Thermocouple/34465A temp CAL

[jonpaul]

Bonjour a tous:

Checking the new Keysight  34465A  all   within sec as far as I can see but temperature is the last one.
Testing is only for the Type K thermocouples we use.: Omega, Fluke and others
Using the Omega mini K connectors.
I made a gold plated banana to Type K female adapter/extender with Omega type K TC wire.
Use old Fluke 51 handheld 2 1/2 digit Tep meter for reality checking. (also  Fischer mercury thermometers,  poor resolution).

The various TCs seem to match ~ 0.5..1 deg F.

Procedure:
_____________________

Boiling:

Distilled water, very slow boil in pyrex Erlenmeyer flask, immerse TC to middle of fluid,
Compensate for Baro pressure. CAL  100 C or 212 F.
So far within 1 deg F.
________________________

Freezing: 0 C or 32 F (not yet setup)

Dewar flask,  Pyrex dim 45 mm ID 125 mm L
Add distilled water,  make distilled water ice, add chopped ice,
decant most of the water, mix, allow to stab, immerse TC to middle of mix

Appreciate all comments and improvements from  The CAL Mavens.

Mille mercis,

Jon

[Kleinstein]

The weak point in using a thermocouple with a DMM without a proper connector is the cold side compensation. So the expected error from the meter is mainly an offset. Beside the meter itself the adapter, plugs used also effect the offset. The scale factor part is quite accurate.

On the other side the thermocouple wire even if totally wrong would still give zero voltage at zero temperature difference. With non ideal materials the exact value for the voltage can be off. So the error from the TC used is mainly an error in the scale factor and not in the offset.
TC wires are roughly in the 0.1 % range with the accuracy of the voltage. This way less accurate than the scale factor of the voltage readings. So the scale factor error should be dominated by the wire, at least for a quality DMM (could be a factor with the old Fluke 51).

For testing the TC function a rather long warm up period many be needed, as the temperature difference between the sensor and terminals may need quite some time to stabilize. Compared to a normal measurement the temperaure difference could have more effect.
It may be interesting to measure the drift during warm up for a TC (at constant temperature) to get am idea how long it takes to stablize. This may be different from the normal warm up speed.
For the TC function the back side terminals may behave different.

[jonpaul]

Cher Kleinstein
 
Mille mercis pour votre council rapide!

The Keysight 34465A has a CJ comp at the front terminals only.

connections 

34465A front gold plated banana ><gold plated banana plugs><Omega Type K ext wire><Omega type K mini connector ><Omega Typ K mini TC

Allowing 1 hr to stabilize.

The ancient  Fluke 51 is just for a reality check and has a CJ offset trim on the panel.

I seek your wise experience on ice and boiling water baths for TC testing.

Can you critique /improve the notes in my orig post?

Many thanks again ,

Jon

[Kleinstein]

For the ice bath and boiling water case the heat condduction along the wire is a weak point. So the TC shloud be immersed quite a bit ( e.g. some 5 cm - depends on the diameter and isolation). It somewhat works with bare thermocouple wires in the ice bath, as the conductivity of distilled water is low. There is a possible error source from electrochemical voltage and a current flow through the liquid, if the wires are exposed. So ideally the wires (or at least one of them should be isolated.

For my experience there is no need to drain off water from the ice bath, just starting with ice and water that is no too warm is OK. It helps to give the ice bath time to stabilize (e.g. some 15 min at least). The ice parts can be slightly colder in the center and pockets of water start a bit warmer. To keep the TC about in the center of the ice, it helps to mount it on some fixture.

[TUMEMBER]

Do not use "automatic ambient temperature compensation" on the front panel. A "simulation of an inverse thermoelectric force" equivalent to the measured ambient temperature is then performed. Use the option of selecting the temperature of 0oC. You will benefit from physical holding / generation of 0oC anyway. This is more precisely, you are missing one element that causes additional error. For the reconstruction of 0 you simply "stick" the "hot" end of the thermocouple into the ice next to the weld "compensation wire + copper wire". Of course, no water penetrated when insulated with a heat-shrinkable sleeve, this way you will find faults in the compensating cable and the connectors.
Basically, it is worth doing this for S (PtRh10-Pt) and R (PtRh13-Pt) thermocouples "precious metals". They have a low temperature coefficient.
Base metal thermocouples have this factor an order of magnitude greater, so a pair of microvolts does not change the result dramatically.
Pay attention to use compensation cables (made of the same material as the thermocouple) and not expansion cables (cheaper and worse class).
*********************************
It is important to drain the water from the ice mixture. The highest density (specific gravity) of water is at + 4oC. It flows to the bottom of the tank and collects there, and in an "unfortunate situation" the joint of the compensating pipe may be there. It is not dangerous for the "full volume of ice in the tank" but the ice cannot "float" in the water. Ice should only be slightly moistened with what will melt from it. It is recommended to "siphon" continuous drainage using a thin tube brought to the bottom of the tank and "threaded" over the edge of the tank below the liquid level. Capillarity phenomenon.

