PCB Heaven

General Category => Maths, Geometry, Physics and others => Topic started by: seekingknowledge on May 12, 2013, 16:48:46 PM

Title: Thermocouple errors
Post by: seekingknowledge on May 12, 2013, 16:48:46 PM
Please excuse my lack of knowledge/understanding about this topic.

I am an inspector in a small shop that manufactures thermocouple probes. Please examine this situation: We have a fairly standard thermocouple probe, type k, stainless steel sheath mineral insulated cable, with an ungrounded junction, and extension wire (thermocouple grade). However, one of the conductors is accidentally grounded to the sheath where the transition is made between the extension wire and the conductors in the MgO insulated cable.

Obviously this would make it susceptible to electronic noise, resulting in erratic measurements. My question is, in the absence of electrical noise, would the reading be affected?

The explanation from my colleagues is that if we have a probe intednded to have an ungrounded junction, and it is grounded at some point, we can determine the location of the ground by heating the junction and watching the reading. If the reading responds to heat as normal, the junction is grounded to the sheath. If one of the conductor wires is accidentally grounded to the sheath at a point others than the junction, it will produce a slow response (jumping up 4-5 degrees F at a time, instead of 50-60 deg F. It is perceived as a slow response by some, I explain it as an innaccuracy.)

This came up because we had some probes where conductors were making contact with strands of the stainless steel overbraid on the extension wire, making it susceptible to electronic interference. I am told that we can determine if it will be susceptible to noise just by testing in over a heat gun and watching for a "slow response", and watch the behavior of the readings up to high temperature. I question if this is a valid method. I would think that the valid way if these probes are susceptible to interference (or to what degree it is susceptible), is to subject the probe to some simulated electric noise.

Also, if this is the case, what would be a good way to simulate this electrical noise, (of course the noise being greater than that generated by a heat gun used for testing).

I hope this rambling makes some sense to some of you. If any of you electrical engineers could weigh in on this, that would be greatly appreciated.
Title: Re: Thermocouple errors
Post by: kam on May 12, 2013, 18:36:49 PM
The problem for me is double:
First, the thermal problem. If the temperature of the surface that the connection touches is different than the temperature where you do the temperature compensation, then there will be definitely an error reading, even without electrical noise

Second, the electrical problem. If the surface is not electrically isolated from the reading circuit, then it may add noise or alter the very small voltage of the thermocoupler. If any slight voltage goes through the TC wires to the reading circuit, it will certainly alter the reading and may destroy its input as well.
Title: Re: Thermocouple errors
Post by: seekingknowledge on May 13, 2013, 00:01:54 AM
Thank you for your prompt response. So would any amount of grounding to the sheath(say one strand of overbraid vs several strands of overbraid) be enough to make the t/c reading to be altered and render the probe defective?

And would this be observable by subjectively observing the reading when holding it over a heat gun and comparing it to another probe with little control over conditions? I would think that it would be hard to determine accuracy that way. I suggested placing a defective probe and a conforming probe as well as a certified RTD or thermocouple as a control in a certification oven as to supply a controlled environment to give us objective readings. This was dismissed as not being able to tell us anything, but simply watching the temperature increase and "reaction time" would tell us if the probe is faulty.
Title: Re: Thermocouple errors
Post by: seekingknowledge on May 13, 2013, 00:07:23 AM
Basically, if there is any grounding of sheath to conductors at any time, I would consider the probe defective and should be marked as non-conforming.
Title: Re: Thermocouple errors
Post by: kam on May 16, 2013, 10:51:30 AM
There are thermocouples that the junction has no galvanic contact with the sheath. This is the best to use.
Depending on the application, there is always the possibility to get wrong measurements from thermocouples without isolated contacts. There are 3 types, grounded, ungrounded and exposed. What you look for is "ungrounded sheath" to use in noisy environment.