Modification: Increasing roll servo leverage.

At high speeds, the control force on the ailerons is significant.  My GRT high-torque roll servo just can't articulate the aileron pushrods without slipping its clutch when flying at speeds above around 140 KIAS.  I had already moved the servo control arm closest to the servo arm pivot point, as shown below from GRT's autopilot installation manual.

With that option exhausted, I needed to increase the mechanical advantage to the roll servo.  Below is what my install looks like now, when the motor is overwhelmed by the control forces at high speed.  You can see the servo pushrod is at the "min throw/max torque" position.  And it's connected to the bellcrank one inch below the aileron pushrod pivot point (as accommodated by the bellcrank design).

Below is what I planned to do.  I wanted to make a new bracket (grey) that moved the attachment point of the servo pushrod (yellow) further from the bellcrank pivot point.  This would give the servo greater mechanical advantage (since large movements of the roll servo would translate to smaller movements of the aileron).

So, I hopped into LibreCAD to fashion some brackets with varying lengths.

• My plan was to make these brackets out of 0.125" 4130 steel stock.  I suspect  that's thicker than it needs considering the bellcrank material is much thinner.  
• The spacing between the aileron pushod rod end bearing and the nutplate on the bellcrank for the servo pushrod's original position is exactly 1". 
• I wanted to use a nutplate to attach the distal rod end bearing of the servo pushrod so I didn't have to fuss accessing a nut on the backside of the bracket in the limited space afforded by the inspection panel.
• The bracket width of 1.2" was dictated by the mil-spec edge distance requirements associated with nutplate rivets (I later found out that this was too thick and would bind on the aileron pushrod).

This would also mean I would have to lengthen the servo pushrod from 3.25" to 4".  So I bought some tubing and used this method to drill holes in the center (since I don't have a lathe or mill) and threaded it.

Here's my collection of brackets ready to go.  In retrospect, I gave myself too many length options.

Here's a bracket (second from the left above) attached.  The geometry of the pushrod didn't work out as the servo arm's angle exceed 30° of travel (causing a potential over-center instance and thus jamming the roll control).  No adjustment of pushrod lengths, across the four pushrods I had, would yield an acceptable geometry. 

I then made this temporary, not flight-worthy, bracket next.  The pushrod geometry with this bracket also did not yield something workable.

Next, I fashioned a new servo arm that has a hole slightly lower than the one on the stock arm.  Below is two of those exploratory, not flight-worthy, servo arms.  The large hole bolted to the servo.  The next hole goes to the shear screw.  The next hole(s) are locations for the pushrod.

As the pair of pictures below show, the servo arm with even "more torque, less throw" also did not have a workable geometry.

So at this point, I have no solution to the issue.  GRT has already checked out the servo.  They are checking with their software group to see what else they can do, yet I don't expect to hear back from them.  I remain of the sentiment that the servo lacks sufficient torque for the aircraft and that's the way it shall be.