This test is the culmination of several efforts to remove contact type temperature sensors (e.g. thermocouples) from the cold well.
The cold well was run without any simulated blood products. Here's the raw data.
We need to know the temperature of the coldest place in the cold well in order to prevent the blood products from freezing. In the next figure, let’s focus on the bottom two traces. Thermocouple 1 is on the refrigerant tube and Thermocouple 2 is on the wall of the cold well, where the refrigerant tube runs just beneath the surface. Their agreement means we can get a good picture of the coolest temp inside the cold well by measuring the coolant tube temperature.
Next, we want to monitor the temperature of the top-most blood products, which will be the warmest. Previously, we’ve tried to do this with non-contact IR sensors, but the sensors weren’t isolating the temperature of the blood products. Their wide field of view also included the (much colder) walls of the cold well. So, I built a blinder that restricts the field of view (see previous post for details).
In the next figure, the blue traces, light and dark, are IR Sensor 1. Red traces are IR Sensor 2, with the blinder. “Temp IR # Object” gives the temperature of things in the sensor’s field of view. “Temp IR # Die” gives the temperature of the sensor silicon itself (which is used by the sensor for internal temperature compensation before outputting the Object value).
Notice that Temp IR 1 Object moves with Thermocouple 1 and 2, indicating that the cold well walls are in the sensor’s field of view. Temp IR 2 Object, from the sensor wearing the blinder, moves with the Die temps, which will vary with air temperature implying that the cold well walls contribute less to it’s reading.
Next, the cold well will be run with a full load of simulated blood products to verify this result.
Then, the cold well was filled to capacity (12 RBCs and 12 FFP) with simulated blood products.
There’s good agreement between TC3 and the IR Object average value.
The phase difference between the two IR Object readings is reduced versus an empty cold well.
The IR sensors have much more ripple than TC3. The Die temps have a fair bit of ripple too, so it’s probably real due to air temp fluctuations. It’s unlikely IR Object ripple will be reduced below Die ripple, still it’s worth trying a narrower FOV.
TC3 and IR sensors values are above 6 C. Products had previously been stored in a fridge measured to be between 0 and 3 C. Perhaps this is due to poor lid sealing. I added weights to the lid to get it it close more completely. The lid has insulation on top.
The IR Die temperatures have increased substantially versus the empty configuration.