Advanced Search

Maybe some power or SI issue at the Flash EEPROM?

The firmware has a 5 ms timeout for page writes (3 ms max on the W25Q16)
The erase timeout is 3 seconds (400 ms max on the W25Q16)
so that should not be an issue.

I was reading the qr_auto_probe_tool.ngc code and this line:

#<calculated_offset> = [#<touch_result> - #<_hal[qtversaprobe.probeheight]> + #<_hal[qtversaprobe.blockheight]>]

Appears that the tool length, #<calculated_offset>, should equal the distance between the probe length and the loaded tool length. It is saved using

G10 L1 P#<tool> Z[#<calculated_offset>]

Which would indicate the length of the tool "should" be the difference between the probe length and the loaded tool length. We can assume the #<youch_result> as zero if we reprobe the probe...I think.

However, I get -70.(Something) For my probe length and -90.(something) For my tool length when the tool is shorter than the probe. At least in the tool table.

I think it would work better if the probe touched off and then, instead of "work piece height" which may or may not be G5? Z0, we should just be instructed to probe the Z0 point and set the offset between the tool setter and Z0 as a constant. Then just add that offset to the probe height and be done with it.

I believe the coding to include block height in every calculation is unnecessary and can cause errors. Plus, and I realize it's not Python, but the code could be more readable if there was a single constant offset then just amending the new tool length by that offset.

I may well be way off base here, because I am new and the code is different than what I am used to. But, I think the height setting could be simplified a bit. Just my opinion. Not trying to step on anyone's toes because it's obvious they put a lot of work into it already.

Ran
./mesaflash --device 7i92 --addr 10.10.10.10 --verify ~/linuxcnc/firmware/7i92/TopEthernetHostMot2.bit

about 10 times passed every time.

Just for completeness tried verifying against another bitfile (not written to flash) and failed as expected.

Hold & WP are tied directly to 3.3v CS is tied hi via 10k resistor.

If the main clk is out, part is marked as 50Mhz, could this throw a spanner in the works ? But then that would affect the Ethernet Interface's 25Mhz clock.

It feels like the write is taking longer than expected, but checking the busy flag would mitigate this. That's why I thought of putting a delay between mesaflash wiritng a page and the "packet read for board syncing"

Thanks a lot!
If someone need it the code is: 
rtapi_module_param

D

This helps..
What is OUTPUT_SCALE = -10800 ?
This config is for fixed motor:spindle transmition ratio.
I have 2 selectable speeds after encoder-motor
These 2 speeds are selected by hydraulic electrovalve that switches gears
position 1 max rpm is 760rpm and position 2 max rpm 3060

How to do it with only 1 button that has function cycle start / pause / resume ?

Hold on a second. I lied. Not intentionally. My spindle motor is a DYN4 servo motor geared 1:1.8 and it has a simulated encoder signal generated by the servo driver. That is the signal I am using for the coordination for rigid tapping. The sensor on the slotted plate it just for measuring RPM.
It's been a few years since I set this up and when you mentioned more slots I thought "I knew that, why didn't I do that?" Then I looked deeper and remembered how it really works.

To actually do this with a slotted plate and sensors you would want more slots and two sensors. Read up on how quadrature encoders work.

As long as you know the ratio between the spindle and encoder you can make almost any arrangement work.

These are my .ini settings.
[SPINDLE_0]
P = 2.0
I = 4.0
D = 0.0
FF0 = 1.08
FF1 = 0.0
FF2 = 0.0
BIAS = 0.0
DEADBAND = 0.0
OUTPUT_MIN_LIMIT = -5400.0
OUTPUT_MAX_LIMIT = 5400.0
OUTPUT_SCALE = -10800
ENCODER_SCALE = 4551.111111
#((2048 * 4) / 1.8) = 4551.111111
#The maximum spindle speed (in rpm)
MAX_FORWARD_VELOCITY = 5400
#The minimum spindle speed (in rpm)
MIN_FORWARD_VELOCITY = 50
#This setting will default to MAX_FORWARD_VELOCITY if omitted.
MAX_REVERSE_VELOCITY = 5400
#The minimum spindle speed (in rpm)default to MIN_FORWARD_VELOCITY if omitted.
MIN_REVERSE_VELOCITY = 50

Sorry for the confusion. I hope this helps.

Can it verify repeatedly without error?
If so if may be something related to write
(how are the WP and HOLD pins terminated?)

It's important to keep in mind that joints (machine movement) and axes (DRO values) are not the same thing. As the X axis is supposed to be perpendicular to the Y axis it would be somewhat strange for the X value to change when moving along Y.
Also note that the forward kinematic is only used to calculate the axis values from the joint home position right after homing and probably when switching kinematics. After that only the inverse kinematic model is used to calculate the joint positions for given axis values.

A little update, changed the EEPROM from a 16MB ST part to a 16MB Winbond part (same as I've used for the EPP\SPI board). Initially no change.

Put a 5ms delay between before the "sync" instruction when erasing & writing a bitfile. Was able to write a bitfile, not everytime, but when it does it verifies ok.

Ran 100 iterations of the Linuxcnc logo gcode, one after another in a loop, and then left it, after 12 hours no comms issues.

If you have nothing connected to the 7I97T but 5V power and any analog
output is not  close to 0V, the analog output circuitry has been damaged somehow.

This may occur if an output was inadvertently shorted to say +24V or
connected before a common ground was established between the drive and
the 7I97T.

OK, that simplifies things, a lot.

Actually if you use simtools there are multiple interfaces.
Serial, udp, write your own...

Every game that is supported by simtools then has the same interface.

I'm working on the Vismach implementation and while playing around i realized, that when i move the Y axis the X axis moves as well. So for so good. But the X-coordinate in Axis does not change.


Note:
X on the Mazak is head up and down in
Z on the Mazak is table left and right

But shouldn't it change? Why doesn't it ?

Kind regards,
Jochen

 

Had the same idea today for a racing/flight sim!
I have no idea what games output for motion, that would help to know, but whatever it is (and i am sure no 2 games use the same output, but maybe through directX...), we would need just translating those to waypoints and from those to gcode.
Taping into games collision detection is out of the question by now as the coding on new games is absolutely terrible and utterly useless.
But, games that have display/gauges for pitch/tilt/angle of attack, should be much easier if they also output somehow somewhere those values.
Then there is the interpreter/LinuxCNC reading those values and sending motion, but should be doable, there are some "camera to motion" things working already somewhere on this forum.
Maybe use that and point the camera at the screen...