[mendip_discovery]

I have a k-type to the banana connector, https://cpc.farnell.com/tenma/72-8368/adaptor-temperature/dp/IN05715 that I keep for this kind of situation. Because as soon as you move the Cold Junction point away from the temp sensor connector that is where I see the errors.

I have a calibrator that is capable of simulating the temps so I often will do a simulated CJ of 0 and a temp of 0 and look at the reading and compare that to the lab temperature gauge as a sanity check that the internal CJ of the unit under test. Often this is where I see the biggest issues as it underpins the maths for the measurements. It's why if I can I place the unit under test and the simulator box into a plastic box so that the unit and the connector can warm up equally, my simulator has a temp sensor in the thermocouple connector so is sensitive to people wafting air over it. I manage an Uncertainty of about 0.5 °C and a lot of that error is down to the specs of the CJ.

It is worth noting the specs for your meter are, 1.0 °C for the CJ and 0.3 °C for the unit. That is before you take into the 1.5 °C of the thermocouple wire.

[TUMEMBER]

https://www.measurement.govt.nz/resources/#collapse-control-1-5

[KT88]

One way to do a calibration is the old-school method by having an actual cold junction (TC in ice water). If you could use the same thermocouple wires there will be no errors from slight differences of the alloys.
An ice bath can be tricky though - you will see differences of up to one K if you let the bath settle and then stirr it up again. I prefer to constantly stirr the bath - it also helps to get a better thermal coupling between the TC and the water.
I would use two different baths for the 0°C measurement to avoid any cunduction between the cold junction and the DUT.

Cheers

Andreas

[jonpaul]

Rebonjour a tous!

As always perfect and précise advise from the  experts.

With just a plastic ice  tray,  partly melted cubes  1/2 water.

Immersed each TC (insulated , Kynar or Teflon) ~ 35 mm depth, adjusted the Fluke 51 zero CAL, then swapped the new Omega and old  (Fluke?) TC

For 3465A used  gold plated bananas to Type K mini adapter with used TC wire (not extension)

SUCESS! I Got consistent 31.9 deg F on the 34465A with either of the TCs.

And the Fluke 51 also reads consistent on both TC.

BRAVO to all who commented, i am very satisfied.

Bon journée et bon weekend,

Jon

[mzzj]

Bonjour a tous:

Checking the new Keysight  34465A  all   within sec as far as I can see but temperature is the last one.
Testing is only for the Type K thermocouples we use.: Omega, Fluke and others
Using the Omega mini K connectors.
I made a gold plated banana to Type K female adapter/extender with Omega type K TC wire.
Use old Fluke 51 handheld 2 1/2 digit Tep meter for reality checking. (also  Fischer mercury thermometers,  poor resolution).

The various TCs seem to match ~ 0.5..1 deg F.

Procedure:
_____________________

Boiling:

Distilled water, very slow boil in pyrex Erlenmeyer flask, immerse TC to middle of fluid,
Compensate for Baro pressure. CAL  100 C or 212 F.
So far within 1 deg F.
________________________

Freezing: 0 C or 32 F (not yet setup)

Dewar flask,  Pyrex dim 45 mm ID 125 mm L
Add distilled water,  make distilled water ice, add chopped ice,
decant most of the water, mix, allow to stab, immerse TC to middle of mix

Appreciate all comments and improvements from  The CAL Mavens.

Mille mercis,

Jon

For thermocouple ice point ordinary tap water should be enough good unless you live in area with extremely salty tap water. I have  measured  max -0,006 cel melting point depression with local tap water.
1% NaCl lowers melting point by 0.6Cel.
EPA limit for (sodium) NaCl is 50mg/L for drinking water, this 0.05% salt concentration would lower the melting point by -0.03 Cel

[jonpaul]

Hello again we are about 0.2 deg F in CAL at ice and boil point, with tap water, filtered,

Both the Fluke and Omega TC K correspond well and the Fluke 51 is also very close, after zero deg CAL

This is far better than I had expected and sufficient for my purposes.

Enjoy,

Jon