Hi,

We have strange temperature peak on a laser cutting machine chiller that trigger an emergency stop too often in our taste. Been searching for the problem for years.

I finally had the time tu put the scope on it and was expecting to see a nice clean dc voltage of lets say 5 volt on one side, and one another clean trace with a lower voltage correcponding to the voltage drop across the PT1000resistor. From that i was expecting to let the scope running a little bit to catch spike that i could measure or a specific frequency induced like 16khz coming from the switching signal of a vfd or something like that.

But instead i got one hell of a mess at the 1sec scale, and when i fine tuned on the signal i seen 95khz spike and i dont understand what i see.

This gave me the idea of plugging the scope on my multimeter probe to see if i could catch similar spike when the DMM is measure a random resistance. I seen somewhat more spaced but 85khz spike similar to the one from the plc.

So i need to understand what i'm seeing in order to be able to make something out of that signal. So please educate myself on that matter.

Also if someone could point me out a way to test that signal to be able in find what is interacting with it i would appreciate.

I dont wrap my mind about why a commercial company would use 300mV for a temps probe in a chiller full of 600V high power compressor/fan that produce lot of EMI... It seems so fragile to EMI...

Thank you and have a nice day!

 

We use PT type temp sensors in wafer furnaces that can melt quartz with fractional C temperature stability. Using a single ended scope probe on a PT sensor in a circuit that doesn't use a simple voltage divider usually results in seeing all sorts of hash. There are more complex 2 3 4 wire PT1000 circuits that cancel external noise and hash

Can you post the PT1000 sensor circuit?

 

Sure the scope trace looks ghastly, but remember that temps change relatively slowly, all that nasty EMI is at frequencies many orders higher than the temp signal.
Typical temp sensing circuits have low-pass filters to make this all go away.

 

The actual signal probably is not referenced to ground like your scope measurements are. It is probably a differential signal which is taken across a sensing resistor or resistor bridge, and reflects current through the RTD rather than voltage between either wire and ground.

You can see the same noise present on both wires. If your scope has a subtraction function, subtracting green from red should show you a much more stable signal that is closer to what your controller is seeing. Or if your scope is usb-powered and not ground-referenced, you can take just one measurement across the two wires with one probe; ground clip on one and probe on the other. Just make sure your laptop is not plugged in, not resting on grounded metal, and USB scope box not on grounded metal.

 

We use PT type temp sensors in wafer furnaces that can melt quartz with fractional C temperature stability. Using a single ended scope probe on a PT sensor in a circuit that doesn't use a simple voltage divider usually results in seeing all sorts of hash. There are more complex 2 3 4 wire PT1000 circuits that cancel external noise and hash

Can you post the PT1000 sensor circuit?

It's a 2 wire pt1000 that goes into a water tank. The temps can stay stable at the 0.3 farenheit at like 63 for a while then jump to 75-76 for a seconds and go back to its original value. as the water cannot change temperature fast like that where the sensor is situated it's clearly electric. We found some answers to the problem, we had a fan blowing cold air straight to the condensor and as the system monitor the condensor pressure it was waiting till the last moment to start the compressor to cool the water. That combined with the random jump in temperature lead to the problem we actually have, that the temperature goes past the alarm point for a sec and it send the emergency shutdown signal to the whole machine... I wrapped the wire 4 time into a toroid core in hope that it will resist the change and absorb a part of the spike cause that, it helped a little but not enough. The kind of toroid i used best absorb past 30mhz and i dont think the spike that cause that is cause by that high of a frequency... I will need to eventually upgrade the wire to a shilded one for sure but i try to find the root cause to eliminate it.

 

Sure the scope trace looks ghastly, but remember that temps change relatively slowly, all that nasty EMI is at frequencies many orders higher than the temp signal.
Typical temp sensing circuits have low-pass filters to make this all go away.

Alright good to know for the low pass filter on the plc thx!

 

The actual signal probably is not referenced to ground like your scope measurements are. It is probably a differential signal which is taken across a sensing resistor or resistor bridge, and reflects current through the RTD rather than voltage between either wire and ground.

You can see the same noise present on both wires. If your scope has a subtraction function, subtracting green from red should show you a much more stable signal that is closer to what your controller is seeing. Or if your scope is usb-powered and not ground-referenced, you can take just one measurement across the two wires with one probe; ground clip on one and probe on the other. Just make sure your laptop is not plugged in, not resting on grounded metal, and USB scope box not on grounded metal.

I tried measuring like you said one time but there is so much 60hz coming from the machine that you lose the signal. Something around 80+ volts of 60hz AC going straight in the scope probes... Like at my working table when i just work on my big stainless benchtop there is like 80+ volts picking up on the probe...
I tried the substration on the signal i saved yesterday and it worked really good thank you. I will have to try it live on the machine to see if i'm able to pickup the spike now that the signal is a lot better. It will have to wait for the machine to run while plugged to identify when the specific subsystem start and i have a peak.
I joined a picture of yesterday signal cleaned with the substraction.
If you have some other idea let me know i appreciate the help!
Have a good day!