Ken Roach
Lifetime Supporting Member + Moderator
Hello, PLCTalk Community, I hope you can help give me a little perspective.
I'm doing an in-depth failure mode analysis on a system that uses a VFD with an ordinary differential line driver quadrature encoder for velocity feedback.
The end customer keeps questioning the reliability of the encoder signal, essentially saying "what happens if the encoder gives you a false reading and you're actually traveling at a different speed than it indicates ?"
I'm trying to convince them that such a failure mode is not a "credible failure mode".
The VFD itself recognizes four failure modes for the encoder:
1. Open Wire
2. Phase Loss
3. Quadrature Loss
4. Peripheral Module Fault
Open-wire detection depends on the differential wiring for the encoder; you have both A and A' signals for Channel A (and likewise for B and Z). If the signals are not opposite one another, you know you have a broken wire.
Phase Loss is a count of Open Wire faults; if more than 30 of them accumulate over 8 milliseconds, the drive declares an encoder phase loss.
Quadrature Loss is when the signal changes on Channel A and Channel B at the same time. Because they are offset by 90 degrees, they should never change simultaneously and if they are doing so you must be seeing a noisy signal.
Peripheral Module fault is a failure of the encoder input module or the serial bus between the drive and the module. That's made up of watchdogs and checksums.
I have been telling them that it's improbable for a noisy signal or other malfunction to create a pulse stream that is not detected by one of those three fault detection methods, but that doesn't satisfy their inquiry.
They want me to attempt to account for other failure modes in which an unforeseen signal failure or firmware glitch causes the encoder to count by multiples or to scale the incoming signal by some unforeseen multiplier or otherwise simply calculate the velocity incorrectly.
The encoder is a Dynapar magnetic ring encoder, so there are no programmable elements inside the encoder itself and a minimum of possible failures of the encoder disc/ring/pickup assembly.
So I guess my real question is: has anyone here ever seen a VFD encoder fail in a way where it gave an inaccurate velocity feedback without also having a bad signal detected ?
I'm doing an in-depth failure mode analysis on a system that uses a VFD with an ordinary differential line driver quadrature encoder for velocity feedback.
The end customer keeps questioning the reliability of the encoder signal, essentially saying "what happens if the encoder gives you a false reading and you're actually traveling at a different speed than it indicates ?"
I'm trying to convince them that such a failure mode is not a "credible failure mode".
The VFD itself recognizes four failure modes for the encoder:
1. Open Wire
2. Phase Loss
3. Quadrature Loss
4. Peripheral Module Fault
Open-wire detection depends on the differential wiring for the encoder; you have both A and A' signals for Channel A (and likewise for B and Z). If the signals are not opposite one another, you know you have a broken wire.
Phase Loss is a count of Open Wire faults; if more than 30 of them accumulate over 8 milliseconds, the drive declares an encoder phase loss.
Quadrature Loss is when the signal changes on Channel A and Channel B at the same time. Because they are offset by 90 degrees, they should never change simultaneously and if they are doing so you must be seeing a noisy signal.
Peripheral Module fault is a failure of the encoder input module or the serial bus between the drive and the module. That's made up of watchdogs and checksums.
I have been telling them that it's improbable for a noisy signal or other malfunction to create a pulse stream that is not detected by one of those three fault detection methods, but that doesn't satisfy their inquiry.
They want me to attempt to account for other failure modes in which an unforeseen signal failure or firmware glitch causes the encoder to count by multiples or to scale the incoming signal by some unforeseen multiplier or otherwise simply calculate the velocity incorrectly.
The encoder is a Dynapar magnetic ring encoder, so there are no programmable elements inside the encoder itself and a minimum of possible failures of the encoder disc/ring/pickup assembly.
So I guess my real question is: has anyone here ever seen a VFD encoder fail in a way where it gave an inaccurate velocity feedback without also having a bad signal detected ?