Re: [Emc-developers] hostmot2 stepgen bug

[Chris M. <chr...@ho...>]

> 
> The problem is its fairly rare, and quite system dependent. I cant duplicate 
> it here even with a base thread, And I dont think Sebastian was ever able to 
> duplicate it.
> 
> Do you still have the very large dir change setup/ hold numbers?
> 
> I suspect they are related in pushing the stepgen correction over the edge
> 

It may be the the two systems I used were particularly sensitive.
Maybe most other always set maxvel and maxaccel to values other then 0
As I said for me even the unchanged sample config would fail.
one other fellow messaged that if he set the base and servo thread just right
it would work fine. (kinda suggests a timing problem as you said)

At some point hostmot2 changed and the problem showed up for me.
I just can't remember now which version of EMC was fine and which was not.
your now talking almost two years ago.
It may sound that I'm complaining, I'm not.
I'm just saying the problem was reported before and has been around for a
while. I reported, and gave any info that was asked (as far as I remember).
setting the max setting as you suggested the last time works for me without
fail and thats good enough for me and PNCconf

I'm not at my machine to check the step timing numbers.
but I would guess that the last config would have timing values of all 1000ns.
I originally had very large numbers because the OPTO 22 opto I was using
were very slow. (I was just testing at the time)
I now removed the optos and directly drive the step controller with the
mesa 5i20. The PARKER step controller has it's own optos.

Cheers Chris M