Effect on CJC correction of using copper instead of thermocouple-type extension wires

[ddml]

Hello folks. This is is a sort of "phenomenological" question, lumping together a bunch of real-world issues. I'm kind of hoping someone has had experience trying this and can give me an opinion.

I am seeking to make relatively high speed temperature measurements with very fine-wire type K or N thermocouples.

I will have very many TC's - around 75 in qty, and hence will, if it works out well enough, wire them single-ended (trading accuracy for quantity considering the two SCB-68's I have).

My complete DAQ system is:

PXI computer

PXI-6225

(2 qty) SCB-68

QUESTION 1: Any recommendation in configuring these either NRSE or GRSE? The device I am near (but not electrically in contact with) is completely ungrounded (combustion experiment). The fine-wire TC's are indeed very fine, and their unshielded lengths will be approximately 12". There is no significant EMF/EMI around my experiment. I will be experimenting but just wanted to know if anyone can rule something out here... 

QUESTION 2:

Important points:

•  I want to use N type for their additional stability over K's at my temperature range, and also because the N-type are available in smaller gauge (0.003") and butt welded.
• Unfortunately, I don' t own a pile of N-type connectors - If I need to use/buy N-type connectors this is a substantial cost (show stopper).
• My configuration has the TC's in physical location "groups" of 7. Considering the single-ended configuration, I'd have 7 anodes and 1 single "ground" for each "group" - for a set of 8 wires coming off of each "group"
• I need to connect on the order of 80 wires into the two SCB-68's. I'm trying to minimize the number of wires I need to deal with.
• Omega does not sell single conductor anode and cathode wire for the N-type as they do for K-type. They only sell duplex wire.
• Omega also does not sell shielded wire bundles in N-type like they do in K-type
• If I use N-type TC's and I use actual N-type extension wire, it's going to have to be duplex (because of the previous two points) - that's something like 80 PAIRS of wiring, with one of cathode wire on most of the wires unused (due to the single-ended configuration).

It seems I have to consider the following tradeoff:

Either use the desired N-type TC's and deal with ~80 pairs of duplex N-type extension wire, OR use the less desired K-type TC's and purchase bundled, shielded wiring to minimize the wiring mess OR use copper extension wire (WHAT?)...

Yes! Why can't I use copper extension wires to the N-type fine wire TC's, and connect to the TC's with some sort of non-metallic connector (such as nylon screws), or solder? The attraction here is that I could use bundled, shielded 8-conductor signal wire to each of my TC "groups". This would reduce and clean up my wiring and not require me to purchase any N-type connectors (TC's would be hardwired, no connectors involved).

It seems to me I could simply configure the cold junction correction (CJC) onboard the SCB-68 to work on N-type. Copper wire would lead from the SCB-68 to the N-type TC's. No metallic connectors would be involved. Thusly, there would be only be the copper/OMEGA-P and copper/OMEGA-N junctions to deal with in CJC. This sounds like it would be the exact same thing as wiring an N-type directly into the SCB-68 except that there may be error between the temperature read at the onboard CJC and that of the actual copper/OMEGA alloy junction. This won't pose significant error because the two locations will be very close in temperature.

If that temperature "error" becomes a concern I can mount a single TC at the location of the copper/TC junctions and make my own corrections (I think...)

Here's a little text schematic (without considering any shunt connections in the RSE circuits...):

Any anode circuit/channel: SCB-68 terminal block -> copper extension wire -> (direct connection via non-metallic connector) OMEGA-P of the N-type TC

Common ground circuit/channel: SCB-68 terminal block -> copper extension wire -> (direct connection via non-metallic connector) OMEGA-N of the N-type TC

This idea assumes that the circuit board within the SCB-68 has copper conductors, such that the addition of copper extension wire would not create an additional junction.

Has anybody tried this? Can anyone offer me some ideas?

THANK YOU in advance!

