Sean, I'll have to go back and have a look at some of the test results we did a while ago. It wasn't so much the margin of error, but the fact that the measurement was often different. Because the laser is able to be slid along it's rails to make the circumference of the readings larger. This then changes the amount of revolutions (if the set time is constant) and the amount of data collated to give the ColorTrack number. So often the start and the finish point vary every time you do a scan. Even when using the same sample. I think for it to be more reliable it has to read only one revolution, starting and stopping at a single point. Otherwise you're taking double readings of some beans, and only single of others, and that's where the variance is. The interface is kinda cool as far as being able to see the range of colours measured, in the line graph, rather than a final average.
Some feedback I was given from the Fresh Roast Systems was told the beam & track width is less than 2mm, ground samples should not be patted down - level, but not compacted, just fill and scrape. Laser should sit in housing just shy of flush (1/4"-1/8"), but doesn't really matter.
Chris, I think the scale is Gourmet for Agtron, I'll have to double check. What you raised in regards to the Agtron sample prep are some of the things the CT was meant to overcome. For the Agtron, we also do both readings (roaster-side and finished product). We have also calibrated our tastes profiles on the delta reading between bean and ground. The readings aren't directly inversely proportional to the Agtron readings either.
Our Colortrack at the moment is a big microwave-looking paper weight until we have time to put into it again.
I think it has potential to compliment the Agtron, but for us at the moment (for us) it could never replace it, until we adjust our taste calibration. We're planning on doing some tests soon.
Happy to throw around ideas if you have any.