Well,
looking at the numbers from Dave's short video I calculate like this:
temp 121GW Gossen U1282A
---------------------------------------------------------------------------------------------------------------------
room 22,60 °C 4,5001 V 4,5006 V 4,4999 V
test 12,80 °C 4,4996 V 4,5011 V 4,5005 V
Temp diff -9,80 °C -0,0005 V 0,0005 V 0,0006 V
total err -1111,09 ppm 1110,96 ppm 1333,36 ppm
-0,11% 0,11% 0,13%
---------------------------------------------------------------------------------------------------------------------
Tempco 113,38 ppm/°C -113,36 ppm/°C -136,06 ppm/°C
0,011%/°C -0,011%/°C -0,014%/°C
temp 121GW Gossen U1282A
---------------------------------------------------------------------------------------------------------------------
room 22,60 °C 4,5001 V 4,5006 V 4,4999 V
test 44,70 °C 4,5000 V 4,5005 V 4,4987 V
Temp diff 22,10 °C -0,0001 V -0,0001 V -0,0012 V
total err -222,22 ppm -222,19 ppm -2666,73 ppm
-0,02% -0,02% -0,27%
---------------------------------------------------------------------------------------------------------------------
Tempco -10,06 ppm/°C -10,05 ppm/°C -120,67 ppm/°C
-0,001%/°C -0,001%/°C -0,012%/°C
Please someone correct me if I calculated wrong...
Also @joeqsmith, tempco is described in detail in Metrawatt Ultra datasheet. It is specified for Volts DC at 0,05%rdg+5d/10°K .. That comes out roughly to 50ppm/°K + some digits.
EDIT: In specs, you can see that it is specified from 0°C to 40°C. Your test at -20°C and 60°C is way out of instrument specification..
Manufacturer was very upfront that this meter perform to it's full spec in the laboratory and controlled temperature environments and we shouldn't expect it to perform much better than specified..
I would also not compare Joe's measurements with Dave's. Measuring in millivolt range introduces increased visibility all kinds of thermal effects in cabling and setup. I would expect that good portion of measured values comes from that influence. Measuring at 4.5 or 10 V is much better to suppress these kinds of errors and to concentrate on actual tempco of an instrument.
Looking at data for those two measurement points separately (room to -10°C, and room to +20°C) you can also see nonlinearities in tempco, that box method (min to max temp delta) doesn't show.
In order for tempco measurement to be useful, there should be measurement point every 10°C (or 5°C, even better ) from 0°C to 50°C with soak time of 45min at every point.
It should be performed at 5-10V range to minimise cabling and setup tempco...
Best regards,
Sinisa
But considering this a quick check, it seems so far that 121GW does pretty well, compared to more expensive meter..