-Dan

[Henrik_Volkers]

TC measurements are noisy and with low voltages (only 30-40µV/K), so if you want to make fast measurements without long timed lowpass filter you have to keep the signal as clean as possible! Keep the signals separate, differential, shielded and twisted paired as long as possible!   

You didn't mentioned the cable lengh, and if there are restrictions on cable types you are only allowed to use... (CAT9 cable is cheap and each of the four twisted pairs is shielded)

Combustion measurements... so we talk about higher temp ... >800°C  -> uncertainty of about 10K (2K for the setup and 1% minimum for the TC) (plus some more due to ionisation effects in the flame;) )

So I won't bother about the type N connectors and use shielded Type N extension cable to the SCBs . If you don't want to mess around 40 pairs in the SCB do the shield grounding and summing to one Type N neg wire outside, however take into account that all currents in all wires sum through that little wire and together with the wire resistance  produce error voltages. You can use a short copper wire if you can life with the additional error that will come due to temperature mismatch between that summing point and the CJC tempsensor inside the SCB.

Connections:small wire to extension: As long as one pair is at the same temperature I would crimp them together. But I also just twisted them together and sometimes even put a drop of solder on it. No problem if you don't have temperature gradients ....

You can build your own CJC terminals near the DUT and use copper (and only one TC or another temp sensor NTC,RTD)  to the SCB and do the corrections in LabVIEW but that will need a good understanding on how thermocouples work.  Another trick here would work directly, but is prone to noise....   time for a coffee 😆

And as usual: 

A good source for TC knowledge:
Manual on the use of thermocouples in temperature measurement,
ASTM PCN: 28-012093-40,
ISBN 0-8031-1466-4

255 pages dedicated to TC

(Page1): 'Regardless of how many facts are presented herein and regardless of the percentage retained,

                all will be for naught unless one simple important fact is kept firmly in mind.

                The thermocouple reports only what it "feels." This may or may not the temperature of interest'

Message Edited by Henrik Volkers on 05-06-2010 10:01 AM

[ddml]

Henrik -

Thank you for the very useful reply. Yes, I neglected to mention that extension wire lengths will be on the order of 1m, so extension resistance should be low, even if common ground is used as part of a single-ended circuit arrangement.

I am am trying to spatially and temporally resolve a reacting flow field (the flaming region, including 2ndary air flow etc) for validation of a CFD model! This is exciting stuff since I haven't done it before and don't know any better :) Yes, I have read a lot of papers on the topic but this is my first attempt at essentially instrumenting a fire.

My actual DUT application is a biomass cook stove. I am instrumenting the combustion chamber basically above the fuel bed. The TC's will be arranged in rakes suspended across the "chimney" at incremenal heights. The mount apparatus at each height location in the chimney is basically a ring-shaped circuit board (without copper), with the TC's being mounted through holes in the board, stretched across the middle. The chimney is arranged in sections. It's kind of tedious but easy to disassemble, etc. (Sure it would be nice to have some optical diagnostics here!)

Before I build the device for my actual experiment of interest, I will conduct a simple test to evaluate some stuff about the circuitry. I will arrange a number of thermocouples as close as possible to each other (spaced < 2mm) in an unsteady, "puffing" flame of a non-premixed gas burner (orifice 0.1m). The resulting flame "puffs" are nice and large and cleanly-defined so at such thermocouple spacing the thermocouples should be exposed to reasonably consistent heat fluxes (function of time). Part of the experiment is to evaluate the response time differences (Yes, a huge topic I know! Not talking about that here!), and part to evaluate the affect the circuit arrangement has on the signal. I plan to wire sets of two (for redundancy) 0.003" fine wire N-type TC's arranged in differential configuration next to two of the same type of TC arranged in NRSE or GRSE configuration. Picture four fine-gage TC's with their junctions within a ~4mm zone. Assuming they are all exposed to essentially the same transient heat flux, I can look at the transient response through the proposed circuitry, and try to make sense of any differences. If by chance all signals are quite comparable then I would prefer single-ended circuitry, choosing to accept XXXX-amount of error (from signal noise) in exchange to receiving almost twice as many inputs.

