>
>
> It just makes difficult for any observer to understand, what is going on
> there.
>
> If it works and You understand it - very good. But remember that if
> You will ever want somebody else to understand it, then it can cause
> frustration. I think that in situations like this it is best to stick
> to KISS principle (Keep It Simple and Stupid).
>

My understanding is that the simple and most well-traveled approach for
servo systems with Mesa hardware is to use pncconf.  If you have a
recommendation for a simpler or stupider solution I am certainly open to
it.  Personally I have found pncconf to be invaluable and just to clarify,
I have not modified the ini or main hal file beyond that generated by
pncconf.  So if you have a problem with the aliasing, etc. that is an issue
with that tool and certainly not of my implementation.

If I would do what you suggest and put my changes directly in the hal file,
then every change I make in pncconf would require me to manually re-merge
my custom changes.  AFAIK, the recommended approach for custom hal via
pncconf is custom.hal (and custom_postgui.hal), and that is where I have my
custom stuff.  If the sourcing obfuscates things I apologize, but I don't
know a better way to keep configurations relatively modular and still use
pncconf.  For what it is worth, I believe in the future pncconf is going to
use the source command as well for this purpose.


> AFAIK having or not having a base thread - yes, it can make a
> difference in the realtime performance of Your PC. If I recall
> correctly, You had set base-thread to 10000 ns. I am 100% sure that
> You are getting realtime error messages.


True with that configuration I have under revision control.  I typically
run at 50000 ns and had set it to 10000 as a diagnostic measure to see if I
could increase the sampling rate.  Again, though, if I can run a base
thread with reasonable performance on all other fronts, why not do that?  I
am talking about frequency-based analog inputs for things like temperatures
and voltage levels where if I miss an encoder count it is certainly not the
end of the world.


> That is what I told since the very beginning - use encoder module in
> 5i23...
> You have loaded 4 pwmgens and 4 encoders and are using only 3 of them.
> So You a spare encoder module to attach the frequency input to.
>

I have tried to explain several times why I want a software-based encoder:
 1) Eventually I will want more analog inputs.  2) Without a custom
firmware I cannot align the encoder with optically isolated I/O.
 Additionally I'd like to try and build a solution that I can document for
anyone to use regardless of specialized hardware.


> BTW why do You have 4 stepgens loaded, if You use only one?
> That also is a little waste of resources :) If You would load 1
> stepgen, then those pins, currently occupied by 3 unused stepgen
> instances would be available as gpio pins.
>

This would help me conserve a resource that is not at all scarce to me.  I
have 4 stepgens configured because if I change the number of stepgens
pncconf wants me to reconfigure everything.  I expect I will want to add a
stepper or two in various future machine configurations and so have
reserved them in that configuration.  The non-isolated I/O alongside the
one stepgen in use goes unused otherwise anyway.

As You can see, there are bitfiles, that offer 8 encoder modules along
> with pwmgens and stepgens.
>

Yes of course.  But as far as I can tell there is no stock bitfile that
allows me to run my three main servos and encoders, isolated I/O for limit,
etstop, etc. along with isolated I/O for a number of other encoders, a
non-isolated stepgen and still make use of the Mesa servo controller card
and isolated I/O daughterboard with their specific set of input and output
pins.

Conversely, if I can get software encoders to work, I can stack
software-based encoders using one isolated input each wherever they fit,
basically giving me as many as I could ever need.

Regards,

Scott
------------------------------------------------------------------------------
This SF email is sponsosred by:
Try Windows Azure free for 90 days Click Here 
http://p.sf.net/sfu/sfd2d-msazure
_______________________________________________
Emc-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/emc-users

Reply via email to