Re: Pulse Rate/Processor Speed
Posted by
Ray Henry
on 2002-09-19 10:15:24 UTC
Jared
This is a rather long and convoluted post but I'm thinking that it comes
round to what I see as the problem at hand for you.
I've got the EMC (BDI-2.16) running on a 166 Mhz Gateway that I
overclocked just a bit. (~200 Mhz). That box was given to me for
installing a new PC for a neighbor. Some of it, like the telepath modem
for windows just doesn't work but the basic thing is there.
I run it with KDE for Linux, the MS like window manager. It's a real
processor and memory hog but I've been able to get 6-7 KHz with
steppermod and about 2-3 KHz with freqmod. That seems to be enough for
my little Grizz mill with the 100 oz/in stepper motors. I presume that
this is lower frequency than would be possible with a dos on the same PC
but then I have the advantage of being able to do other things while the
machine is running.
IMO it may be more valuable to start thinking from the mechanical end of
things and figure out how many steps you need to get the performance you
want. Let's use the recent rotary axis posts for an example. Someone
mentioned , "If you are set up to half step the motor then one revolution
of the table is 28800 steps." Another post gave a different figure,
"Your motor assuming it is a standard 200 step unit connected to the G201
needs 2000 steps per motor revolution. So you have 2000 x 72 = 144000
steps per 360." One of these is assuming half step and one 1/10
microstep but either of them would be deadly slow on my EMC (0.8 to 4
rpm) so I'd need the ability to play freecell while it's traveling.<g>
If I were designing for a rotary axis I'd want it to take no more than a
couple seconds to rotate a complete turn. Since the Sherline is 72 to 1
that means I'd have to spin the motor about 2160 rpm. For full stepping
that's 7200 pps. Half stepping 14,400 pps and 1/10 microstepping with the
Gecko 201 would require 72 Kppm. The Gecko can do that last number
without problem but you'd be hard pressed to make that many pulses with a
quick PC.
The first conclusion I'd come to is that the Gecko 201 isn't for me but
the 210 might do the job. Then I look at the chart of torque v velocity
and see that there isn't much left for my little motors at that speed so
I will need to set accel low in order to run that fast and then I'd loose
a second or two getting to speed and back. Suddenly that little surplus
servo with a Gecko 3xx is beginning to look real attractive. Since it
has a 500 cpr encoder it would be the same 2 Kppr as a 201 microstepped
drive so I'd want the 340 with its pulse multiplier which would get me
back to the 7200 pulse rate and the old Gateway and steppermod.
The next to last question I've got is accuracy. 200 pulses per rev times
72 rotations of the motor per turn of the rotary plate means that I can
divide a degree into 40 parts or a bit less than an arc minute accuracy.
The last question I have is related to the rotary fourth axis capability
of the EMC and I'll have to study that one for a while.
Ray
This is a rather long and convoluted post but I'm thinking that it comes
round to what I see as the problem at hand for you.
I've got the EMC (BDI-2.16) running on a 166 Mhz Gateway that I
overclocked just a bit. (~200 Mhz). That box was given to me for
installing a new PC for a neighbor. Some of it, like the telepath modem
for windows just doesn't work but the basic thing is there.
I run it with KDE for Linux, the MS like window manager. It's a real
processor and memory hog but I've been able to get 6-7 KHz with
steppermod and about 2-3 KHz with freqmod. That seems to be enough for
my little Grizz mill with the 100 oz/in stepper motors. I presume that
this is lower frequency than would be possible with a dos on the same PC
but then I have the advantage of being able to do other things while the
machine is running.
IMO it may be more valuable to start thinking from the mechanical end of
things and figure out how many steps you need to get the performance you
want. Let's use the recent rotary axis posts for an example. Someone
mentioned , "If you are set up to half step the motor then one revolution
of the table is 28800 steps." Another post gave a different figure,
"Your motor assuming it is a standard 200 step unit connected to the G201
needs 2000 steps per motor revolution. So you have 2000 x 72 = 144000
steps per 360." One of these is assuming half step and one 1/10
microstep but either of them would be deadly slow on my EMC (0.8 to 4
rpm) so I'd need the ability to play freecell while it's traveling.<g>
If I were designing for a rotary axis I'd want it to take no more than a
couple seconds to rotate a complete turn. Since the Sherline is 72 to 1
that means I'd have to spin the motor about 2160 rpm. For full stepping
that's 7200 pps. Half stepping 14,400 pps and 1/10 microstepping with the
Gecko 201 would require 72 Kppm. The Gecko can do that last number
without problem but you'd be hard pressed to make that many pulses with a
quick PC.
The first conclusion I'd come to is that the Gecko 201 isn't for me but
the 210 might do the job. Then I look at the chart of torque v velocity
and see that there isn't much left for my little motors at that speed so
I will need to set accel low in order to run that fast and then I'd loose
a second or two getting to speed and back. Suddenly that little surplus
servo with a Gecko 3xx is beginning to look real attractive. Since it
has a 500 cpr encoder it would be the same 2 Kppr as a 201 microstepped
drive so I'd want the 340 with its pulse multiplier which would get me
back to the 7200 pulse rate and the old Gateway and steppermod.
The next to last question I've got is accuracy. 200 pulses per rev times
72 rotations of the motor per turn of the rotary plate means that I can
divide a degree into 40 parts or a bit less than an arc minute accuracy.
The last question I have is related to the rotary fourth axis capability
of the EMC and I'll have to study that one for a while.
Ray
> From: J <jaredts@...>
> Subject: Pulse Rate/Processor Speed
>
>
> Anyone using an older computer to drive their CNC? I am a newbie and
> really don't have a feel for how much an older pc could affect the
> maximum pulse rate and other performance issues of a stepper system. I
> was hoping to find an old surplus computer to run my system when I get
> it together, but I don't know what processor would be unacceptable. I
> have looked at some controller software and have been surprised to see
> minimum requirements over 400 Mhz. What operating system would allow
> me to get the most bang for my buck--dos or Linux (I'm sure its not
> Windows)?
>
> Jared
Discussion Thread
J
2002-09-18 19:30:52 UTC
Pulse Rate/Processor Speed
JanRwl@A...
2002-09-18 20:26:30 UTC
Re: [CAD_CAM_EDM_DRO] Pulse Rate/Processor Speed
Carol & Jerry Jankura
2002-09-18 20:29:18 UTC
RE: [CAD_CAM_EDM_DRO] Pulse Rate/Processor Speed
bjammin@i...
2002-09-19 05:20:21 UTC
Re: [CAD_CAM_EDM_DRO] Pulse Rate/Processor Speed
Steve Blackmore
2002-09-19 06:14:32 UTC
Re: [CAD_CAM_EDM_DRO] Pulse Rate/Processor Speed
Ray Henry
2002-09-19 10:15:24 UTC
Re: Pulse Rate/Processor Speed