开云体育

Steve:

The debounce inetrval now applied to the index pulse. This means that when
the index sigmal goes active, it must stay active for the number of debounce
cycles. Lets say your dedounce is set to 20, this means that any signal,
including the index signal, must be active for at least 20 interrupts or it
will be ignored.
So, if your using 25Khz, then one interrupt is 40us and times 20 is 800us,
so the width of the index pulse must be at least 800us to be recognised.
This should elimiinate any noise or reflections from triggering it.
If you have a 1ms pulse width, then it should read fine.
I think I will have to give you a separate debounce interval for just
the spindle beacuse the width of the pulse will, of course, vary with the
RPM of the spindle. This means at high speeds, you may not read the index at
all if the debounce is too high.

I'd be interested in what your spindle RPM now reads and if its more stable.
Thanks,
Art
www.artofcnc.ca

Hi Art;

Must be going blind. Looked at wrong place in dialog box.

My Bad

Bob Harvey

Hi Art;
Dwell option doesn't appear in my logic tab dialog box.
Only word Units is present on my machine.

Looking good
Bob Harvey

--------- snip ----------

Good man Art !
Jim

On Thu, 04 Sep 2003 19:27:39 -0300, you wrote:

7) Index signal is now debounced to eliminate spurious noise. This means the
width of a spindle index must be equal to your interrupt period * Debounce
interval before it is assumed to be valid.

Hi Art - thanks for that, can you explain what you mean here and where
you set it?

If it's the same as microswitch debounce interval on logic screen, are
the values for spindle going to suit switches?

--
Steve Blackmore

Hi All:

Well, whata we got this week....(I'm now keeping a better filing system so
I'll list from my copius notes...

1) Close file button added.
2) Reload last session file button..
3) BackLash error fixed. (Thanks Isak)
4) Rutex stability in PID variable readouts.
5) CV-OFF LED OEM function is added (OEM #56)
6) Not pressing enter when entering a jog incr will no longer cause a 0 to
be entered. (Which was crashing the program)
7) Index signal is now debounced to eliminate spurious noise. This means the
width of a spindle index must be equal to your interrupt period * Debounce
interval before it is assumed to be valid.
8) Dwell G04 may be in milliseconds or seconds. Selected on the logic
screen. Small bug to be fixed for next version is that you must restart
program after changing this value.

Thanks,
Good luck, (Special driver has been updated for it as well.)
Art
Art
www.artofcnc.ca

Here is a formula for CONSTANT acceleration (Had to dig out my old
dynamics book)

X = .5 (Vo+V)*(V-Vo)/a

X is the distance that the axis will move
Vo is the starting Velocity of the axis (0 if the axis starts from a
stop)
V is the Final velocity (The rate that you want to be at)
A is the Acceleration

If you all think that this looks good I will do an Excel sheet tonight

Brian

PS I will take care of the units in the spreadsheet

Hi All,
I have two questions:
1. When tuning servos with Mach2 and Gecko drives can someone give
me a proceedure for setting Gain, Damping, Max Speed, and
Acceleration?

2. Can anyone give me a non-calculus approximation of distance for
starting and stopping with a given acceleration.

(As I was typing this it occured to me that Excel might do the
calculus; anyone tell how to set up the formula?)

Thanks, Jerry

Ron:

I will look into offering a settable units for dwell on next release which
should be by Sunday or so of this week.

Thanks,
Art
www.artofcnc.ca

Don:

Unfortunately, there is no way to add a second port to a laptop. (Just one
of those things.) Are you running on batteries? Perhaps the current level
has dropped? If on a Power Supply that can't be the case of course.
Is is possible that something else is interferring, or a ground wire is
loose, this can also affect the available current and make pins drop out and
such. Pin 1 and 11 are notorius for this as they have alsways provided less
current capability than pins 2-9.

Thanks,
Art
www.artofcnc.ca

--- In mach1mach2cnc@..., Steve Blackmore <steve@p...>
wrote:

Have a try with that RS opto if you can, it's cheap and should work
fine.

Have ordered two plus two different reflectives to try (219-2634 is
Omron SY410 and 455-0969 is Optek OPB760T)

I wish I had a fast digital scope with advanced triggering, maybe

then

I could figure out what is happening :(

Sadly I'm still in the Tek 465B (100 MHz) era - thought it was "the
last word" when I bought it - is still working though and very
reliable delayed triggering off a level if repetition rate is high
enough to actually see the trace!

John Prentice

On Thu, 04 Sep 2003 14:06:17 -0000, you wrote:

If you are slowing the edges of the waveform with the R/C network
then a schmitt after the filter would square them up. A little noise
with the waveform going slowly through a switching point with no
hysteresis might give hi lo hi when sampled every 40 uSec. The
software debounce will probably do a better job anyway.

I'll leave it for the minute then.

I will try very hard to rig up a step/dir servo spindle to join your
tests - it must be lonely!

It is <G>.

Have a try with that RS opto if you can, it's cheap and should work
fine.

I wish I had a fast digital scope with advanced triggering, maybe then
I could figure out what is happening :(

--
Steve Blackmore

--- In mach1mach2cnc@..., Steve Blackmore <steve@p...>
wrote:
<snip>

Hi John - it is a Schmitt trigger device, it's RS part No 304-560.

I was thinking of reducing the iF value on the diode a bit to see

if

dimming would help.

