog: it kinda flow-drills plastic
01:00 PM polprog: its not perfect, but it works
01:10 PM polprog: i also got some of those things you use to thread holes for screws
01:11 PM polprog: so several items in my room have screws sticking out of them in weird places
03:09 PM z64555: Helle: what about sticking a magnet on the actuator arm? It's been awhile since I last opened one up, but I think it's just a screw inside of a piston
03:18 PM Helle: z64555: you mean to have it turn around and once per rotation get a magnetic blip on a sensor, possibly, that is very much relying on the internal design of something I am not 100% certain of the exact build of
03:19 PM z64555: so take one apart
03:19 PM z64555: if you can, that is. If you're on a budget you'd probably not to accidentally break a part you need
03:21 PM Helle: Exactly
03:21 PM Helle: haven't bought them yet, wouldn't be surprised if they are riveted shut or some nonsense
03:24 PM z64555: I've seen a few that had the tubes riveted shut. so... yeah.
03:48 PM Helle: exactly and I just can't get the affordable ones, otherwise to my spec, but with potmeter, I'd have to use 4 instead and syncing up 4 is a headache I am not touching