Thermometer calibration

[Regina Castor]

We have an old policy saying that the thermometers in the same storage device must agree within 1 degree C. Does anyone have a reference for this?

Also, We had to pitch all of our mercury thermometers and buy new ones. We even bought a new NIST standardized thermometer against which to compare all the others. Problem is, our brand new thermometers don't correlate at all to the new standard. We have as much difference as 2.5 C from the reference. Has anyone had this issue before and could you offer some suggestions where I go from here?

[Lcsmrz]

The problem lies in measurement uncertainty, as an expert explained to me. A reference thermometer that is accurate to +1 C per certificate is compared to another that is +1 C, and you get -- at best -- correlation of +2 C with additional uncertainty adding at another 1 C . Yet, we continue to define their accuracy to one decimal point! I think there's still a small discussion in the Tech Manual that sets the difference at 2 C.

Interestingly, I found out that the thermometers in most BB storage devices are accurate to about 0.05 C within operating range, yet we "calibrate" them using liquid-in-glass NIST thermometers with a resolution of 0.5 C and an accuracy of 1 C. Like using a POCT glucose meter to verify our chemistry analyzer. Old habits die hard, I guess ...

The answers are to hire a metrologist, use a very accurate reference thermometer and research all of the things that affect temperature measurements within a cabinet, or reduce the measurements to its most accurate temperature representation given the measuring device, probably just an integer.

[rickkaestner]

Enjoyed your post. However, I find it hard to believe that thermometers used in BB storage devices are accurate to 0.05°C. We sell temperature sensors and a sensor that has that type of accuracy, at least a thermistor, costs well over $ 250.00. Might you be talking about repeatability rather than accuracy?

Rick

[Lcsmrz]

We were beginning a project of cross-calibrating the probe connected to the digital readouts on our storage devices, the ones on the central monitoring system for alarms, and our usual NIST Liquid-In-Glass thermometer. The outside guy started discussing the many problems measuring the cabinet air temperature between adjacent devices -- the science was well over my head -- and basically talked us out of the entire project due to cost!

He did say at one point in the discussion that his probe and probably the one in the refrigerator we stood next to was accurate to half the display (0.05 C) and to keep my NIST thermometer out of the project completely as comparing apples and watermelons.

The refrigerator manual does not specify the accuracy of the digital display. An email to the refrigerator manufacturer went unanswered, so I could not confirm or deny his statement, but a web search indicated that probes and controllers consistent with his statement were available.

I found very little information in the online engineering literature comparing temp measuring devices, except that they are hard to compare. However, I did notice that the FDA food inspectors use non-contact infrared thermometers in their inspections, rather than thermistors, so it appears that the subject of accuracy in blood banks may not be as important as I think!

[rickkaestner]

Measuring the air temperature of a refrigerator or freezer accurately is very very very difficult to do. To calibrate a thermometer inside a refrigerator it must be placed inside a container with a large thermal mass and shielded from any moving air.

We calibrate inside a large heavy metal box. This is the only way to get a steady fairly accurate measurement. The air inside refrigerators and freezers is always moving and there could be as much as several degrees difference just one or two inches apart. Even thermometers placed side by side could read differently because of the many factors (thermal mass of the thermometers, built in temperature buffering, position. etc...). So if you have a thermometer calibrated by a metrologist it will be accurate to the specified accuracy but that won't necessarily mean that it will measure the actualy temperature of the unitit is placed in. Comparing the temperature accurately of one refrigerator/freezer to another is impossible.

We sell thermistors, digital sensors and thermocouples which have an accuracies of between 1.0° and 0.1°C. When we calibrate them we must take extreme measures that you would probably never be able to duplicate in your blood bank.

My suggestion is to take the accuracy stated by the manufacurer and accept it. If you need the thermometers calibrated your best bet is to get some else to do it. The accuracy of the thermocouples built into the refrigerators and freezers are probably good enougy but even the manufacuter is reluctant to tell you how accurate their thermocouple is becuase of all of the many factors inherent in any individual refrigerator/freezer.

Hope this helps

[mwilliams]

Some information on accuracy vs. resolution in thermometers can be found

here: http://www.mackinfo.com/accuracy_resolution.htm

Matt

[rlmirand]

Has anybody replaced mercury thermometers with something else.I would like to know more about my options.

