Weird problem: either all dynamixel AX12's discovered or none.

[makegeneve]

Hi,

I've built a PhantomX Hexapod mkIII and have a strange problem with it. Either all the 18 dynamixels are detected or none. I've added a bluetooth serial port to the standard build and this just confirms it.

Sometimes it powers up and moves after detecting all 18, then stops, and debug checks indicate no dynamixels detected again.

Surely a loose cable would just affect one string no?

It's a standard build based on the Arbotix-M board with the AX/MX power expander. I'm running it from a 12v 10a PSU for now.

Any tips on this type of fault?

Thanks in advance.

[jwatte]

Some things to check:
If there's an intermittent short somewhere, it could make the TTL bus not work right.
If there's a power loss to the serve power rail, it could affect all of them.
If the bus baud rate is unstable/marginal, or changes, then it could affect all of them.

Do you have a logic analyzer (like a Saleae Logic) or an oscilloscope, to check the TTL bus traffic when it works / doesn't?
Might be useful to measure at the controller board as well as at the end of each strand, perhaps.

[makegeneve]

Thanks. The power seems OK, as the servos are holding position. I'll put a scope on the TTL bus and take a look.

An intermittent short sounds like it would give this symptom. Is that likely to be a bad cable, or worse?

[KurtEck]

When I have seen things like this, it is often a cable, that maybe got caught in a joint and breaks through the insulation

Or maybe a servo has gone bad. If I get into this state, I try to unplug all of the cables from the MUX and then only have the servos for one leg connected and see if it works, if not, try different leg. If it does, try to add another one... And see if I can isolate the issue.

[jwatte]

If it affects all servos, it's more likely a short than a loose connection.
it could be a bad connector (more likely than a cable) or it could be a bad solder joint on a board/hub or it could be a board/hub being a little too close to some metal on the underside.
It could also be a cable itself, if the insulation has been breached/pinched, and makes contact with another wire or other contact point (such as chassis.)
It could also be a bunch of other things :-) I wouldn't count out bad baud rates, for example.

[makegeneve]

Thank you!

Replacing cables (4!) seems to have done the trick. Unfortunately I can't tell what is wrong with them. I'll leave that for another day.

But the robot is working again - for more than 30 minutes :-)

I think I'll put some cable braid around my cables where they are going through the chassis/legs.