SM-STAHL CNC 8 axis robot losing all home positions when powered down

Assuming with home positions you do not actually mean the KSS Home Positions but the mastering positions: do not only check the batteries for your PLC. Batteries for your PLC do not influence how the robot controller cabinet of the KRC behaves on shutdown. If the batteries of the KRC control cabinet are down you loose mastering after every boot. Replace batteries with good ones.

Usually if batteries are down the KRC immediately shuts down when turning the off switch. If they are ok it should take a bit until KRC is completely shutdown since it is buffering stuff like your mastering.

Very bad situation. We had 3 SM STAHLS in our facility years ago… which model do you have. I think I have some spares in my stock if you will need.

Does robot move in some direction? Does robot arm behavior is strange (I mean if arm goes in some directions by itself)?

As I remember, you can manage something in Q parameters, also check batteries( depending on which type of machine you have).
Better to connect tour computer to machine.

Quote from Tiamat067

What happens is whenever they cycle the cell, or randomly it seems, the robot will lose reference positions.

there are different types of encoders. most robots use pulse counting (quadrature encoder) or resolvers (sort of rotating transformer). few use true absolute encoders (they know position any time, regardless if axis is moved by servo or external force while servo was switched off).

to compensate for this, it is common that non-absolute feedback types store information somehow.

on axes with pulse encoders, counters are keeping track of position, so even when robot is powered off, there must be battery to keep counters alive.

most common problem here is batteries failing or accidentally disconnecting them (or power).

resolvers are absolute feedback but only for rotation that covers single motor pole. most motors are 3 or 4 pole. on a 3-pole motor, resolver is absolute for 120deg of motor rotation. on a 4-pole motor, it is 90deg. so if external force somehow makes motor shaft turn precisely 120 or 90 deg, robot will not know that position was changed. fortunately such exact movement is highly unlikely so changes are easily detected by comparing to previous value that was stored somewhere.

but... if data is also lost while robot is powered up, something else is going on:

1. broken wire or oxidized connector contacts.
2. strong interference from some source of noise (or defect in the robot itself)
3. lost encoder shielding ground connection
4. ground loop when encoder shielding is grounded in multiple places. if installation is purposely done with shield grounded at both ends, then shield itself must not carry significant current (add external GND conductor with large wire size). this is common on robots used for welding, plasma cutting etc where strong and high frequency interference is present.
5. worn out cable. when flexed for too long (or too much) strands in cable may brake. flexible wires (and shield) are always multistrand so single break is not an issue.... unless... the end of the broken strand is digging into insulation. this becomes easier with higher temperatures (insulation becomes softer) and faster motions (inertia of the cables causes increased wear). so what happens when one of the demon strands pearces insulation and touches GND or another encoder wire? yes, it would be pretty much like what you are experiencing. to troubleshoot this one would need to replace encoder cable and run robot to see if issue persists (at least temporarily, does not need to be pretty, can be hanging ). the other test is insulation test but you would want that done carefully by someone that knows what they do so you do not fry anything, because insulation test is done with high voltage (500V, 1000V, ...)

etc.

note:

if only one encoder is disconnected, only that axis should be affected.

but if multiple encoders loose positions simultaneously, then it must be something that is common to all of them... an example of this is a loss of supply power to encoders or their logic boards (counters/RDC/whatever). so perhaps replace batteries by some temporary source (must be suitable voltage) such as another battery or even power supply. intermittent power loss in a circuit can be monitored using suitable devices (power analysers, loggers, electronic version of trigger latches etc.)

if you have more robots of same type, you can try swapping parts until you have narrowed it down. good luck...