[adiescast]

We replaced all of our mercury thermometers with the red liquid filled thermometers. I am not at work, and I do not remember exactly what fluid they use in those thermometers. We have not had alot of problems with accuracy, but we are not trying to peg the thermometers to .5. I require them to be within 1 degree C of the NIST on annual calibration. I do not require two thermometers in the same large device to agree within 1 degree as long as they are all within the acceptable range for the device. Any large temperature controlled storage device is going to have a range of temperatures throughout the device (see other threads discussing temperature mapping).

[rlmirand]

Thank you so much!

[L106]

We also replaced all of our mercury thermometer with red liquid filled thermometers. (Available from any typical supplier such as Cardinal Healthcare, etc.) We have not had any problems with them. We require all our thermometers to be +/- 1 C of the NIST thermometer.

[SRMC BB]

We use the Temp-Chex Red Spirit thermometer manufactured by Streck Labs. They were purchased through our suppliers-- Infolab and Cardinal. Have not had any calibration issues with these against the NIST thermometer (which is sent out annually for calibration)

[dal6164]

Has anybody replaced mercury thermometers with something else.I would like to know more about my options.

We replace ours with streck thermometers that supposingly never needs to be calibrated

[Alternatives]

We replace ours with streck thermometers that supposingly never needs to be calibrated

I think you will have some compliance problems if you use some sensors which are not calibrated...They must be calibrated!

Benoît

[RR1]

In the UK we are expected to have all thermometers recording fridge temperatures or used for fridge mapping purposes to have been calibrated against a national standard thermometer, and be within 0.5'C of that calibrated instrument. This would then give a tolerance of 1'C between the instruments (+/- 0.5'C) ; the importance of this is that your fridge could be recording 6'C ...but this figure may in reality be anything between 5'C and 7'C. If the tolerance was 2'C the "real" range of the storage temperature would be even greater (4'C- 8'C) etc.

Edited March 6, 2010 by RR1

[M_Allan]

Interesting topic, but one important thing nobody mentioned.

All samples or units in a fridge or freezer need to be stored within a certain temperature limit.

It is not only, that temperature fluctuates because of air movement, there is also a permanent

difference in temperature in different locations of the fridge or freezer.

Usually, I notice, that many people place a thermometer in the front of the fridge, so they

can easily take the reading.

Most of the time, the temperature in the front of the fridge, can be many degrees higher then in

the back.

It is of importance, to map a fridge or freezer, to pin-point the difference in temperatures in

several locations.

If you located the coldest and hottest spot in a fridge, you could place a thermometer or

monitoring system probe there, and by taking the readings, you have a certainty, that the rest of the

fridge, is within these two readings.

For a freezer it is easier, since it is only really important what the highest temperature spot in your freezer is, so one thermometer/probe would be enough.

Marc

[M_Allan]

And, yes,

Another important issue is the hysteresis of the fridge or freezer.

As you see in my avatar to the left, this is a graph of a fridge.

Every fridge or freezer does the same thing. The compressor switches on and off,

and depending on the quality of the thermostat and compressor, the temperatures

can fluctuate a lot.

Sometimes even, for an extended period of time, the fridge might freeze, or go above the maximum

limit.

This is hard to monitor with a standard thermometer, and I think, it would be wise, to, at least,

once every 2 months, use an electronic thermo-logger, that will show you these fluctuations, and

if it runs within the accepted temperature fluctuations.

Especially, when a fridge has been closed over night, the temperature usually goes down a lot, and

I have seen plenty of fridges, that you could better use to make ice-cream.

Marc

[RR1]

Interesting topic, but one important thing nobody mentioned.

For a freezer it is easier, since it is only really important what the highest temperature spot in your freezer is, so one thermometer/probe would be enough.

Marc

Hi Marc, it's also important to know the lowest temperature of the freezer , as blood bags become brittle and fracture easily at very low temps. This is obvious when folks start seeing a higher than average breakage of their FFP on thawing.

[M_Allan]

Hello RR1,

You are right, but I think, the critical temperatures in regards

to the stability of the samples and units are more important, and that

is were a couple of degrees difference can cause a lot of problems.

Marc