Cold junction compensation sensor

[derZockerMP]

Hello, I want to implement cold junction compensation on a Type T thermocouple. This is for a scanner card connected to a K2001 Multimeter. The multimeter can integrate a temperature sensor on Channel 1 to measure the temperature of the reference junction and compensate for drift. In the original Keithley 2001-TSCAN card, this temperature sensor is an AD590. I want to rebuild this small circuit externally using a copper block, with the wires inserted into a hole. The temperature should be measured with a sensor, but the AD590 is priced around 60€, which is quite high . Is there another sensor that can be integrated into this application, or are there any alternative projects for this?

[Kleinstein]

The AD592 is essentially the same as the AD590, but in a TO92 plastic case. In most cases good enough.

[derZockerMP]

Thanks, that's a good option and available at Reichelt, too! Do you have a recommendation for a good Thermistor for the calibration procedure that fit to the 0.01°C accuracy?

[Kleinstein]

The calibration may work with a few temperature fix point's. 0.01 K accuracy is however really difficult and would likely need a water tripple point cell. The simpler fix points (melting ice, boiling water, meting of Ga or tin or zinc have usually larger uncertainty).

For high accuracy the sensor of choice is usually PT100. Good ones are however also expensive.

With the typical thermocouple accuracy the AD592 may not need an extra calibration. A a single point, like meting ice (use with the TC) could be good enough for normal use.

[derZockerMP]

OK, I will try it out, thanks

[EC8010]

What do you intend the copper block to do? Copper is expensive and horrible to machine because it's sticky. Drilling holes is tricky and tapping them is even worse (very easy for the tap to clog and snap). Drilling and tapping using full fat milk as a lubricant works, but you need to clean up perfectly afterwards to avoid gone-off milk smells. For most purposes aluminium is very nearly as good and an awful lot easier to deal with. Methylated spirits is a good lubricant when drilling tapping aluminium.

[CalibrationGuy]

I would use the RTD since the meter supports it directly.

TomG.

[TimFox]

What do you intend the copper block to do? Copper is expensive and horrible to machine because it's sticky. Drilling holes is tricky and tapping them is even worse (very easy for the tap to clog and snap). Drilling and tapping using full fat milk as a lubricant works, but you need to clean up perfectly afterwards to avoid gone-off milk smells. For most purposes aluminium is very nearly as good and an awful lot easier to deal with. Methylated spirits is a good lubricant when drilling tapping aluminium.

Yes, copper is very difficult to work.
I (carefully) drill holes (with lots of lubricant) and use a screw and nut to hold things together instead of tapping (and breaking the tap).

[mzzj]

The calibration may work with a few temperature fix point's. 0.01 K accuracy is however really difficult and would likely need a water tripple point cell. The simpler fix points (melting ice, boiling water, meting of Ga or tin or zinc have usually larger uncertainty).

For high accuracy the sensor of choice is usually PT100. Good ones are however also expensive.

With the typical thermocouple accuracy the AD592 may not need an extra calibration. A a single point, like meting ice (use with the TC) could be good enough for normal use.

Water triple point for 0.01 K accuracy is bit overkill and Gallium melting point would be also overkill. Commercial Gallium melting points are with <0.001K uncertainty.
Ice-water bath is better than 5mK with little preparations and 10 usd budget.

Of course anything like that is overkill for most thermocouple cold junctions.

[arcnet]

Thanks, that's a good option and available at Reichelt, too! Do you have a recommendation for a good Thermistor for the calibration procedure that fit to the 0.01°C accuracy?

As already mentioned RTDs are the way to go if such an accuracy is really needed but even the old 1/10 DIN PT100 are not that accurate without calibration (+-0.03 K @ 0 °C). There are some thermistors that would fit like USSensors (now Littlefuse) USP3986 (600€) or Fluke 5640 series (afair >= 2500€). Assuming that each input of such a scanner card has the same temperature could also be a bit optimistic.
Usually ignored because not relevant for almost all thermocouple measurements are the errors of the inverse functions that are used to calculate the final temperatures. For Type T in the 0 °C to 400 °C range these are +-0.03 °C (see https://its90.nist.gov/InvFunctions)

[Overspeed]

What do you intend the copper block to do? Copper is expensive and horrible to machine because it's sticky. Drilling holes is tricky and tapping them is even worse (very easy for the tap to clog and snap). Drilling and tapping using full fat milk as a lubricant works, but you need to clean up perfectly afterwards to avoid gone-off milk smells. For most purposes aluminium is very nearly as good and an awful lot easier to deal with. Methylated spirits is a good lubricant when drilling tapping aluminium.

Hello

Copper is not really horrible to machine :

The first problem is a lot of cutting oil brand stain the copper , a cheap and easy way to machine copper is plain milk
The second is the cutting angles I use HSS Co8 with cutting angle for aluminium
The third you need to flush all chips to avoid to smash chips between the tool and the part.

I agree than aluminium is very close and far easier / cheaper , use 5000 or 6000 serie  , 6000 and 6061 / 62 are easier to machine that soft 5000 alloy .

Regards
OS

[derZockerMP]

I want to use the copper block for thermal stabilization of the cold junctions (four of them). I plan to drill a hole in the block and mount the copper/constantan cold junctions in it. The copper/copper junction shouldn't be a problem because it uses the same material. The temperature sensor for compensation should be mounted on the copper block. I'll try using the copper block because I already have one at home.

[derZockerMP]

Here some photos of it.

[Kleinstein]

The wires can conduct quite some heat to the TC junctions and sensor.  Ideally one would want extra thermal contact from the wires to the block.
I once build a very similar cold junction block. The wires, especially the copper ones were additionally wrapt 3/4 around the bock for additional thermal contact. 

[EC8010]

If you drill a hole in the copper for the TO92 device and epoxy it in you will get a much more stable long-term thermal contact. That tape will loosen over the years.

Kleinstein: Good point about the leads to the device. That suggests that thin wire from device, wrapped around the block, then connected to the (probably thicker) wires going to wherever they need to go. These sorts of things are all about the detail.

[Kleinstein]

AFAIR I drilled holes for the AD592 sensor and the TC junctions. In my case type K thermocouples and thus 2 junctions TC wire to copper for each TC channel. AFAIR it was 3 TC channels with the wires going to the yellow plastic TC connectors and the copper cable going to some SUB D type connector to a scanner input at the DMM. It that case the nice point was that the DMM scanner could also measure current and turned off the channel when not used. So the sensor supply could be a fixed battery inside the same box.  For the AD592 keep the supply voltage low (e.g. 6 V) as the self heating of the sensor is visible and a possible, though not major issue.

[derZockerMP]

I will leave it and watch for it. The Voltage for the AD592 will be 5V and the temperature is meausured via a 200Ohm resistor in series to get an voltage drop acros it. It is also made in that way in the Keithley 2001-t-scan card.