HAL hand holding please

I am afraid it is not entirely clear to me what you are trying to do with that network of HAL functions.

Can you explain in short words?

Zig wrote:

ONESHOT and OR gates are to be connected to X and Y axis inphase and quadrature encoder signals to generate a pulse stream of constant pulse width.

That might not work. With a Mesa card config your servo thread typically runs every millisecond. There may well be very many steps in that time period, and you will miss them. You could add a faster base thread to track the step pulses, but that seems silly.

D type flipflop will take Z axis DIRECTION flag on data input and latch it high untill the next Z axis physical move is called for by the virtue of positive edge presence on Zaxis STEP signal.

That seems like it should work.
(combining posts)

My question is do I setup a new thread or tack this onto servo thread?

You either addf the new functions to the servo thread, or add a base thread and add them to that, depending on how frequently you want to call them.

How do i actually identify the input nets and connect the above to those nets ?

You identify the pins either by poring through the docs (www.linuxcnc.org/docview/html/) or by looking in the HAL config window in EMC. You can also put a "show all" line in the HAL file just after the addf block, and that will print out all the pins and threads to the terminal (only any good if you start emc from the terminal though)

Personally I think that exactly copying the hardware solution in software might not be the best plan. As I understand it you want to adjust the laser power according to current velocity. I would do that by adding a pwmgen to the Mesa card, then passing the axis velocities to a hypot hal function (hypotenuse, just sqrt(a^2 + b^2 +c^2) ) and setting the PWM value to that velocity (with the pwm scale value set accordingly)
HAL can cope perfectly happily with analogue values, you might as well use that.

A base thread of 0.1ms is perfectly feasible, in fact .05mS is common and low-latency machines can go down to 0.02mS