I have not validated or tested the theoretical design presented below nor have I provided any detail on how to safely implement such a system. Even worse, I am human and make mistakes, OFTEN, sometimes with severe financial and/or life threatening consequences. Use of this information is at will and completely at your own risk.
Here is my proposal for optimizing a hydraulic conversion for PT105 style bender.
I propose designing for 110* of bender arm travel. This should account for any design slop in the dies/bender as well as springback to make 90 degree bends in a single stroke, without repinning.
Determine a suitable mounting point near the end of the bender. I propose 22" from the center of the dies because this works out nicely for a 36" stroke cylinder and should work fine on a PT105 or JD2 model 3 bender. You don't have to use this 22" dimension exactly if there is a better mounting point inboard of this, but you will need to scale the flow rate down based on scaling this dimension down (the end result will be the same bending speed, but higher operating pressure). If you go outboard of 22", you will need longer than the suggested stroke or you will be limited to something less than 110* of travel.
Angle the cylinder so it aligns with an imaginary line that runs from the starting point of this mounting location to where the mounting location will end up at maximum bender travel (see image for a visual). For 110 degree design travel, the mounting angle should be 35*. Mount the rear of the cylinder so this angle is achieved in the collapsed position. This mounting angle is what optimizes the conversion of linear movement to rotation when using a fixed displacement pump / fixed linear rate.
You can also use a trunion-style front mount which will reduce the overall size of the bender and may be more convenient for mounting. In this case, you want the same mounting angle with the imaginary line above to pass along the cylinder and the trunion mount.
I suggest using a 2" diameter cylinder. This is the smallest cylinder that is readily available in this stroke length and will operate at very reasonable pressure (max ~1500psi) with this geometry. Alternatively you could use a 2.5" cylinder and up the flow rate per the suggestion in the flow rate section below; there will be no performance drawback in terms of speed or capacity, but a larger cylinder will be heavier and may present a tighter fitment onto the bender; it will also operate at a lower pressure.
The calculations provided assume a maximum capacity of 5,000ft-lb, which I believe will be capable of bending 2"x0.120 DOM or 2"x0.095cromoly. This is based on calculations I did many years ago and scaling some test/measurement values from samples of smaller 1-3/4" tube.
Based on the capacity above, for a 1hp power unit, you should be able to use up to 1.13gpm pump. For a 3hp power unit, you should be able to use up to a 3.4gpm pump. If these aren't readily available, consider using a 2.5" cylinder with up to 1.77gpm for 1hp or 5.31gpm for 3hp. If you go with a lower flow rate, the bender will simply operate more slowly and consume less than the available power. If you use a higher flow rate, the motor may be overloaded or stall.
Using the numbers above (ether for 2" or 2.5" if you select the pump accordingly), you will be able to do 110* of travel in 26sec for 1hp or 9sec for 3hp.
What makes this an optimized design?
The goal of this design is to fully utilize the available power from a fixed-displacement pump. The biggest challenge to doing so is converting the linear travel of the cylinder to rotational travel of the bender arm. There are many ways to mount a cylinder and get ample torque and rotation out of the bender, but doing so in a non-optimized geometry will leave speed on the table. The same goes for selecting a power unit and pump that flow at a lower-than-optimal rate.
For example, the outstanding hydarulic conversion built by Jay uses a 1hp/1.25gpm power unit with a 3" cylinder and 24" stroke for completing a 90* bend. While this will get the job done, the design is not optimized. At 1.25gpm, it will take about 35seconds to complete 24" of stroke with a 3" ram for that 90* bend whereas the optimized design above will do the same bend, using the same power capacity and overall bend torque capacity, about 10seconds faster. Also because the cylinder in Jay's design is mounted at a shallower angle with respect to the bender arm, it initially requires a much higher force to produce the needed bending torque, which necessitates a larger diameter cylinder and puts increased loads on the bender and it's pins.
Is this the "right answer"?
There is no "right" answer. Some answers are better than others, but there are infinite ways of doing this, many of which will give close to the same performance. This is a system where many different elements may be fixed based on availability of components and one or some other parameter may need to be adjusted to optimize for a given set of equipment, so if you have a particular constraint such as already have a particular cylinder or pump or that only a certain size is available of something, post up and we can try to work through making the most of that component.
Can it go faster with smaller tube?
The design above is based on a fixed geometry and fixed displacement pump. Assuming you are not changing the motor, if you are bending smaller tube, you can trade the lower bending torque required for a faster speed. There are a number of ways to do this. For example, you could provide alternative mounting holes in your bender and frame so the cylinder can be moved closer to the center of the bender (if you do this, be sure to maintain the angles mentioned above). You could also use a variable pulley system between your pump and motor, so you can increase the flow rate when bending smaller tube.
Additional calculations to support above design. msg your email address if you want the Excel file.