Sensible practice for e-stop?

I've had a bit of a search on the forum and a few posts came up but nothing to really answer my question, which is - what is considered the sensible way to implement e-stops on a Bridgeport-sized machine to provide reasonable protection without making life difficult.

Apart from guarding against finger trouble or senior / human moments, there is the (more remote?) possibility of the hardware or software having a wobbly. That would normally suggest a hardware override. However, it would be very easy to go over the top and end up with all sorts of events shutting the thing down. I've been there in the past with motor / dyno testing so I know you have to get the balance right.

The other input to an e-stop circuit might be the sh*t or bust limit switches. It seems that LinuxCNC doesn't really need limit switches for normal operation, as it can manage fine from the home position. In that case, it seems to me that the limit switches if fitted could connect in to the 7i76 (to protect the machine from me) - or hard wire into the e-stop circuit as a last line of defence.

As well as telling the 7i76 that the e-stop is depressed, I have 3 servo drives and a spindle motor that can be inhibited or stopped very rapidly. Sounds like a sensible plan to do make that happen.

Finally, I would normally expect an e-stop circuit to be latching. Is this naive here?

What's considered a reasonable / pragmatic path here? I'd love to hear any suggestions.

Thanks

A lot of what might be best will depend on the machine and parts in question.

High speed spindles are often best stopped by a controlled shut down of the VFD with braking. (I've seed a spindle coast for hours after a power loss.)

Heavy vertical moving parts of the machine, need to be counterbalanced and or braked to prevent them from crashing down.

These and other things should be considered when planning your safety strategies.

An e-stop switch that breaks a circuit is generally safer than one that has to close one. I've worked on some machines that were set up with normally open e-stops, and it really sucks when you hit the big red button when you need things to stop and nothing happens because the buttons contacts are dirty. (That was a factory designed machine too.)

They are usually a latching circuit with a series of normally closed contacts. Open any of them and you have to re-arm the thing.

The VFD has a braking resistor so could stop pretty quickly. The servos have an enable input but this would de-energise them so they would be free to move (or spin on).

All the machines I've worked with disable the servos and stop the VFD when e-stopped. The VFD usually has a digital input that is dedicated to the e-stop signal, that stops the drive. The nicer machines I use, have a lockout pawl pin that shoots in to lock the Z screw to prevent it from dropping. It is usually a single acting spring loaded air cylinder with position sensors that pops out when air pressure is removed, and when the Z servo is enabled an air solenoid pulls the pin back. If the pin sensors don't sense the correct position, all sorts of alarms are set.

If you connect your limits to the e-stop, or servo enables, be sure to include a way to override and get yourself off of them if something goes wrong. (Especially if the machine can't easily be manually moved.)

I will add that e-stop should be separate from LinuxCNC as far as possible. But it is helpful to tell LinuxCNC that an e-stop has happened to force a re-enable cycle. You probably don't want everything powering back up when you release the e-stop button.

On my Mill I do this with a separate NO contact on the e-stop button as an input to LinuxCNC. But on the lathe I am planning to have LinuxCNC monitor the main coil voltage directly.