A few more observations from testing the old model of CNC4PC card.
1. Some pin outputs perform much better than others. For example I
tested the same servopack with same cables and same motor on two
different pins. On one case after several experiments I lost no
pulses on the other I lost about 2 or 3 for every thousand.
2. I use a Dell Dimension and the logic ground pins 18-25 on the
parallel port are definitely connected to the earth. This completely
screws up the Yaskawa recommendations to segregate Earth ground from
signal ground.
3. The suggestion to use a UPS will help isolate the logic ground
from Earth and prevent ground loops IF one cuts down the earth pin on
PC plug that connects to the UPS. This ofcourse will cause your
electrician and your insurance and other safety authorities to
frown!!!
3. Digikey sells a true optoisolator that can isolate ground loops
for DB25 connectors but unfortunately it has only four channels and
it good only for serial ports.
4. Yaskawa allows to jog the motor using a hand terminal. Things work
fine but when I put the osciloscope on any terminal pin including
earth ground and signal ground I get this huge pulses at ~17KHz even
when the motor is stoped. I can pick this huge pulses with the scope
even on a thick copper wire that I am using to connect to earth! When
one hooks up all 5 motors on the breakout out board the pulse stream
combinations that I pick from the ground terminal can be seen filling
the whole scope screen.
The fact that I am losing pulses only on a few channels and not on
all of them may indicate that the performance of the old CNC4PC card
is marginal and the improved model may do the trick. I also ordered
200 ferrite clipons!
John
--- In mach1mach2cnc@..., Steve Blackmore <steve@...>
wrote:
On Mon, 29 May 2006 10:12:42 -0000, you wrote:
I did not know some computers did not have the ground connected
earth
in the power supply. Do you think this could help John?
It may, but impossible to say without knowing how everything is
wired
including the mains supply and going over everything with a meter.
Always helps diagnosing this sort of problem to do a hand drawn
schematic and then go over every connection and device.
If it's really "noise" it can only come down a wire or be EMI/RFI
induced.
I told him
previously that it could be interesting to use a UPS o a really
good
voltage regulator in the computer (and breakout board), but the
way I
see it, the problem is really related to how the servos drivers
are
hooked up.
I may be.
I also have a fully optoisolated board that isolates the grounds
(from the computer), but in any case the grounding issue will end
up
in the board. If someone thinks using one of these boards could
be a
solution, I could send one for testing.
The board I made for my lathe is totally isolated on inputs/outputs
including grounds. I deliberately did it to try and fix a noise
problem,
which in the end turned out to be a faulty driver, but it did prove
that
whatever was making the stepper move was not coming in from the PC
side.
I had convinced myself it was "noise" and wasted hours and lots of
money
on what was a 5 minute fix!
But on the bright side, there are now no connections anywhere on my
lathe or mill that connect data,signal or power grounds to earth.
Both
machines have earth bonding. All mains inputs to PC, VFD's etc all
have
filters on. 5V & 12V power supplies are double insulated with no
connection to earth either.
Steve Blackmore
--