As I indicated, my limitations are funding, "clutter"/complication, and the other practical issues I noted about what wires are and aren't available.

Here's my latest logic !

• If my experiment shows that the single-ended NRSE or GRSE arrangement response "closely matches" the conservative differential response, I will choose single-ended to get more sample locations.
• Since I know the extension wire / TC junction at the DUT will be almost isothermal with the SCB itself, I would love to use copper extensions, and accept (and possibly correct for?) the temperature-based error. I would use wire such as this (recalling I need 7+ and 1 ground per grouping of 7 TC's at my DUT) -> http://www.newark.com/jsp/content/printCatalog.jsp?cat=c127&page=1199&display=zoom
• If in examining the previous point I find that copper extensions are NOT practical, and I need to use shielded TC-metal extensions the whole way, then I will select a K-type TC over an N-type. The reason for this is that there are many more options for shielded, bundled K-type extension wires (N-type is apparently kind of new and not very popular). However if copper extensions are feasible, then I will use the N-type TC's, motivated by their superior high-temp oxidizing atmosphere stability, and the fact I can get them in a butt-welded configuration at 0.003" gage!

If any of this logic seems faulty or unrealistic I would really appreciate any more input! I do know this will be a lot of work!

Thank you once again,

-Dan

[Henrik_Volkers · Accepted solution]

---

ddml wrote:

Henrik -

Thank you for the very useful reply. Yes, I neglected to mention that extension wire lengths will be on the order of 1m, so extension resistance should be low, even if common ground is used as part of a single-ended circuit arrangement.

...

Here's my latest logic !

• If my experiment shows that the single-ended NRSE or GRSE arrangement response "closely matches" the conservative differential response, I will choose single-ended to get more sample locations.
• Since I know the extension wire / TC junction at the DUT will be almost isothermal with the SCB itself, I would love to use copper extensions, and accept (and possibly correct for?) the temperature-based error. I would use wire such as this (recalling I need 7+ and 1 ground per grouping of 7 TC's at my DUT) -> http://www.newark.com/jsp/content/printCatalog.jsp?cat=c127&page=1199&display=zoom
• If in examining the previous point I find that copper extensions are NOT practical, and I need to use shielded TC-metal extensions the whole way, then I will select a K-type TC over an N-type. The reason for this is that there are many more options for shielded, bundled K-type extension wires (N-type is apparently kind of new and not very popular). However if copper extensions are feasible, then I will use the N-type TC's, motivated by their superior high-temp oxidizing atmosphere stability, and the fact I can get them in a butt-welded configuration at 0.003" gage!

If any of this logic seems faulty or unrealistic I would really appreciate any more input! I do know this will be a lot of work!

Thank you once again,

-Dan

---

Dan,

I marked the important sentence 😉  what you actually do is shifting the CJC from the SCB to the TC/copper connection. Every temperature difference will directly add to the error.

However don't try to save copper with your 7+1 approach , ask the IT guys for another 12m more Ethernetcable and go with 7*2 plus one pair reserve and do the summing at or near the SCB. IF you will need a differntial measument setup later, you will have to rebuild all connections. (And don't just add a new cable for all the ground connections, it's the twisted pair that does the magic 😉  Yes, even vor 1m! Remember you are dealing with µV! And since you want fast readings without a lot mean filtering, don't try to save $10 for about 12m cable. Somehow I have the feeling you will try it anyway 😉

[ddml]

Henrik -

Vielen dank! This is exactly the type of experience I was hoping to tap into. 

As this is not my usual area of work I am having to come up to speed on the details of noise induction and reduction, etc. I will try to get some bundled cable featuring 7(+) x 2 twisted, shielded pairs, performing the ground summing at the SCB. I think this will work well.

In case you are curious about the result I will post back to this thread with some of my findings.

Cheers,

-Dan