5069-OBV8S and Safe Torque Off

[ASF]

Hi all,

Just a heads up on an issue that a colleague of mine was unfortunate enough to stumble upon recently.

The new 5380 Compact GuardLogix safety outputs (5069-OBV8S) have a no load detection "feature" which can't be turned off. Safe torque off is not enough of a "load" to avoid faulting the outputs on no load. So, you can't wire directly from these safety outputs to STO without something else in parallel to act as a load (the official RA "solution" is a 2.4k resistor).

I would imagine that this would also apply to the 1756-OBV8S and the new Flex 5000 safety outputs (excluding the safety relay outputs).

Enormous oversight by RA I have to say. Hopefully there'll be a firmware fix in the pipeline, but as usual, so far nobody knows anything.

If you want to see the super-detailed official response from RA:
Technote 1091239 (Access: Everyone)

 

[AustralIan]

Sounds like this problem should have been OBV[iou]S at module design time.

[Yes, yes it did sound funny in my head]

 

[ASF]

I would have thought so too (on both counts)

 

[Jeev]

Good to know!

the official RA "solution" is a 2.4k resistor

Welcome to technology in 2019, folks.

Enormous oversight by RA I have to say. Hopefully there'll be a firmware fix in the pipeline, but as usual, so far nobody knows anything.

Either that or they want you to buy their ERS2/ERS3/PF527 models that have CIP Safety.... And cost more.

Sounds like this problem should have been OBV[iou]S at module design time.

I approve of this humour

 

[ASF]

Either that or they want you to buy their ERS2/ERS3/PF527 models that have CIP Safety.... And cost more.

Just you wait, in early 2020 there'll be a Rockwell Cat 4, PLe, SIL3 Safety Rated 2.4k Resistor on sale.

 

[alan_505]

Just you wait, in early 2020 there'll be a Rockwell Cat 4, PLe, SIL3 Safety Rated 2.4k Resistor on sale.

Don't give them ideas like that.

 

[AustralIan]

ASF, have you done the Safety Calculations adding the resistor yet? can you still get similar PL?


What "chance of dangerous failure" did you use for your resistor?


What solution have you used instead of OBV8S + (Drive with high impedence STO)? What PL or SIL was required?

 

[drforsythe]

Thank you for the head's up! My colleague and I just finished two separate designs that utilize 5069-OBV8S and Powerflex drives. We would have gotten hung up for a while, for sure. Thanks again!!

 

[Archie]

I am now starting up a system using the 5069-OBV8S card and ran into another problem created by not being able to turn off this feature. The card does an open wire check using a pulse test. It essentially switches the output off for maybe 500 microseconds every 30-50ms (was trying to judge this on an old oscilloscope). This is independent of configuring the output as a pulse test output, which means even if it is set as a Safety Output, it still does the pulse.

My system is wired directly from the output to an Emergency Stop input on a Universal Robot controller. About every 5-10 seconds the robot detects the pulse and triggers an E-stop condition.

Since this "feature" cannot be disabled, I am unsure of my solution just yet. I suspect I will have to wire some relays to the output and wire the robot to the contacts. The relay should be able to float through the pulse.

To AB's credit I did find mention of this buried in a small foot note in the manual:

“An open wire test and main switch pulse test can also generate a pulse on a safety output even in safety mode.”

 

[janner_10]

Just stick a couple of positive guided relays in between.

 

[Archie]

Just stick a couple of positive guided relays in between.

I was trying to avoid the additional components, but it looks like the OBV8S isn't going to allow it to work without relays or some other components in between.

 

[janner_10]

I was trying to avoid the additional components, but it looks like the OBV8S isn't going to allow it to work without relays or some other components in between.

I have tried and tried, but this was the only way I could get it to work.

 

[Archie]

I have tried and tried, but this was the only way I could get it to work.

I tested this and checked the signal going into the robot on the scope and it prevents the robot from falsely throwing an E-Stop

 

[Jeev]

I tested this and checked the signal going into the robot on the scope and it prevents the robot from falsely throwing an E-Stop

Anyone using this example, please note the value of the capacitor. Capacitance will add to the switch-off delay (Discharge time) of the inputs. If you're doing a validation/verification of the safety system or any reach or safety distance calculations, you will need to factor this latency in.

 

[Archie]

Anyone using this example, please note the value of the capacitor. Capacitance will add to the switch-off delay (Discharge time) of the inputs. If you're doing a validation/verification of the safety system or any reach or safety distance calculations, you will need to factor this latency in.

While determining the size of the capacitor, I took the latency factor in consideration. I started with a 0.05 capacitor and watched on the scope and saw practically no impact. I then tried 0.1 and also near immeasurable impact. Finally using the 1 microfarad I could see the pulse was reduced to only about 1 volt of drop. I did not measure the actual latency time because the threshold voltage of the robot input would have to be known and I would need a good digital scope.

The robot I am using is a slow moving cobot, so I did not feel the need to be concerned about a reaction time in the few millisecond range. I did the non-scientific test of watching the robot pendent while pressing the E-stop to ensure the reaction was in a fraction of a second.

In an application using a high speed robot, the OBV8S card may not be a viable solution because no matter what you use between the card and robot, it must float through the 500+ microsecond pulse of the OBV8S. Relays and contactors float through this because their off response time is typically in the 10-50 ms range.

This does make me curious of the response time of drives with STO inputs. Since AB says only a resistor is needed, that means their response time is greater than the 500 microsecond pulse width of the OBV8S.