High amount of noise in thermocouple reading

[Kavanakt]

I would say that any additional filtering would be unnecessary. First, additional filtering requires you to take your TC+ and TC- and create a filter off of these.  This adds an additional junction that will offset your TC measurements which then you would need to compensate for the new junction you have created.  If the ambient temperature of your junction and the ambient temperature of the 9213 junction is the same then compensation is not needed.  If they vary at all then compensation is needed to give you accurate measurements.  The other thing, the A/D on the 9213 is a sigma delta.  By nature of this A/D a filter is created removing all frequencies above 1/2 of your sample rate.  By the time you get to 300-500kHz this signal has now been attenuated to the noise floor of the A/D.  Also, there is a differential filter stage before the A/D which provides additional filtering.  A shot in the dark, I am assuming the bandwidth of the differential amplifier providing the pre-A/D filtering is below this 300-500kHz pulses of the probe (this is speculation, but if this is true then essentially another pole has been added to this circuit making it even less likely that additional filtering would be needed.)  Outlaw is most likely more on track of what is going on.  How are you using the COM of the 9213?  Are you using any other channels on the device or just the one?  Also, is the liquid you are submerging the TC and RF probe in conductive? Proper grounding and shielding techniques should be followed as explained in the white paper Outlaw pointed to.  A quick test to try, take the TC out of the liquid and the probe out of the liquid. Then place the two close to each other and turn on the probe. You should be able to inductively heat the TC without introducing this erratic behavior. If the erratic behavior does show up, then you are in fact picking up the RF somehow.  If the erratic behavior does not show up and the TC works as expected then this points to the setup as being the issue. Hope this helps.

- Kyle     

[Henrik_Volkers]

---

Jeff·Þ·Bohrer wrote:.  You need those wires very well shielded to prevent inducing some rather strange currents. 

---

Common shielding is good against electric fields, but fails for magnetic fields. Good magnetic shielding is not wise near induction heaters, unless you want to heat it 😉

BUT: Cable based temperature measurements are made in induction heaters:  You just have to know how the magnetic field 'looks' like, and provide a wirering along the field lines. 

[Jeffrey_Zola]

Jeff, Lynn, Henrik- I applaud your your use of physical principles!!!!

Jeff

[Peaches0204]

Given the nature of the electrode, we cannot take it out of the liquid to perform that test. It is immersed in a 0.9% saline solution as to mimic the desired environment. I know this poses a problem in that it is now a conductive solution. I can try the TC outside of the liquid with the electrode submersed but as I said unfortunately cannot remove the electrode with it activated. The feeling that I am getting is that there is some amount of voltage being transferred to the TC which is offsetting the reading since the TC reads on the microvolt level. That is why I have been leaning towards external filtering to solve our problem, try to eliminate this problem of excess voltage (if that is in fact the problem)

[johnsold]

It sounds like you need a shielded, ungrounded thermocouple. The TC junction is inside a sheath but insulated from it. The sheath is connected to the shield which is grounded at the signal conditioner. The shield blocks the interference from the heater while allowing the thermocouple to make the temperature measurements.

The details depend on the physical size of the electrode, the quantity of saline solution, whether it is stirred or not, the proximity, power, and frequency of the heater, the required response time, and probably a few dozen other factors.  Omega (among others) sells several types of shielded thermocouple wire which might be suitable.

Lynn

[Peaches0204]

---

@johnsold wrote:

It sounds like you need a shielded, ungrounded thermocouple. The TC junction is inside a sheath but insulated from it. The sheath is connected to the shield which is grounded at the signal conditioner. The shield blocks the interference from the heater while allowing the thermocouple to make the temperature measurements.

The details depend on the physical size of the electrode, the quantity of saline solution, whether it is stirred or not, the proximity, power, and frequency of the heater, the required response time, and probably a few dozen other factors.  Omega (among others) sells several types of shielded thermocouple wire which might be suitable.

Lynn

---

Lynn, 

We are going to order a shielded TC in order to try to see if that solves our problems. Secondly as to some of the specifics you are talking the output voltage of the electrode is variable has a 340V RMS output on the setting we are using though it is variable to almost any voltage output. The saline solution will be a constant flow of new saline and the proximity is going to vary as we are attempting to use as many TCs as possible. 

[camerond]

Sorry I'm coming in late here, but there seems to be one piece of information missing. Have you scoped your TC? If the ridiculous reading is due to noise from the RF probe, you would be able to see something at high frequency on a scope (be careful to prevent the scope leads from acting like their own antenna). If there's no RF-type noise coming into the scope, it's not RF from the probe that's causing your problem.

If there is RF-type noise, try the TC inside and out of the water. If it works normally in air, then all you have to do is insulate it. How fast does it have to respond, i.e., can you get by with placing it in a capillary which is then placed in the tank? If that is too slow, then there are readily available coatings with names like "Liquid Electrical Tape" which stick well after applied and also insulate electrically (some even come as a spray so a very thin coating can be applied).

If it doesn't work normally in the air, then, as Outlaw said, you most likely have a ground loop in your physical circuitry. At this point, LabVIEW isn't going to help you, you're going to have to track it down yourself.

Good luck,

Cameron

[Peaches0204]

The TC works fine in air. And it works fine when submersed within the saline by itself. It is only upon the introduction of the electrode that things go hay wire. We are going to scope it now and I will be back with the results soon.

[camerond]

Do you mean when you first put the RF probe in the water, or when you turn it on? Since you've said "introduce" twice, it's unclear to me which you mean.

Cameron

[johnsold]

Consider that your TC voltage will be less than 5.269 mV (Type J at 100 C). So that signal is ~96 dB weaker than the 340 V electrode signal which is the interference source. That means that the total of EVERYTHING you do separate the TC signal from the interference must add up to much more than 96 dB of attenuation, which is approximately the dynamic range of a 16-bit A/D converter.

You need to make sure that the voltages at the thermocouple inputs (1) do not exceed the specified common mode range (Channel to COM = +/- 1.2 V) and (2) do not exceed the signal input range (voltage measurement range = +/- 78.125 mV). These are hardware limits. Exceeding the first may result in damage and exceeding either will prevent valid measurements. If the voltages at the inputs exceed these limits there is nothing you can do in software to recover the data.

If the coupling between the heater and the TC is primarily capacitive, a simple aluminum foil wrapped shield will probably reduce the interference sufficiently. If the coupling is inductive, then shielding alone will not do a lot. The ferrite cores will help a lot. The optimum selection of core material, size, number of turns, and so on can be quite complex, but finding a combination which will provide significant attenaution should not be too hard. Also keeping the TC wiring as far as possible fomr the heater wires and at right angles if possible will be very helpful.

Lynn