I exported the log into Excel and it looks like the largest errors are at 10 gps and less. I did lower the value at 10 (added the values at 20 & 30 and used that number for 10) and got them closer, but I have nothing to compare that to in my logs and wouldn't know for sure if I was making it better or worse.
I guess I could just filter out anything less than 10 gps and see how it looks. I'll give it a try when I get some time next week.
Thanks for all the help.