Omega HH-11 Thermometer Calibration

[gnuarm]

I found an old Omega HH-11 thermometer in a junk drawer and don't remember where it came from.  Digging around it seems this is a pretty old product as they don't actually list this model in many web sites.  What is listed is the HH-11A, HH-11B, HH-11C.  These all look a bit different with a bit of a wasp waist.  This unit is squared off with straight side.  None of the other models have the same buttons and so, features.  The thing read around 50°F in a room that was around 63°F, so it needs calibration.  I adjusted the offset and stuck the probe into an ice bath in a quart styrofoam cup with a lid.  The probe is inserted where the straw would go and after allowing the temperature to stabilize, the offset was adjusted until the measurement read 32.0°F.

Allowing the probe to warm back up to room temperature, it reads 59°F next to the thermostat which reads 63°F.  That seems a bit much to chalk up to the thermostat being off.  I realize they are not precision devices, but usually even inexpensive devices are close at room temperature. 

Does this sound normal?  I'd like to have a decent thermometer, even if it's a bit clunky.  The HH11A is rated as ±(0.3%rdg + 2°F), so not a lot better than the difference I am seeing.  I suppose the two errors combined could be 4°F.  There's also error in the thermocouple itself apparently.

Any advice?  I'd like to calibrate it at another temp, but the only one I could produce "easily" would be boiling water, and that's not so easy to get accurately really.  It's only 100°C at equilibrium.  The manual talks about five voltages from a reference to adjust five internal trim pots, but I'm not sure this device even has those same adjustments and I certainly don't have a mV accurate reference.

[Harfner]

Did you use distilled water? The freezing point of water shifts with slight impurities. That may add up to to a few F.

[jonpaul]

ice bath 0c boil distill water 100c compensation for altitudes and baromètre pressure

j

[Kleinstein]

When reading the room temperature there is essentially no error from the thermocouple itself, as this gives essentially zero voltage plus or minus a few percent if really bad. The qualtity of the termocouple gets relevant at temperatures much different from the cold junction temperature.

A test with a termocouple would bring the quality of the thermocouple as a possible source of uncertainty. So the adjustment of the readout circuit is better done with the mV range voltage(s).

For a first test a divider and DMM may be enough to at least check if the scale factor is way off.
The observes 4 F difference would be a huge error if this an error of the scale factor.
There should be mainly 3 points to adjust / calibrate:  2 constants for the cold junction zero and scale factor and one scale factor for the thermocouple. The K type is relatively linear and the nonlinear corrections may also want a trim, if done analog, but this should not have very much effect.
The points that effect the room temperature reading should be the 2 cold junction parameters.  The test with the external voltage is for the TC scale factor and maybe nonlinear corrections.

[jpanhalt]

Is there a "slope" adjustment?   

Do you want a high temp or low temp for your second point?   Your body temp might be good enough for one.  For temps much higher than 212°F, there are waxes with calibrated melting points.  Personally, I would probably use one of my old mercury thermometers for comparison and calibration.   

[gnuarm]

Is there a "slope" adjustment?   

Do you want a high temp or low temp for your second point?   Your body temp might be good enough for one.  For temps much higher than 212°F, there are waxes with calibrated melting points.  Personally, I would probably use one of my old mercury thermometers for comparison and calibration.

There is one "offset" adjustment externally accessible.  There are a bunch more internally which apparently impact the scale factor on different portions of the curve.  They are set by providing an input that emulates a wide range of temperatures and setting each adjustment at a different temperature. 

Looking up the equipment they specify, they refer to using a Temperature Reference Probe.  This seems to be a fairly special device that provides a thermocouple connector using the appropriate thermocouple material with a transition to copper for connection to a source voltage.  This junction must be maintained at 0 °C to obtain the correct voltage at the input to the thermometer.  They have an ice point reference that has a sealed chamber with distilled, de-ionized water and controls the amount of ice vs. water through a volume sensor.  Yeah, really!

So I'm not going to mess with any of the internal settings.  To give an idea of how sensitive this is to the small voltages from the thermocouple probe, just touching the meter causes the temperature reading to change.  The meter has to be stationary to get stable and accurate readings.

So I've set the offset control with the ice water bath and when I find another good reference, I check another temperature, otherwise this is good enough.

[jpanhalt]

Melting point of indium is about 313.9°F.  That would be a decent high point, but not terribly high. It is cheaper than I thought.  Amazon has 20g for $13.41.

[Kleinstein]

For the calibration of the display part it would be better to use a mV source instead of a real thermocouple. On would still need some cold junction compensation or a real ice bath to interface the TC connector.

[gnuarm]

Melting point of indium is about 313.9°F.  That would be a decent high point, but not terribly high. It is cheaper than I thought.  Amazon has 20g for $13.41.

I could use boiling water at 212°F.  It's probably a bit hard to get the temperature at an "equilibrium" at that temp.  We'll see.  It's probably working good now.  I just don't have a way to confirm that.

[Kleinstein]

To conform correct operation 2 test points are not enough. There are 3 main parameters to effect the reading: Offset, gain of the cold junction correction and gain for the TC voltage.

