Pretty poor form for your mud logger and the other company not to be able to provide the info you requested. (I’d run them off if it were me!). I would ask to see the daily calibration check the mud logger and the system person should be doing. It could be that at lest one of them is “dog housing” the calibration.
How each detection system extracts the gas from the drilling fluid will also play a huge role in the amount of gas seen at the detector. Placement of those extractors is also important, both of the detectors your subs are using should be placed as close to the flow line as possible in the possum belly just before the shale shakers. A good comparison of the two systems would be to analyze the results of a carbide lag test the mud logger should be running on a daily basis. Also remember the many things affect Total Gas measurements including; mud viscosity, mud weight, annular velocity, polymer additives, & lost circulation material.
Although the last time I mud logged was almost 30 years ago there should be some parameters which should have remained fairly consistent. The measurement of total gas from your drilling fluid used to have a calibration standard of 100 units being equivalent to 10 percent methane in air. The detector of choice used to be a whetstone bridge devise known as a hot wire. Two different hot wires were required as one (Catalytic combustion)was accurate only as long as there was sufficient O2 to support combustion, above that point a Thermal Conductivity sensor was used. One of the problems which comes from these types of measurements are that your drilling fluid has more than just methane entrained in it. Adding additional gases causes the sensor reading to deviate from the expected calibration curve. Flame Ionization Detectors (FID) style detectors also are calibrated in the same 100 units = 10% methane but again the addition of gases in addition to the methane in your drilling fluid will effect your values. So the bottom line here is that you may have different types of detectors being used which one shouldn’t expect to be able to directly compare.