External ESTOP trouble

Hi, 

I am retro-fitting my 1968 Bridgeport Series 1 Papertape NC machine with rpi5/Mesa7i96s/Gecko203v.

I've got everything working pretty well it seems, except for the ESTOP.

I have (and want) an external/manual ESTOP-in button.  It is currently setup Normally Closed and appears to be successfully plumbed through to the GUI, which is to say that the GUI ESTOP button tracks the status of the hardware ESTOP-in button perfectly.

The issue is that my ESTOP-out is no-longer tracking either the GUI ESTOP or the hardware ESTOP-in.  It just stays asserted all the time with no change of state.

To be clear, before I attempted to plumb in the hardware ESTOP-in button, my GUI ESTOP was correctly plumbed through to my ESTOP-out so that it would allow me to disable the steppers.

Aside: I have not been able to find any documentation that really explains how ESTOP is expected to work?  I'm unclear on if I should work with estop_latch or not?  The estop_latch documentation is also pretty thin and does not appear to include any examples for either simple or complex configurations?

Here is my current configuration... (generated by pncconf)
Thanks,

B

That is a cool trick! Glad to know about Show Hal Configuration. Thanks.

I am using the Axis GUI.

When I watch the ESTOP-ext variable (is variable the correct term here?) I see that it changes state when I toggle the hardware ESTOP. Hal Show Configuration shows yellow when the hardware ESTOP is not active and red when the hardware ESTOP is active. That seems right to me.

When I watch ESTOP-out, it just stays red the whole time regardless of what I do in hardware or software. This seems wrong to me, but it is also consistent with the problem I am having.

Billy wrote:

Aside: I have not been able to find any documentation that really explains how ESTOP is expected to work?  I'm unclear on if I should work with estop_latch or not?  The estop_latch documentation is also pretty thin and does not appear to include any examples for either simple or complex configurations?
 

Yes, you should use estop-latch if you have an external estop button (or a safety relay but that's another story). Its pretty well described in the man page and there was a hal tutorial about it on the forum.

Actually, to be compliant for safety, a machine cannot depend on LinuxCNC to manage the estop. The compliant external safety circuit should be advising LinuxCNC when the an environment is safe via the iocontrol pins. That's how I have set up my last few machines.

And yes, your 15 tonne machine is a perfect case to adopt modern hardware. The servo drive needs to have a Safe Torque Off  (STO) input triggered by the redundant safety circuit so the servos are safely brought to a halt. In an ethercat environment, there is specific modules to manage this. To your Beckhoff EK1100 coupler you need to add a EL6900 safety PLC and a EL1904 that contains 4 compliant safe inputs. In the absence of STO, You can place two force guided relays in series to trigger the normal servo enable/disable. Siemens make  a 5 pole force guided relay and two of these would allow compliant shutdown of 5 servos (4 axis plus spindle) But you  would also need 2 more single pole force guided relays to fire the signal back to LinuxCNC's iocontrol to tell the controller an estop situation existed.

I'm sorry Tommy, but none of those 20 machines you retrofitted are compliant in today's safety concious world.

Rod, I agree with Tommy, the Linuxcnc E-stop Latch component isn't necessary for estop compliance. Linuxcnc doesn't even need to know about the hard E-stop loop for a machine to be compliant. The soft E-stops in Linuxcnc really have nothing to do with it, and shouldn't (I'm not even sure that it even can) be counted on if you want a compliant E-stop system. It is how the hard E-stop is implemented that makes a machine's E-stop system compliant.

Billy wrote:

The issue is that my ESTOP-out is no-longer tracking either the GUI ESTOP or the hardware ESTOP-in.  It just stays asserted all the time with no change of state.
 

I don't have a 7i96s, but out of curiosity I checked it on my 7i92...

You can make two types of mistakes:
1. Electric.
2. Logical.
It looks like you did both.
First, about the electric one. It is true that there are inputs that detect all possible states (high, low and high impedance), but I have only seen such inputs in logic states analyzers. Typically, the input detects only one state or no this state. For example, shorting the input to ground causes the output to change to the opposite state, but it does not matter whether we short the input to Vcc or leave it unconnected anywhere. In this case, if the output of the previous gate is high, closing or opening the serial switch does not change anything because the high state and the high impedance state are interpreted the same.
Now about the logical one. By pressing the serial switch we do not change the input state, we only cut off the signal. And this is not the same as what I explained above.

Your configuration either doesn't work at all, or works in such a way that it cuts off the iocontrol.0.user-enable-out signal. This means that the serial switch does not turn off the machine, it only prevents it from being turned on/off via an icon in the GUI, which is definitely not what you wanted.

Can this be fixed?
Probably yes, but it makes no sense because you have a ready-made and 100% reliable solution - the estop-latch component. A switch between the input and ground is connected there, so that each time it is switched, the input status changes.

This is from my config:
And now a few words about the emergency stop switch. As the name suggests, it is used in an emergency situation when we want to stop the machine immediately to prevent it from causing damage.
Sometimes the emergency switch must also disconnect or enable something (for example, turn off the motors and apply the brakes), so simply connecting it to the control program may be insufficient or unreliable (the program may hang or respond with a delay).
You have to remember this, and the bigger the machine, the more seriously that should be treated...