Temp measurement and accuracy

[swmcl]

Hi all,

I have been stuggling with temp measurement. I started out with a cheaper ($200 - China) multimeter with temp probe. Then I borrowed a little solid temp probe from work that had a long straight metal shaft with the probe inside. This second probe read 2.3 degrees C different to the original. I then thought I'd buy a thermometer from B&H (the Patterson one) with as much accuracy as possible (0.1 degree graduations). Alongside that purchasing decision, I went and bought a calibrated multimeter as well (an expensive Extech 570A which is really 'accurate' ...) (This mulitmeter will replace the other Chinese one for all functions - it aint just a temp meter !)

The Extech and the thermometer are about 1.5 degrees C different. The Extech and the original Chinese one are about 2.5 degrees C different. The Extech is reading similar to the probe from work.

Which one do I trust ?? Just how accurate are the thermometers ?? I know the thermometers are certainly slower to repond to variations.

Given that 1 degree matters in processing ...

Cheers,

Steve

[Vaughn]

It does not matter how accurate they are (at least within several degrees of "normal"), as long as they have good precision. In other words, as long as they give the same temperature under the same circumstances every time.

Do all your work and testing with the one thermometer and you are set. If you get another thermometer, calibrate it to your original.

Vaughn

[Heroique]

It does not matter how accurate they are ... as long as they have good precision. In other words, as long as they give the same temperature under the same circumstances every time.

I like what Vaughn says, so let’s give it give it an example:

If you like results when your thermometer reads “68 F,” then you want to be sure that each time your thermometer says “68 F,” you're getting that very same temperature from job to job – even if it might be different than the “real” temperature.

What makes inferior thermometers so exasperating is that w/ each indication of, say, “68 F,” you may actually be getting varying temperatures. Maybe 67 for one job, and 71 for another, even though your thermometer always says 68…

I like dial (analog) thermometers, but I’ve noticed they have a tendency to vary w/ barometric pressure changes! Maybe the better dial thermometers don't do this.

[Doremus Scudder]

Many good labs keep an expensive calibrated thermometer in a safe place to check other thermometers against occasionally and make adjustments.

I've got an old Kodak Process Thermometer that I use for that purpose and a few of the Kodak color thermometers as well. They all seem to read within less than 0.5°C of each other.

The analogue dial thermometers are another story. I've got stickers on them telling me the error at my processing temperature (e.g., "reads +2°F") for the ones with no adjustment. For the others that can be twisted to adjust the reading, I adjust against the process thermometer at processing temp. If it's wrong, at least I'm consistent.

Acquiring a good, calibrated thermometer (the one Leigh suggests would be fine if it is that accurate through its entire range) and using that to calibrate other thermometers to.

BTW, there are lots of other threads on this topic here. Do a search for even more than you wanted to know.

Best,

Doremus

[Jim Jones]

Many good labs keep an expensive calibrated thermometer in a safe place to check other thermometers against occasionally and make adjustments.

I've got an old Kodak Process Thermometer that I use for that purpose and a few of the Kodak color thermometers as well. They all seem to read within less than 0.5°C of each other.

The analogue dial thermometers are another story. I've got stickers on them telling me the error at my processing temperature (e.g., "reads +2°F") for the ones with no adjustment. For the others that can be twisted to adjust the reading, I adjust against the process thermometer at processing temp. If it's wrong, at least I'm consistent.

Acquiring a good, calibrated thermometer (the one Leigh suggests would be fine if it is that accurate through its entire range) and using that to calibrate other thermometers to.

BTW, there are lots of other threads on this topic here. Do a search for even more than you wanted to know.

Best,

Doremus

Yes, indeed. Analog dial thermometers should be checked often against a reliable meter. If they are stored together in a darkroom with stable temperature, this is easy to do at room temperature whenever you use the dial thermometer. Keeping the darkroom, chemicals, film tanks, and wash water at the desired developing temperature saves a lot of bother.

[swmcl]

Leigh,

It doesn't matter much ... if McMaster is the only place we know of then the cost must be borne ...

I'll send a PM to discuss further ..

Cheers,

Steve

[Louie Powell]

It does not matter how accurate they are (at least within several degrees of "normal"), as long as they have good precision. In other words, as long as they give the same temperature under the same circumstances every time.

Do all your work and testing with the one thermometer and you are set. If you get another thermometer, calibrate it to your original.

Vaughn

Vaughn has the right idea, but it's not expressed clearly.

The issue isn't precision. The term 'precision' refers to the 'fine-ness' of the measurement. A thermometer that measures to 0.1 deg us sufficient for photochemical processing.

The critical issue that Vaughn meant to say is 'repeatability'. That is, if the liquid is 100deg, then the thermometer should register 100 deg every time it measures the temperature of that liquid.

The third measurement concept is 'accuracy' which refers to how closely one instrument compares to an absolute reference instrument. Accuracy is important to achieve predicable results, but the photochemical process does allow for minor variations in temperature if compensating changes are made in time.

But as Vaughn said, of the three concepts - precision, accuracy and repeatability - the most important is repeatability because if your thermometer is repeatable, then the results that you will achieve will be both predictable and repeatable. No one cares about how accurately you measured the temperature of your processing solutions - all they care about is whether your negatives are properly developed.

[Leigh]

...of the three concepts - precision, accuracy and repeatability - the most important is repeatability...

Absolutely correct.

As long as a thermometer gives you the same reading every time for a given temperature,
it will work fine for process control.

Any actual error between what the thermometer says and the true temperature will be compensated automatically
when you calibrate your process. We all calibrate our development process, yes?

- Leigh

[Leigh]

Hi Steve,

I expect your multimeters are using Type K thermocouples, since those are most common and cheapest.
Consumer-grade Type K thermocouples have a rated accuracy of ±2.5°C (~5°F).

Even the best laboratory-grade Platinum thermocouples are only rated ±1.0°C (~2°F).

======

I just looked through the offerings at McMaster-Carr. They have quite a selection.
The one I would get is catalog nr 3569K58 at $53.92.
It's rated accuracy is ±0.4°F (~0.2°C). It comes with an NIST certificate of calibration, so you know it's right.
It looks like this: The site is http://www.mcmaster.com
Great folks to deal with. No minimum order and no handling charges, just actual shipping.

======

You can calibrate any thermometer that has sufficient range (some have a very narrow range of readings).

Mix up some ice water (preferably distilled water for both) in an insulated cup. Swirl it around to make sure
it's reached equilibrium. The thermometer should read 0°C or 32°F.

Put some distilled water into a pot and bring it to a good rolling boil. Your reading should be 100°C (212°F).

While distilled water is not essential, any significant mineral content or contamination
may change the actual temperatures from the ideal.

- Leigh

[SergeiR]

Put some distilled water into a pot and bring it to a good rolling boil. Your reading should be 100°C (212°F).

Nope. Depending on the altitude boiling point may be different from 100C.


Anyway - as long as under same conditions measuring device will give same results - not really important. Can just tape up scale to the desired point even if it reads 100 at 90 - who cares as long as its 90 every time