stepper pulse rate idea

[Ray Henry <rehenry-at-up.net>]

List

Those of you who have applied the EMC to steppers that run from the parport 
know that you have to design into the machine a compromise between speed and 
resolution.  

This compromise is fixed in part by Jon's PPMC stepper rate generator card.  
This card goes a long way toward expanding the range of stepper powered 
machines to which the EMC can be applied.  (this is not just a blatant 
advertisement but an attempt to clarify where the rest of this post is going 
or not going.)  I'm thinking here of a direct parport driven system.

What I'd like you to think about is a two scale pulse train for each axis.  
One pulse train would run at a low rate for high speed travel, the other at a 
high rate for low speed resolution.  Let me suggest a couple of ini variables 
because that may be the easiest way to think about this.

INPUT_SCALE_RAPID =  5000 0000
OUTPUT_SCALE_RAPID = 5000 0000

and 

INPUT_SCALE_FEED = 50000 0000
OUTPUT_SCALE_FEED = 50000 0000

As I see it we would not change the resolution of the axis hardware itself.  
We would change the resolution that the EMC sees or imagines to se out there 
at the machine.  I think this thought experiment has software, hardware, and 
value components.


1 - Software

What kind of a nightmare would we create for the software if we ran the 5000 
step scale for any rapid move and the 50000 step scale for any feedrate move? 
The example is deliberately exaggerated to show that there will be a rather 
slow feedrate above which you can not cut.  

Some sort of scaling would have to happen in several places including 
accel/decel, following error, position, world model, etc. 

Some sort of output would have to tell the hardware to change the pulse 
scaling.


2 - Hardware

I spoke with Mariss at Geckodrive this morning about applying this notion to 
his step and direction servo with the pulse multiplier card (G340).  His 
initial response was, "don't even go there."  I suspect that pretty well 
matches the initial thinking of most the readers of this post.  

As you can see from the scale numbers above this is a ten to one ratio.  This 
is one of the ratios available with the Gecko card.  Others are two and five 
times the input rate. 

Mariss listed a dozen reasons why it couldn't be done and then came up with a 
couple of ways that it could.  In fact he even thought about building a new 
pulse multiplier card that would do it.  In typical "Ray steps in IT again" 
fashion, I offered to be an alpha/beta test site for such an animal.  There 
would need to be some small delay (70 ms in worst case) for the pulse 
multiplier to complete it's work before the pulse train from the other scale 
begins.

Although I have not spoken with Jeff at Zylotek, I suspect that it would be 
as easy to accomplish the same thing with that board except that the scaling 
would be full, half, quarter, eighth and you would be dealing with 
microstepping rather than servo stepping.  You could do the same thing with 
the G210 from Gecko. 

Or we could simply build a small pulse multiplier card that would do the work 
somewhere between the PC's parport and whatever step and direction device we 
have out there.  Matt would be incined to build one of these into one end of 
a parport cable or one of those little Radio Shack db25 jumper boxes.


3 - Value ( your chance to vote or express your thoughts)

3 - a  Do you see any value to such an approach to expanding the range of 
stepper speeds that the emc can reach?

3 - b  Assuming that we can accomplish the hardware side of this in one of 
several ways, would it be worth doing on the software side?    

3 - c  Ray needs to reduce his caffeine intake?

(snip replies to save)

Ray