Need HELP on daisy chaining pls

Lynxmotion Robot Arms
1

This is my second post as there was not much response for my earlier one (Daisy chaining…).
This is an industrial project and as the wt. is less (200-250 gms) and the positioning accuracy needed is also less (around a mm precision will do), I thought of doing this with 5995 or as suggested by robot guru Nick, 5955s.
The arms will work around 15minutes and then stop and this is done around 10 times in 24 hrs. So it will work for 150 minutes in 24 hrs. It rests for most of the time

The servo side is ok. But can anyone help on linking many arms together to do the same work? Is it possible to do this?

2

while there is certainly nothing to prevent you from feeding the same servo output to the same servo position in multiple arms, it’s done with parallel servos on r/c control surfaces all the time with Y-cables, it seems that the problem will be alignment for each of the arms due to mechanical tolerances of the parts and positional tolerance between the different servos as well. you might be a lot better off to put a small controller, maybe one of the low end 8 output serial ones you can buy from other robot specialty stores, in each arm and then daisy chain the controllers (each with it’s own address) back to a single computer. This way you could individually control offsets and endpoint positions for each axis of each arm in software instead of finding some way to allow for it mechanically on the arms themselves. The difference is taking time to write slightly more capable software routines vs. adding either more (and more expensive) parts or swiss watch assembly and adjustment time to each and every arm. Holding 1mm positional precision over 4 or 5 axis of control and expecting to do it across many arms without lots of individual tuning time is more fantasy than reality. The trick, imo, is to put the tuning where it would have the least cost impact over the project.

3

Hmmm…
It all comes down to the fanout of the microcontroller that you’re using.
Theoretically, you could PWM to an infinite number of servos using a single microcontroller’s output pin.
In the real world, though, there’s a minimal amount of current necessary to properly convey a high or low signal to whatever you’re microcontroller is speaking to.
In your case, it’s speaking to each servo’s internal microcontroller.

A typical fanout for a digital I/O pin is 10.
In other words, you could probably get away with controlling 10 different servos all with the same pin.
That wouldn’t leave you with too much of a safety factor, though.

So, what to do when you need to chain more than 10?
I suppose that you could just use a transistor, but I’m not sure if that would be fast or clean enough.

My digital electronics textbook tells me that buffers are the way to go.
They can be used as signal boosters to kick the fanout up significantly.
If you can’t get the boost you need from a single buffer, then you should be able to tack on more buffers as needed.
Doing a quick search, I found that the 74LS07 seems to be a popular buffer.
I’m not sure if that one will suit your speed requirements (you’ll want to retain a resolution within a few microseconds, if possible, since the usual pulse length of a servo is a couple thousand microseconds or less).

Oh, and rather than truly daiseychaining them, it might be better to daisy chain them in such a fashion that all leads are the same length (i.e. bring all leads back to the controller and tie them together there).
This wouldn’t ordinarily be a problem, but if you’ve got a lot of bots, or just have a lot of distance between a few bots, it might be significant.
If you truly daiseychain them from one bot to the next, the juice will take a lot longer to travel all the way to the poor bum at the end of the chain.
It’ll mean extra cost in wire, since you’ll need much more length, but signal wire is rather inexpensive, anyhow.

Keep in mind, that I haven’t yet started my digital electronics class and am only going on what I’ve read.
That said, hopefully this’ll be enough to getcha pointed in the right direction.
And, hopefully I haven’t accidently lied to you.
:laughing:

4

This is what happens when I take a half hour to write a post.
Someone always beats me.
:stuck_out_tongue:

To add on to what Eddie said…
You could use a servo programmer and individually program the offsets into the internal chip of each servo.
But even after all that work, 1mm would still be unrealistic.

Maybe a 1DOF arm could achieve 1mm resoultion…
:laughing:

You could use complicated software tricks, like the program RIOS does to calculate for the distortion of the end-effector position due to gravity, but even after that I doubt that there’s a 1mm resolution.

5

sounds more like a job for cnc machine with a gripper,or ten of them

6

another idea . if the item and target are always in the same location, a flat arm (rotate only) with just a wrist and gripper (up and down only) and mechanicle stops should be able to thread the needle so to speek,and about daisychaining , uhh many ideas .but not enough know how