Advanced Search

The issue with pendants not working in 2.9.5 and 2.9.6 has been identified and a fix has been pushed.
However there is currently a problem with the buildbot so a manually build 2.9.7pre .deb file is available here:
forum.linuxcnc.org/21-axis/57364-axis-co...-9-6?start=10#336658

1. What daughtercard is used to drive the step/dir outputs?

2. A relatively common cause of parallel expansion failure is voltage exceeding the -0.5 to +7V range applied
to any I/O pin (even for an instant). One easy way to do the is to plug in the P2 connector when the system is
powered and there  is no common ground reference between the 7I97T and the connected device.

3. When damage occurs, the typical damaged component would be U22, this can be checked with mesaflash
assuming W21 is in the up position and P2 unconnected, running:

mesaflash --device 7i97t --addr [card_ip_addr] --rpo 0x1008

Should respond with 0x0001FFFF (all P2 pins read high)

if not, its likely that U22 has been damaged

(U22 is ~$1.50 in ones and relatively easily replaced if you have hot air tools)

Hi new here.

Using linuxcnc 2.8.4-1 and having issues with g38.5 commands.  The probe I am using is a magnetic encoder on a rail and spring system to measure distance.  My hal is setup so that a displacement of >0.005in will trigger the probe input.  The gcode starts with a g38.3 to probe the piece to get its location.  Then using g38.5 command to determine the edge of a part by moving in z until the probe falls off.  I've had issues with "probe tripped during non probe move" so I am controlling the activation of the probe input signal with a digital io pin.  I am still getting "probe tripped during non probe move" errors and confused as to why it is still happening.  

Hal config:
#Negate Encoder Position
setp mult2.2.in0 -1.0                    
setp comp.0.in0 0.005                   # Threshold value for encoder
net enable-probe    <=    motion.digital-out-03    =>    and2.1.in1

# Encoder position and comparator
net negated-encoder-out <= mult2.2.out
net negated-encoder-out comp.0.in1       # Connect to comparator input
net encoder-threshold-met    <=    comp.0.out   => and2.1.in0

# AND logic to trigger only during probing motion
net probe-trigger and2.1.out motion.probe-input

Gcode:

M64 P3    (enable probing) 
G38.3 F15 Y1.6  (probe vertically into the part)
G38.5 Z4 F25 (drag off edge) 
M65 P3    (disable probing) 
G4 P0.1 (dwell to make sure the signal was sent to disable probing)
 

Now you have have a label with a LED that has a HAL pin so you can monitor just about anything.



JT

Thank you. I used that however when I get to freecad it doesn't want to engrave all the edges that don't form a closed loop. Even if I separate the groups it soems it still doesn't see closed loops as closed.

I'm missing something.

Thanks again

Thank you guys for the quick fix...  

cheers miniwini

I’ve been working with a Mesa 7i97Т (Ethernet) board for several months, building a custom machine for forming stainless steel strips (bender).
After long tuning and debugging, the system was finally running perfectly — all axes worked, PID loops were tuned, encoder feedback was stable. Everything was ready for production.However, at the very final stage, a strange and critical issue appeared:
→ All step/dir outputs stopped working.
The configuration hadn’t been changed, firmware is the same, LinuxCNC starts normally, encoder feedbacks are alive, analog channels respond, isolated I/Os work — but there are no pulses on stepgen pins anymore.I’ve checked:

• power rails — OK (5 V stable);
• FPGA communication — OK (hm2_7i97.0.readhmid works, pins visible);
• configuration and HAL wiring — unchanged, verified against backups;
• Ethernet communication — stable (no packet loss).

It really looks like something went wrong inside the P2 connector circuitry.
There is no visible damage, but all signals related to the P2 step generators are either completely absent or there are slight jerks on some axes.The issue happened suddenly, after months of correct work.
That’s why I suspect a hardware failure — possibly in the P2 line driver section of the FPGA or the buffer network.I’ve dedicated months learning, configuring and testing this system — and it’s extremely disappointing that such failure occurred when everything was already finished and tested.
Before I try to order a replacement board (delivery will take weeks), I’d like to understand:

• could this be confirmed as a hardware failure on P2?
• are there any known cases of P2 buffer/transceiver damage on 7i97?
• can I temporarily bypass stepgens via another interface (for example, through TB4 SmartSerial → external 7i85S module)?
• any diagnostic HAL tests to check P2 FPGA I/O status?
  
