See my "Servo Booster" thread for details... That shows the simple NPN booster I came up with which works very well. On the v4.2 release Bryan changed over to using a MOSFET instead of an NPN because that board is running all MOSFETS and no NPN's.
Here is a shot of the MOSFET booster Bryan used on the v4.2.3 test board, which works just as well....
The other big improvement that has been made which makes longer CAT5 cables work better is the RC filter(s) Bryan added to the probe leads. These filters remove noise that is picked up over long cable runs and makes the probes read MUCH smoother whether connected directly to the HM or over a CAT5 cable.
These filters consist of just two components, a 100K resistor in SERIES between the probe and the ATMEGA, and a .1uf cap from the probe lead to ground. Here is a pic of the RC filters from the v4.2.3 schematic so you can see the position and orientation of the components.
The 10K resistors are the pullups that are present on your board now, unfortunately since the filter resistor is in series with the probe leads you would have to lift that lead and insert the resistor inline, the caps can just jumper from the probe tip lead to ground.....
Edit:
The forthcoming HMv4.2 release will have the servo booster and filters built onto the board, it has all the components completely re-arranged to isolate the power circuits from the sensor circuits, has an improved ground plain design, has a feedback circuit added to the blower driver to support additional (less noisy) operational modes. The HMv4.2 design is improved in lots of ways that will extend the viable length of your CAT5 cable. I'm not sure how much improvement you would net from just adding the filters cause I haven't tried that yet. (HMv4.2 also has optional onboard thermocouple pit probe support that works really well in my initial experiments) I encourage you to jump into the HMv4.2 project when it becomes official because it is just that good, a big step forward from previous releases IMHO...