One test should include changing the temperature of the meter itself and not just the probe temperature. This could be with having the probe in an ice bath or other stable temperature and than let the meter chance temperature. Ideally this should not change the reading and no really need for a measurement of the actual temperature.  So one could start with a warm meter (e.g. some 40 C) and than let it cool down to room temperature and compare the 2 readings.

For the test with an external mV ref, there is no need for an extra cold junciton compensation: just copper wires to the TC plug and than looking at the difference in the temperature should be good enough. Zero voltage should read the room/ meter temperature.

[gnuarm]

To conform correct operation 2 test points are not enough. There are 3 main parameters to effect the reading: Offset, gain of the cold junction correction and gain for the TC voltage.

Sorry, I don't follow that.  In a sufficiently small segment of domain, any function is approximately linear.  So that would only require two settings for the approximation. 

What do you mean by "gain of the cold junction correction"?  The cold junction temperature causes a voltage that must be subtracted from the measured voltage.  As long as the cold junction temperature does not vary a lot, that would pose a very slight error indeed.  I've seen many devices that simply subtract a fixed value and don't worry with the cold junction temperature, assuming it is room temperature.

One test should include changing the temperature of the meter itself and not just the probe temperature. This could be with having the probe in an ice bath or other stable temperature and than let the meter chance temperature. Ideally this should not change the reading and no really need for a measurement of the actual temperature.  So one could start with a warm meter (e.g. some 40 C) and than let it cool down to room temperature and compare the 2 readings.

For the test with an external mV ref, there is no need for an extra cold junciton compensation: just copper wires to the TC plug and than looking at the difference in the temperature should be good enough. Zero voltage should read the room/ meter temperature.

None of this matters.  I don't have the equipment to produce a mV accurate reference, so it's good enough to ship!

[Kleinstein]

A linear function can still have 3 parameters - that happens if there are 2 input variables (room temperature and TC temperature).

Normally such a thermometer should not drift much with time and if it did drift it is really not clear which of the 3 main factors is off most. It could be the offset (e.g. offset of the mV amplifier or cold junction sensor) or it could be one of the scale factors (TC or cold junction). Without knowing the circuit it is hard to judge which parameter is more likely to drift, or which part was poorly adjusted.
Adjusting the offset at the ice bath is a guess, but one could as well use the room temperature point. At the room temperature point one could at least exclude the TC scale factor as the culprit.

[mzzj]

Did you use distilled water? The freezing point of water shifts with slight impurities. That may add up to to a few F.

Effect is less than 0.1F unless you use sea water. 
(Ordinary tap water around here was 0,006 C lower than pure water)

[AlienRelics]

I could use boiling water at 212°F.  It's probably a bit hard to get the temperature at an "equilibrium" at that temp.

I don't follow. I've always been told that all that is required is a rolling boil, and don't let the probe touch the sides or bottom.

[gnuarm]

A linear function can still have 3 parameters - that happens if there are 2 input variables (room temperature and TC temperature).

Normally such a thermometer should not drift much with time and if it did drift it is really not clear which of the 3 main factors is off most. It could be the offset (e.g. offset of the mV amplifier or cold junction sensor) or it could be one of the scale factors (TC or cold junction). Without knowing the circuit it is hard to judge which parameter is more likely to drift, or which part was poorly adjusted.
Adjusting the offset at the ice bath is a guess, but one could as well use the room temperature point. At the room temperature point one could at least exclude the TC scale factor as the culprit.

Since there is only one primary adjustment, it would seem there's no point in considering so many factors.  The other adjustments require very specific settings of the input voltage.

To set the thermometer at room temperature, one would need to know room temperature.  Without measuring room temperature with a thermometer that is more accurate, that would not be fruitful. 

[gnuarm]

I could use boiling water at 212°F.  It's probably a bit hard to get the temperature at an "equilibrium" at that temp.

I don't follow. I've always been told that all that is required is a rolling boil, and don't let the probe touch the sides or bottom.

All melting and boiling points are actually specified at equilibrium as well as a specific pressure.  I am not far from sea level, so that would not impact me so much, but the boiling point of water is impacted enough by pressure that cooking recipes have to be adjusted for elevation in places like Denver. 

Just as with the melting point, where the water is typically a higher temperature than 0°C and the ice is lower than °C, boiling water is actually a bit above 100 °C and varies across the volume, hotter in contact with the burner and cooler where it is evaporating.  If the systems are in equilibrium, so no actual boiling, and the heat is applied on all sides (actually, not heat, just a constant temperature) then you get an accurate 100 °C.  Maybe this effect is not so important.  I was a chemist at one time and there were times when we would be very cautious in measuring such temperatures to get accurate values.  Usually it was a melting point.  With a pure substance the melting point is very accurate and precise.  When a substance has impurities, the melting point is lowered, and broadened so it doesn't all melt at one temperature. 

In some of the old movies they would make a big deal of measuring the melting point of drugs to determine the purity, but they made it into a drama scene, not at all realistic.  I do know you have to let the temperature rise very slowly at the transition to get a good reading. 

[Conrad Hoffman]

It's overkill and I'm not sure how to view the pictures, but this might be of interest- https://www.scientificamerican.com/article/tackling-the-triple-point/#:~:text=regenerate%20the%20vacuum.-,TRIPLE%2DPOINT%20CELL,around%20the%20inner%20test%20tube.