We are working on a project where we want to have multiple stepper drivers driven by Teensy's.

I will start off by saying I am not a low level designer or developer, but I've made PCBs and written code, but nothing really sophisticated.

My design partner and I worked on two different stepper driver boards, a 2-driver board using a T3.2, and a 4-driver board using a T3.5 (mainly for the FPU, and initially thinking we needed 5V tolerance).

We designed the boards to be flexible and accommodate 3 styles of drivers:
1) The popular DRV8825-type Pololu driver footprints (socketed onboard)
2) A higher current 3A driver from Panucatt, Bigfoot BSD109A (socketed onboard)
3) Outputs from the board to i/f with external "industrial" type drivers, e.g. DMT542's (header output)

The Pololu's are fine with 3.3V signals, but the Panucatt's and most external drivers want 5V, so we incorporated 74HCT245's to manage and output 5V signals (with insights from this forum).

Paul seems to have kept (edge) pin numbering of the USB end of the short (T3.2) and long (T3.5 / T3.6) boards consistent, which sort of implies we could make a single 4-driver board design, that if a short size Teensy was used, could be a 2-driver board, using one 74HCT245, or if a longer T3.5 / T3.6 was used, becomes a 4-driver board, and needs to be populated with a second 74HCT245.

We are currently using the AccelStepper library, but in the process of transitioning to (a modified version?) of the TeensyStep library. We are not managing the low level timing and outputs, but relying on the libraries for that.

We vaguely recall reading there is some issue with doing this, but I can't find any post that clarifies / identifies the issues with this level of "plug and play" between the Teensy boards.

Can someone explain what differences / problems / pitfalls we might run into if we try to consolidate the design?

Not sure if I should ask about the T4 here or on that thread, but obviously we will have the same question about compatibility with that when available.