I am having a variety of problems sensing process temperatures accurately in an ice rink application including temperature drift and cross talk between channels.
First about the system: I have a SNAP-AIRTD-8U with 4 100ohm 3 wire PT sensors attached. The sensor leads are soldered to 3 wire, 16 awg unshielded cables. The cable lengths are 160 ft, 100 ft, 100 ft and 5 ft. The cable leads connect to a screw terminal strip and then 20 awg wires, about 18 inches long, connect from the terminal strip to the SNAP-AIRTD-8U. These are stranded wire with the ends soldered where they insert in the terminal strip and the SNAP-AIRTD-8U.
I need this system to repeatably measure temperatures around 20F (20F =/- 10F). The absolute accuracy is not as important at the repeatability, meaning that if a sensor reports 24F for the conditions, I need it to always report 24F for the same conditions, even if a mercury thermometer would give a different value.
Here are the problems.
moving the cables changes the reading
The cables run from my machine room to points on the refrigeration system. For half the run, they run next to each other before splitting up to their final destinations. If I grab a cable in the middle of the run and move it a few feet, the temperature reading will immediately change a few degrees and mostly stay at the new value.
abrupt changes on one sensor changes all the readings.
One of the sensors measures outside air temp (OAT). I first noticed that all the readings seemed to drift with time of day so I tried an experiment. When I put the OAT sensor in a cup of ice water, the reported temp drops quickly as expected but the other three readings also immediately drop a few degrees. Stirring my coffee with the outside air sensor has the opposite affect on all the temperature readings.
cable temperature sensitivity
Also as an experiment, I collected up some of the snow that we have a lot of at the rink and covered about 10 feet of the cable bundle. This caused a 6F drop in all the readings. When the snow melted, readings returned to the former values.
I clearly have some design flaw in my system. I am on the forum hoping someone can point me in the right direction to fix this.
Thanks in advance and I am sorry if this question has been asked and answered before. I searched and couldn’t find it.
Rob,
Remember that RTD is simply a resistance reading. Therefore, doesn’t take much to upset that because the energy to measure that resistance is very small. First of all, make sure that the junctions are not creating tiny outputs due to dissimilar metals as TC junctions. Next make sure that all connections are tight and not causing tiny voltage drops.
Are they commonned together in any way? If so, change that to all isolated.
The use of unshielded cables for such long runs is a bad idea. These are very low energy. Use 18 - 22 awg shielded twisted pair cables with drain, I prefer 20 awg. The current flow of RTDs is probably in the uamp range. Make absolutely sure you ground the drain at the control panel only. Make absolutely sure you have the panel grounded to a quality ground source. I would use those silicone filled crimp connectors (typically used on telephone lines) to connect the RTD leads to the cable. They are supposed to be as good as soldered.
Make sure that any VFD drives in or near the system have outputs that are run in either VFD shielded cable or metal conduit (not PVC). Make sure that these VFDs have filters on the 3 phase input lines and filters on the output lines if the lead length to the motors are at or over 100 feet. Make sure that the VFD control panels and motors have serious quality ground systems. If any of these exist, that’s your problem.
I cant stress grounding and shielding enough. If you are getting high amplitude low freq (< 1 meg) emi, the cable shielding is not enough, the cables then need to be in metal conduit that is bonded to a ground.
Shielding and metal conduit is next to worthless if it is not bonded to a quality ground source.
VFDs, Inverters of any kind, servo drives are all large RF transmitters in freq ranges from 2 khz to 1 meg. The longer the unshielded output leads are, the more effective the transmitting antenna they have and the same goes for the length of the instrument leads as receiving antennas.
the thin foil on instrument cable shielding is intended for 1 meg + only, and has minimal effect on freq below 1 meg. Only mass will provide the shielding against these lower frequencies such as a steel or aluminum control panel or metal conduit, Liquitite flex with metal, etc. These all have to be bonded to a quality ground.
Thanks so much, Barrett. I’ll get on the shielding and bonding on the leads. I’m curious to see if this also cures the crosstalk between the different sensors. It seems like that is something else.
I really doubt it is cross talk, probably seems like it, but these signal levels are soooo low, that I’m having a hard time envisioning there is enough field strength to induce noise from one to the other. On the other hand, individual shielded cables would eliminate that possibility.
