Analog output board calibration

[magnetar]

Recently analog output boards were installed in incubators in the lab where I work. They output information such as temperature and CO2 percentage to a building management system using a 4-20mA current signal. The problem is that the output boards only have a precision of about 1 degree and sometimes a little more. For instance, if the display of the incubator reads 37 degrees C, the BMS may read the output as high as 38.2 degrees.
The BMS is scaled properly, and the output of the board depends on the display of the incubator(there are not separate temp and CO2 sensors for the output boards). According to the company that sold us the boards, there is no way to calibrate the output at the incubator and the calibration adjustment must be made to the BMS software. There are no adjustment pots on the output board or the primary board.
The idea of tweaking the BMS software was proposed to the engineer that runs and configures it, and according to him this is a far less than ideal way to calibrate the signal.
I have a little electronics experience and fashioned a simple circuit that sent the output signal through 250 ohms of resistance (the same as the BMS resistance) and drew a few microamps off the top to drive a transistor powered by a 6V source. The emitter current was then sent through the BMS and could be adjusted with a pot at the base of the transistor. With this setup, I could get the output to match the display very closely at operating temperature. If the temp changes though, the accuracy falls apart, and at 29 degrees the signal through my transistor makes the BMS read 25 degrees.
My question is, is there any circuit that I could conceivably add to this output board that would allow me to precisely calibrate it? Maybe something with zero and span calibration pots? I realize I may be well out of my depth, but any advice anyone could offer would be greatly appreciated. Thanks!!

 

[Mikebits]

So which reading do you believe is more correct, the incubator or the reporting device? If the incubator is more accurate, then I would tweak the software to match the incubator. Is the error linear? Sounds like whoever procured the hardware underspecified the boards. If the boards were specified with a tighter tolerance and are not meeting spec, then you should have the board vendor correct the problem or promplty refund your purchase. That is my opinion anyways.

Also for your transistor mod, not sure exactly what you did, but semiconductor voltage drops changes about 2mv/deg C.

 

[crutschow]

Correcting the readings in software is a common way to calibrate sensors. The parameter is measured with an accurate instrument at several points over the measurement range and compared to the sensor readings. The difference is then placed in a lookup table and linear extrapolation between calibration points is used to correct the sensor readings during normal operation. This is simple to do in software and the operating engineer such have no problem in doing that.

Trying to correct the readings with an added circuit often is difficult and can add errors of its own. And any nonlinear sensor errors are generally difficult to correct in hardware.

 

[magnetar]

thanks!

Thank you both for your input, it is very helpful.
sincerely,
joe

 

[magnetar]

Mikebits,
I'm sorry I might not have been clear, it is when the measured temp of the incubator changes that I am seeing the problem with accuracy, not the ambient temperature of the room. Although I am sure that that the ambient temp has a noticeable effect on measuring such small currents (I allowed several minutes for everything to stabilize before I tried to calibrate the circuit).
thanks,
Joe