I'll see if Art's debounce works, in the meantime it's easy enough

to

solder those on to the board and see what happens ;)

If you are slowing the edges of the waveform with the R/C network
then a schmitt after the filter would square them up. A little noise
with the waveform going slowly through a switching point with no
hysteresis might give hi lo hi when sampled every 40 uSec. The
software debounce will probably do a better job anyway.

I will try very hard to rig up a step/dir servo spindle to join your
tests - it must be lonely!

John Prentice

On Thu, 04 Sep 2003 10:17:30 -0000, you wrote:

Getting bored sitting at my screen here and typing, I chip in! ...
Sounds like a case for a bit of old fashioned R/C filtering on the
pulse. Art is sampling every 40 uSec (or less), you have a pulse 3.7
mS wide if my calcs are right, so (low pass) 4K7 series resistor and
0.1 uF capacitor on input of a schmitt trigger gate giving a time
constant of 500 uS might help?

Hi John - it is a Schmitt trigger device, it's RS part No 304-560.

I was thinking of reducing the iF value on the diode a bit to see if
dimming would help.

I'll see if Art's debounce works, in the meantime it's easy enough to
solder those on to the board and see what happens ;)

--
Steve Blackmore

Hi,

I am currently using CNCPro to run my plasma table. I would like to
convert to Mach2 but have run into a bit of a snag. CNCPro uses
milliseconds for the G04 pause time (ie G04 5500 for 5.5 seconds)
and Mach2 uses seconds (ie G04 5.5 for 5.5 seconds). I have a large
library of gcode files that we use and don't want to have to
reprocess them. Is there any way to modify this in Mach2? Or how
about a batch process way to change the G-code files?

Thanks

Ron Routledge

Steve:

It is not possibel to trigger on both edges. Mach2 sets an
internal flip-flop when the leading edge is encountered and flips
back to sensing when the trailing edge is hit.

BUT, the software debounce is turned off only on this signal.
Perhaps I should try adding this signal to the debounce circuit, I
removed it in the design phase thinking that it may screw up
calculations, but now that I think about it, the calculations would
be the same, only lagging slightly in time. Mach2 is probably sensing
noise.

I will add debounce to the current development version, I'm willing
to bet this will fix the problem....

Thanks for the testing, great results really...
Art

Hello,
Does anyone know how I can add another port to a Compaq laptop?
I was trying the demo version of Mach2 and every thing was working ok,
then my X+ limit switch started acting up. Everything seems to be
working with the exception of pin 11. That was the pin for X+ limit
switch. All other pins seem to be working but I can?t get the motors
to run as smooth as they were. Any help or suggestions would be
greatly appreciated.
Regards,
Don

--- In mach1mach2cnc@..., Steve Blackmore <steve@p...>
wrote:
<snip>

Moving the slotted disc very close to the detector gives a "low"
reading, moving it towards the emitter gives a higher reading for

the

same selected motor revs. We are talking +/- 80 rpm!!
The only explanation I can get from this is Mach2 is so sensitive

it

is triggering sometimes on the leading and trailing edges of the

same

pulse giving spurious rpm readings. The actual spindle speed is NOT
changing.

I tested this by setting the speed to 1000 rpm. Moving the disc has
less effect, still there but only 20 rpm or so.

The sensor disc slot is about 2mm wide on about a 40mm radius (not
measured). The only options, in order of preference, I have is,
de-rate the emitter voltage, close the gap or change the sensor and
hope they help, but my gut feeling is that Mach2 maybe too

sensitive?
<snip>

Getting bored sitting at my screen here and typing, I chip in! ...
Sounds like a case for a bit of old fashioned R/C filtering on the
pulse. Art is sampling every 40 uSec (or less), you have a pulse 3.7
mS wide if my calcs are right, so (low pass) 4K7 series resistor and
0.1 uF capacitor on input of a schmitt trigger gate giving a time
constant of 500 uS might help?

Ok back to work,

John Prentice

On Wed, 03 Sep 2003 14:34:14 -0300, you wrote:

Thanks, the numbers help. It seems to indicate that the "normal" spincount
is too low and the current is almost 3 times the normal. This means a 3:1
reduction in speed is being sensed. This is obviously wrong and probably a
bug on my part due to spindle speed step/dir. I suspect the delay is
causing a screwed up number being computed.

Hi Art - just a follow up on painting the sensor disc, after that was
done Mach2's measured rpm changed slightly for the same spindle speed
settings, measured rpm dropped, which hints that the sensor was
picking up reflections. I can't see them on the old analogue scope, I
can only get it to trigger on the major pulses.

Moving the slotted disc very close to the detector gives a "low"
reading, moving it towards the emitter gives a higher reading for the
same selected motor revs. We are talking +/- 80 rpm!!

The only explanation I can get from this is Mach2 is so sensitive it
is triggering sometimes on the leading and trailing edges of the same
pulse giving spurious rpm readings. The actual spindle speed is NOT
changing.

I tested this by setting the speed to 1000 rpm. Moving the disc has
less effect, still there but only 20 rpm or so.

The sensor disc slot is about 2mm wide on about a 40mm radius (not
measured). The only options, in order of preference, I have is,
de-rate the emitter voltage, close the gap or change the sensor and
hope they help, but my gut feeling is that Mach2 maybe too sensitive?

What do you think?
--
Steve Blackmore

2 each: Compumotor Zeta57-102 Stepper Motors and Compumotor ZETA 4
Drives. Components are in like new condition and include all cables
that are unique to the motors and drives along with a manual.

If interested make an offer to servowizard(at)cox.net.