Gantry Limit switches & homing question.

More
21 Jan 2019 02:43 #124506 by superlen
I am working on a wiring diagram for my gantry system and had a question about limit switches.

I envision two over travel switches (one on both ends of the axis) that are "oh sh@t" triggers and would drop out power to the servos, and then an additional home switch on the axis. Is this a typical setup for gantrys? My plan was to only wire the home signal directly into LCNC & have the overtravels simply wire in series with the estops. My estop relay circuit interrupts power from the drive to the linear motors, and I was also planning to have a LCNC input read the status of the estop so it at least knows an estop happened.

Does this sound appropriate, or am I missing something? Also, would there be a problem interrupting power between the drive and the motor, or would it be better to drop the power feed going to the drive?

Thanks in advance.

Len

Please Log in or Create an account to join the conversation.

More
21 Jan 2019 08:40 #124508 by rodw
The limit switches are your "oh sh@t" triggers and should sit outside of your soft limits and should never be triggered under normal operation.

You don't need to hook up the limits to estop as they are already designed do what you want to do and you can jog off the limits with ignore limits to recover from an error..

On my machine, I have maximum and minimum limit switches and a home switch on each joint. I have one shared home/limit switch on the Z axis due to space considerations. I found this the hardest to configure and you loose travel becasue your homing sequence needs to move off the home switch or the limit will be triggered as it comes out of homing. I also found you needed to ensure the home switches needed to stay triggered until the axis coasted to a stop.

My first estop button physically disconnected mains power to the stepper power supply. But then I added a couple of other remote switches (including one on a wireless pendant) so I added a relay which sent 5 volts to the stepper disable signals on the stepper controllers which is activated on an estop. This also disables the stepper motors.

Also, my magnetic breakaway sensor triggers an estop. All my estop signals are chained together in software using the estop latch component.

The only downside to dropping power to the steppers is that you will loose position when you hit estop so precise positioning might need rehoming...

I kinda think you are overthinking things as the LCNC dev team has done the thinking for you!

Please Log in or Create an account to join the conversation.

More
21 Jan 2019 14:11 #124540 by andypugh

superlen wrote: I envision two over travel switches (one on both ends of the axis) that are "oh sh@t" triggers and would drop out power to the servos, and then an additional home switch on the axis. Is this a typical setup for gantrys?


Hardware E-stop is typically only used to avoid injury to humans. You seem to be using it to try to avoid the machine damaging itself. I can't go as far as to say that this is a bad idea, but it isn't normally considered necessary.

What is the machine? Is it something that will actually destroy itself if it hits the physical end stops? I have seen videos of large commercial machines that have done this, smashed castings and all. But the typical stepper-controlled gantry can probably survive the experience.

Once a LinuxCNC machine is homed it should never exceed the soft-limits, so the switches are there in case something goes wrong.
if LinuxCNC is crashed badly enough to not see the limit switches activate then it will be too crashed to make step pulses. If you are planning a servo-motor controlled machine then this isn't always the case.

The typical LinuxCNC machine uses a limit switch at each end of travel connected only to the software and separate from the hardware e-stop. It is also very common to use one of the limit switches on each axis as the home switch. (bear in mind that a mid-travel home switch needs to be set up with a very long target or similar, so that the system can tell which side of the switch it is on so as to know which direction to move to find the switch.

Please Log in or Create an account to join the conversation.

More
21 Jan 2019 16:59 #124556 by superlen
Andy,

Thanks for the response.

This is a large gantry 5x10 with linear motor/servos on it. The Y-axis carriage can move 14' and my fear was that if LCNC or the servo drive gets lost & blew past the SW limit, I wanted a non SW backup. This is the way the original machine (large banner printer) operated, so the heavy duty mechanical stops are already there and the limit switches (Allen Bradly mechanical switches) are already there. There are also Banner prox switches that I am planning to use for home signal. Note: These are just for the X and Y axis. The Z axis I have to add a stepper as the original machine had no Z control. The operator would set the Z height with a hand wheel. On the Z I can easily design in the pair of end point sensors as you mentioned.

For a servo system is it common for the Estop to interrupt power going to the drive or interrupt the output of the servo drive feeding the motors? I would like to interrupt just the servo drive output as this would still allow LCNC to read my linear encoders & still retain a known position.(see note below) However I am concerned that disconnecting the output of the drives when they might be at full output driving the carriage could cause some issues for the servo drive.

Currently my linear encoders wire back to the servo drive and there is a slave encoder output on the drive that will feed LCNC via Mesa cards, so without the drive powered up, I lose encoder feedback.

Len

Please Log in or Create an account to join the conversation.

More
21 Jan 2019 17:07 #124557 by superlen
Rod,

The way you have your system is very similar to what I was thinking. Although I believe you have your "limits" also going to LCNC not outside in some sort of hardware protection circuit (estop related or not).

I made some comments when responding to Andy about interrupting power to the drive or between the drive and the motor. The drives also have enable signals, but if the drive crashes, the enable signal will likely not stop a runaway axis.

Thanks for the feedback and the posting of your gantry. I need to re-read your thread as it has a lot of good info.

Len

Please Log in or Create an account to join the conversation.

