What are good test setups

1

right now I am fighting through some development stuff using this frame

maslow_small_frame
maslow_small_frame1920×2550 531 KB

I have hardware on order to be able to setup additional test systems some will be just boards that I flash to but have not motors attached

but I will also be able to build test setups that have motors/encoders/etc attached.

I want to get people to think about what a useful test setup would be.

one without any stops that just has motors and encoders (no belts) will allow for some testing

but what else should I setup?

@ian_ab you commented about testing all boards, what setups are needed for that? does it really need a full machine (and how can you test that isn’t manpower intensive to watch the tests)

2 Likes
2

Unfortunately I can’t think of a non-user intensive way of testing cable connectors unless we could design a robotic way of locating and plugging all the ports on the control board. What I would suggest is we have a setup that test the motor ports putting out a voltage into a resistance and measuring for power and polarity and confirming they are within limits. The Z motors could drive a stepper motor in each direction and trigger a switch to confirm they had reached the limits. I would suggest the USB port could be used to confirm the wifi signals controlling the controller under test are behaving as expected. I would see this as a static test setup without belts or motors apart from the steppers. The fan port could be tested in a similar manner to the belt motor ports. Possible this could be only used for sample testing of a new batch of boards, and depending on the reliability, may not be needed at all.

1 Like
3

Unfortunately I can¢t think of a non-user intensive way of testing cable
connectors unless we could design a robotic way of locating and plugging all
the ports on the control board. What I would suggest is we have a setup that
test the motor ports putting out a voltage into a resistance and measuring for
power and polarity and confirming they are within limits.

by hooking up motors with magnets on the shaft and encoders, we can see the
motors move as expected.

It would need special test routines (we couldn’t just ‘pull tight’, but we can
test direction and possibly different speeds.

The Z motors could drive a stepper motor in each direction and trigger a
switch to confirm they had reached the limits.

yes, hook up the switch as a probe contact would let us test that as well

I would suggest the USB port could be used to confirm the wifi signals
controlling the controller under test are behaving as expected.

yep, that is one of the tests that can be done on a board with no motors

I would see this as a static test setup without belts or motors
apart from the steppers. The fan port could be tested in a similar manner to
the belt motor ports. Possible this could be only used for sample testing of a
new batch of boards, and depending on the reliability, may not be needed at
all.

I’m not just thinking in terms of Bar running tests in the warehouse, I’m
looking for tests that can be run automatically for every new build (or at
least daily) via automation.

I am planning to have 4-8 controller boards setup and hooked to relays that can
power cycle them and trigger other things.

David Lang

4

In a previous life I used entry level cognex vision cameras for quality inspection. One could be mounted above the frame and used to test xy movements.

Dano

1 Like
5

thinking out loud here, and leaving everything in so that others can be inspired by something I move away from and/or correct me

I can get a raspberry pi with a camera module (or a USB camera) for that.

If we want to have a system that can do calibration, we need a way to simulate detaching the anchors. Something like a 2nd set of motors that can pull the belts out and hold them in place (needs to be higher torque than the main set, but can be slower, I don’t thing extend-all powers out at anywhere near full speed)

make some custom spools so that when something goes wrong and it unspools to far we can reconnect (or at least just print new spools and re-use the belt) rather than have to buy new spools and belts by having them permanently glued together.

it would be fantastic if we could figure a way to have ‘fake motors/encoders’ or a set of steppers to drive the encoders to produce the results we want (while measuring that the motors are turning in the right direction)

actually, a set of low-res encoders on the motors and steppers to spin magnets in front of the encoders may be the right thing to do

the encoders are 40mm/rev, so 200 step/rev motors would be 0.2mm/step, we can microstep to at least 1/2 step to get to 0.1mm/step (encoders measure to 0.011mm/step, but I don’t think the test rig needs to get that precise, unless we run into oscilation, and if we do I think the answer would be to add an option that let us define the encoders to 4mm/rev for the test rather than getting really complex or add high gear ratios).

since the speed of the motors doesn’t really matter that much, we could go with low-res encoders to measure them (and use motors without gear trains, they just need to move slow enough for us to read encoders.

for that matter, we can define a ratio between motor movement and encoder movement (it doesn’t matter if it varies a bit)

oh, but we do need to have a way to add a lot of load to the motor (to trigger current limits. this means the simulator hardware needs to know where it is, not just let the motors go.

could we have a resistor we drop in parallel with the motor to burn current?

hmm, more to think about.

1 Like