Advanced Search

Output scale should be set to the maximum RPM of the spindle

I thought I had, I edited the section with the IP address to my card IP address of 10.10.10.11. Obviously I’ve made a mistake, I’ll go back through it again to see.

Looking at my INI file, I see pnfconf reset my OUTPUT_SCALE = 25263, but I had it set as OUTPUT_SCALE = 1 before the test.

Mozmcks script requires editing to match your IP address setup
or it will not setup the irq mask.

Note that my simple script just sets the mask temporarily and would
need to be re-run after re-boot. In addition, if irqbalance is not disabled
the mask setting  will be eventually undone by irq balance.

The main point of my script is for testing

The i915 driver supports basically all Intel graphic chips except the GMA500 series

That kernel command line is fine

Okay, so maybe I am over my head, but I thought I had this one figured out, but I was wrong.  I followed the suggestions from several posts, including [url=https://162.243.45.186/27-driver-boards/48095-cannot-make-pwm-work-on-mesa-7i96s

Edit: I am not sure what I did, but much of my original message is missing; I have tried to recreate the just of the message below:

Okay, so maybe I am over my head, but I thought I had this one figured out, but I was wrong.  I followed the suggestions from several posts, including [url=https://162.243.45.186/27-driver-boards/48095-cannot-make-pwm-work-on-mesa-7i96s

With my old GRBL controller, I used a PWM-to-voltage module to run my Spindle, an AMB 1400 FME-P DI .  The module generated 0-10v while being supplied with 24v.  I thought the 7i96s Analog Spindle Interface would replace this module.

The only power I have in the new controller is +-5v, +12v, and +48v.  Based on my interpretation of setting hm2_7i96s.0.pwmgen.00.offset-mode to TRUE, it would allow me to supply -5v to SPIN- and +5v to SPIN+ and then shift the voltage on SPIN_OUT to 0-10v.

So, I wired it up and tested it with a DVM; the voltages looked fine, at least for the speeds I tested.  So, I decided to hook up the Spindle.  I ran "m3 s3500", and the Spindle started; I cannot verify the speed, but it seemed correct.  Next, I tried "ms s10000"; the Spindle sped up, and it, too, seemed fine.  So, I decided to try "m3 s18000", (the speed I use most often), the Spindle sped up, but based on the pitch of the noise from the Spindle I think it was running at a higher speed.  I was reaching for the eStop when the Spindle went crazy, speeding up, and then there was the terrible burnt electronics smell.  My hand was over the eStop, so it only lasted a second.  I am afraid even to try to start it, fearing the worst.

As I mentioned, I used a PWM-to-voltage module previously and thought it would be safest to return to it.  The 2-PWM bit file uses STEP/DIR 4, which is my A-axis.  I did read a post about a person using software-driven PWM on an output pin, but they were using a 7i96, not a 7i96s, and it was unclear, at least to me, how they had wired it.  I would guess they wired OUT+ to PWM+ and GND to PWM—but that is only a guess.

I am at a loss as to what to try next.

Any help or suggestions would be greatly appreciated.

I am not sure whether to scrap the idea of using the Analog Spindle Interface or try to figure out software-driven PWN and return to the PWM-to-voltage module.

I do not know where the i915 came from, this is a vanilla install from the distro and the 2.8 was the same. Ah, I see, it is in the rt_setup script pinned to the start of this thread. I did run that in both setups trying to bring latency down.

Strange, for some reason I thought the rt_setup script had already done that.  Obviously I was mistaken as that does seem to have done the trick, have the original servo_thread value and have run multiple times through the test program. 

Really can not thank you enough and sorry that I missed that in all my other troubleshooting.

Question, why and where did you get the info about using i915 in the kernel line?
Remove that, update grub and try again.
-
Pretty sure i read hundreds of times about i915 having to be blacklisted a few years back, so using it now seems really strange.
It was an intel chipset that lasted till i965 or i945, so that is superseded, probably 15 or more years ago.

Can you try pinning the Ethernet IRQ to the last processor:

 

Use by

mv pinirq.txt pinirq
chmod +x pinirq

sudo ./pinirq enp2s0

and then try LinuxCNC again

NOTE: Look at the script before you run it as with any script
from unknown sources, especially scripts that need sudo...
 

Here the results from the Debian 12 system that is not working:

And here the results from the Debian 10 system that is working (same computer, just swapping HDD back and forth):

 
Can you post the results of running this script:



to run

mv checkmask.txt checkmask
chmod +x checkmask

then

./checkmask enp2s0

besriworld wrote:

It is normal for the position to change when you close the program.  That's why we have homing or motors with an absolute multi-turn encoder so that linuxcnc can know the new position.  It may be only a few microns, but the position is changing. this is the enabling signal on my drivers. 
If the signal is missing the motor will not move.

 

Thank you very much for your reply
Yes, your answer is logical.
But according to what is in the litexcnc docs
 
I noticed that this problem does not exist in speed mode
So putting the board in speed mode when I press the estop button solved the problem
 

This is a hobby robot project with mostly Nema 17, so I was targeting 5 & 25 vdc, on a budget.

Hobby Robot

JET has been replaced with Flex GUI... gnipsel.com/linuxcnc/flexgui/index.html

JT