More
21 Jan 2019 18:49 #124559 by rodw
Yes, I stayed within the LinuxCNC machine paradigm and wired my limits back to software inputs. I think to include them in a hardware circuit would be creating a rod for your own back becasue you loose the ability to recover if you do hit a limit stop. You will have to manually move the gantry to clear the error rather than ignore limits and jog off.

I think also it is wise to have separate home and limit switches on a machine with a lightweight high speed gantry (21 m/min in my case) as the machine coasts to a stop while homing over say 50-70mm. A combined home/limit sensor would reduce your travel by at least this much. My limit switch sensors are designed as Andy has described for mid travel just I'm only in the field of travel far enough to account for the coasting.

I think your servo drives can be considered as part of the hardware and you only need to protect from a software (eg LinuxCNC) fault. You seemed to want to include them in your fail safe. I think to implement your additional failsafe, you just need to add separate overtravel switches that disable the servo drives. You probably only need one switch at each end of the table to achieve this instead of on each side.

Somewhere you need to trust your motion controller and I think if you can't trust LinuxCNC perhaps you should use another motion controller. I saw a badly twisted plasma gantry end made from 20mm aluminium during the week where the machine must have racked the gantry. Which begs the question. How do you protect LinuxCNC from a hardware fault? I don't know the circumstances of the crash but I think one servo must have failed (it was a commercial machine). You can't twist a gantry by hitting limit switches!

BTW, 10 ft is not a large gantry. Some I've seen are 90 metres long!

Please Log in or Create an account to join the conversation.

More
21 Jan 2019 18:59 #124561 by andypugh

superlen wrote: For a servo system is it common for the Estop to interrupt power going to the drive or interrupt the output of the servo drive feeding the motors?


Many stepper drivers blow instantly if you disconnect an energised motor.
Every VFD I have read the manual of has strongly warned against switching between the drive and the motor.
I can't speak for many servo drives, but I would suspect that they would be equally unhappy to have the motor disconnected while energised.

Please Log in or Create an account to join the conversation.

More
21 Jan 2019 20:08 #124562 by superlen

rodw wrote: I think to implement your additional failsafe, you just need to add separate overtravel switches that disable the servo drives. You probably only need one switch at each end of the table to achieve this instead of on each side.
...
BTW, 10 ft is not a large gantry. Some I've seen are 90 metres long!


Rod,

The AB limit switches that are on there now are the overtravel switches & I wasn't planning on tying them into Linux, just into a hardware disable of some sort. For the X-axis (14') there are only two...one left, one right. I'm unclear on your comment between each end vs. each side. I *think* we are saying the same thing. I will be adding 1 home prox to each axis that of course will tie into Linux.

5x10ft seems large in my shop....90 meters is insane!!!

I have considered the fact that I would have to manually move the carriage off the stop, but I'm ok with that...In theory this should never happen, only in an actual error or if an axis was manually moved during power off. FWIW, two of my pick and place machines (pcb board assy) are wired this way. If I am working on the machine and leave an axis all the way to one side & I try to zero...it will complain & I have to estop, raise the safety cover, move the axis off limit.ect. It's not too bad and doesn't happen often.

Len

Please Log in or Create an account to join the conversation.

More
21 Jan 2019 20:11 #124563 by superlen

andypugh wrote:

superlen wrote: For a servo system is it common for the Estop to interrupt power going to the drive or interrupt the output of the servo drive feeding the motors?


Many stepper drivers blow instantly if you disconnect an energised motor.
Every VFD I have read the manual of has strongly warned against switching between the drive and the motor.
I can't speak for many servo drives, but I would suspect that they would be equally unhappy to have the motor disconnected while energised.


Andy,

These are my thoughts as well. I know VFDs get unhappy. I have little experience with drives, but surmise they are they same. I'll just kill power to the entire drive. Thanks.

Len

Please Log in or Create an account to join the conversation.

More
21 Jan 2019 20:24 #124564 by rodw

superlen wrote:

rodw wrote: I think to implement your additional failsafe, you just need to add separate overtravel switches that disable the servo drives. You probably only need one switch at each end of the table to achieve this instead of on each side.
...
BTW, 10 ft is not a large gantry. Some I've seen are 90 metres long!


Rod,

The AB limit switches that are on there now are the overtravel switches & I wasn't planning on tying them into Linux, just into a hardware disable of some sort. For the X-axis (14') there are only two...one left, one right. I'm unclear on your comment between each end vs. each side. I *think* we are saying the same thing. I will be adding 1 home prox to each axis that of course will tie into Linux.

5x10ft seems large in my shop....90 meters is insane!!!


Len

I know about playing Industrial Tetris. And my table is only 6' square.

I was thinking you'd Add your home switches and use your existing limit switches as LinuxCNC limit switch inputs and add 2 more switches, one at each end of the table for your hardware override. But I guess you'll need to do that on the gantry axis as well.

But personally, I'd just set it up as LinuxCNC expects rather than step outside their paradigm even i it needed revised wiring. I've never seen LinuxCNC hang when running a job so I trust the paradigm.

Please Log in or Create an account to join the conversation.

Time to create page: 0.094 seconds
Powered by Kunena Forum