Advanced Search

Question. If I use speculative memory access via _mm_prefetch (SSE instruction), does it affect the Linux CNC kernel? I am using this on RAW L2 Zero Copy sockets. I ran it, it seems to work, Axis (Linux CNC) doesn't break. Are there any restrictions on what can be used programmatically when writing drivers for Linux CNC?

Question 2.

Trying to free up the high speed encoder function on my 7i96s by moving the pendant wiring all over to a 7i84u and might as well move all of the other pendant wiring over at the same time.

I read that if I run the 7i84u card in "mode 2" there are 2 spots on TB2 that set up as encoder channels (inputs 16-19)
Would that be two separate A/B single ended encoders or is that one A+/A- B+/B- differential like I have on the 7i96s ???   The pendant MPG has differential output

it also says that the thresholds automatically go to 2.5v to be compatible with MPGs.
Is that's regardless of field power level or not?

Currently I have Field A and B both at 24v, my original thought was  to use the split field power option and have field A be my 24v and move all of the axis selection, jog rate, e-stop pendant inputs  as well as the existing prox limit/home and output onto TB3
Then make field B into 5v and move all of the pendant wiring over from the 7i96s to TB2. I could tap into the VFIELD power input block TB1 for my power and ground for the pendant. I end up with a pile of 5v IN/OUTs that I'm not sure what to do with but maybe later


For switching to MODE 2, software selectable so from what I understand:
what I currently have in the HAL
loadrt hm2_eth board_ip="192.168.1.121" config="num_encoders=1 num_pwmgens=1 num_stepgens=3 sserial_port_0=0xxxxxxx" 

changes to:
loadrt hm2_eth board_ip="192.168.1.121" config="num_encoders=2 num_pwmgens=1 num_stepgens=3 sserial_port_0=2xxxxxxx"  



Anyway, looking for any input before I blow up my second 7i84u
Thanks





 

Oh, I did not realize that there is a pcnconf utility. I used stepconf for my old machine, so hopefully this will work as well.
Deleting all the estop latch stuff did not work.
I will give pncconf a try next week.
Thank you for your help so far!

Not sure but something about the estop latch is is not working.

I would probable delete all the estop latch stuff as it's really not
needed here.

You might also be better off starting with a pncconf configuration


 

So ive tried the new repo code and it still hasnt enabled dc-phase or shown any value for PLL-ERR or PLL-out except 0.

Im not sure if i plotted it correctly because i couldnt find the servo thread in halscope (is it servothread.time? ). However, please see attached image for pll-err and app-phase like requested.

Sync0Shfit was changed to 300000 aswell like you suggested

Mesa Hostmot2 supports homing to index on step/dir systems.
It requires firmware that supports step/dir Index (normally
such firmware has "SDI" in it's name). It also requires one free
input per step/dir channel for the index signal.

This allows homing to index  and spindle synchronized motion  on
step/dir systems without encoder feedback to LinuxCNC.

I will give it a shot. Thanks for the Feedback regarding my files 

Thanks! Unfortunately I don’t have time to test this today, but will do it tomorrow at the earliest or Monday at the latest.

@ Atsu

I made a small change in my repo, get the new code, and make a new build then test, see if pll-err and pll-out get a value this time.

Hi, just a note if you tryto use the on-screen keyboard, this will not work.  It results in:

This is due to an unmet dependency.  is included, but not .  Install that package also and onboard will then work.

grandixximo wrote:

@atsu

Your servo drivers support a period of 1250000

????

Yes, they should. The manual states the following:

In the work mode, the synchronization cycle must be an integer multiple of 125 μs. Otherwise, the servo
drive will report Er74.0 (EtherCAT synchronization cycle setting is incorrect.)

Initially, I ran the drives with a period of 1000000 ns, but increased it to the current value couple months ago. Did it because in some rare instances I got "Unexpected realtime delay on task 0 with period 1000000" error in UI. I can try lowering it back to 1000000 ns to see if it makes any difference.

@TAKUYA

I see that you did not set pdo 1600 for your X drive was this on purpose?

I don't think the EK1100 needs the dconf line at all, but that should not effect, your DC seems ok, can you open the halscope, plot pll_err and app_phase? also servo-thread

also Sync0Shift should be half or 30% your cycle time, not 0
 

forgot to mention granite devices have a good article about feedback systems and even mentions Linuxcnc
granitedevices.com/wiki/Dual-loop_feedback_position_control 

@atsu

Your servo drivers support a period of 1250000

????

Hi ihavenofish,

I am also in process of retrofitting a TC225, however I am in a different situation where the machine was completely gutted and I am starting from (almost) scratch. When you get a chance could you send us that pinout? I have the Oriental Motor brake pack and gear motor and am confident in that part, but I haven't been able to find any information on the encoder. All I have right now is a harness that has 8 wires with the following labels: 49, 50, 51, 52, 53, 54, 55, and E. My next step was to start probing each wire to determine which ones are 0V, +24V, and the 4 BCD signals. E is obviously earth ground.

I appreciate it!
Jeff