Running a refinery gas analyzer and getting a lot of variance in the response of all the peaks on the FID. The TCD channel is very reproducible. The gas is injected by a valve with a 0.1ml sample loop. All the valves are turning properly. I have leak checked all of the connections and replaced liner, gold seal, septum, and column ferrule and none of that has helped. Any ideas on what I should try next? Thanks!