John Kasunich wrote:
> Jeff Epler wrote:
>> [thread moved to emc-developers]
>> On Thu, Aug 14, 2008 at 09:17:31AM -0600, Sebastian Kuzminsky wrote:
>>> Peter suggested a fix for the 5i20 stepgen problem. I've implemented
>>> it, but I dont have access to the machine with my test hardware so this
>>> is basically untested. It compiles...
>>> If someone wants to try it out before I test it tonight, it's available
>>> here (http or bzr):
>>> This is against CVS TRUNK, though the change should work identically on
>>> 2.2.6. The only change is in src/hal/drivers/mesa-hostmot2/stepgen.c.
>> This does allow the step generator to move, but it behaves oddly. Here
>> are the links to the two halscope screenshots I sent earlier on irc:
> I'm just guessing based on the regular stepgen, so this may be totally
> wrong. Eric's earlier observation that the mesa stepgen doesn't export
> a max-accel pin surprised me, because the "position loops" built into
> stepgen _need_ an accel limit to do their math. I don't know if Seb
> implemented position mode the same way that stepgen does it (it would be
> nice if he did). But the velocity strangeness in your second scope pic
> seems like something that could happen if the accel limit was way off
> (or missing).
I intend to model the hm2 stepgen behavior on the software stepgen, i
just haven't finished implementing it yet. What's in there is, ahem,
pretty rough. I'll make stepgen a priority and it should get pretty
complete in the next week or two.
I appreciate all the feedback.
Computer Science for life, that's my direction
Instead of b-balls, my homies throw exceptions -- MC Plus+