  I really liked the features of this board, but I am also very disappointed at the moment.

Any insights from Mesa developers or users with similar experience would be greatly appreciated.
Thanks in advance.

 I always think of them as rotor blades like a helicopter. That said I will have to give it another try.
The last time I tried it they looked terrible. Lets just say they needed a lot of sanding, to say nothing
about being balanced. I should have taken a picture of them as a what it shouldn't look like. LOL

Moved to CAD/CAM.
-
If i understood correctly:
Group (CTRL+G) = a group of separate things
Combine (CTRL+K) = one thing with separate stuff inside
So try with Combine. In Inkscape.

I have a drawing of the outline of a dog (our Rodesian Ridgeback Koa) great dog.  I first used Inkscape and did a trace bitmap to outline the edges.  Then it suggested grouping the separate edges that were not and shouldn't be connected to make it one group of nodes.  I then imported it and was able to extrude it to give it some shape.
then the fun begins.  Every time I try to combine the segments (node to node) all I get is "not a closed loop"  some are and some are not and shouldn't be.  
I think I am missing some understanding of just how Freecad looks and treats some of these segments.  I like to create drawings and be able to import them into engravable or cut out objects.  I work with a lot of artists who look for this sort of capability.

Would greatly appreciate some insight as to what is going on and what I am not understanding.

Many thanks
Routerworks

Hi all, first post.

Could anyone point me to the right direction about using a 446 based board (Fysetc S6) with no SD card slot?

What would be the best, going with static config approach (I don't know how to do it or use which version/fork) or using an SD card adapter and have them talk over SPI? Using SDIO/SDMMC interface would not be so straightforward I guess.

Many thanks to those who made this project real. It was a fun read and I've learned so much. (Not a software guy)

mdi(string) send an MDI command. Maximum 254 chars

MDI executes the command

program_open(string) open an NGC file

Open loads a file into memory but does not run the file.

JT

Yes, you can put home zero at the middle of table, config the x and y size as +100 and -100 if the table is 200 mm. Put a mark on the x and y gentry as zero point. That will save lots of time to home the machine z is always at the top.

I made a detailed study of the probing.
I didn't quite understand it. Here is a detailed contact.
 
1. Start at the current position
2. Probe down with Z axle's max velocity (4000 mm/min)
3. Probe Height
4. Run with Probe Speed (600 mm/min)
5. Contact (voltage lower than limit). Brake with Z's max acceleration
6. Go up at 1 micrometer / cycle
7. Contact release (voltage higher than limit). This sets the material height.
8. Go up with Z axle's max velocity
9. Pierce height (measured from the new material height)
10. Piercing starts (voltage goes high) after 0.6 secs?.
11. Pierce Delay seconds after Pierce start
12. Go down with Z axle's max velocity
13. Cut Height
14. Cut for 5 mm at cut speed.
15. Arc has stopped. 
16. Go up with Z axle's max velocity
17. Safe Height

The big take-away from this is that the material height is set when the probe goes up.
Focusing on contact criteria is not the right way.
Focus on non-contact criterium.

Also, to reduce the sensing time, one wants to
- have probe height close to the material
- use not very high probing speed, because of the distance before Z can stop
- use insanely high acceleration on Z
one gets punished by the probe-up speed being only 1 um/cycle.
Overshooting a few mm, which can happen with too high probe speed and low acceleration takes a few seconds to come back from.

I also wonder about the current method to set material height on probe-up.
Admittedly I have only looked at a few hundred contacts so I have not seen everything.
But what I have seen is that contact is consistently found and non-contact, release from contact is more hesitant.
For example here is a contact where the tip went into a water puddle on the table.

 

It failed to register ohmic sensing contact on probe-up. Would have had no problem to register the contact.
I can see it makes sense for the float switch to register material height on probe-up.
For ohmic sensing I think it is more straight-forward to register material height on probe-down.
I'll see if I can find my way in the plasma.comp code and make this mod and see how that works.

If you are talking about pncconfs scale calculator, the steps per turn number
is intended to be the number of full steps per turn. For almost all step motors,
this number will be 200. So for a standard drive you would leave the "Motor steps
per revolution" at 200 and then say if your drives are set for 16 microsteps per full step,
you would set the "Microstep Multiplication Factor" to 16. This would result in 3200
microsteps per revolution.