Only Z axis is working / HP DC 7600 Tower

Dear Community

I had a working Plasma cutter until I upgraded to 2.8. I wanted to integrate PlasmaC and the Proma THC150.
After the upgrade the hard disc made problems. Finally the latency when from under 12 500 to over 120 000. The HDD is onboard, not reasonably replaceable.

After having seen two HP DC7600 in the latency table with good values I bought one (Tower with the Intel 940) .

Debian with Linuxcnc 2.8 was installed. Most function in the Bios are deactivated, I do not have a lot of possibilities in the given Bios. No option about HT etc ...
Ram was increased from 1Gb to 2Gb, this had little impact on the max jitter.
The 25um max jitter is around 56 000 and the 1um is around 37 000. Has somebody an idea how to get it quicker?

Furthermore is it not possible to operate Linuxcnc with this values?

What also drives me nuts is that I only get the Z axis moved with the stepconf axis test (also with the laptop but with this it moved well down and poorly upwards). I thought it must come from the too short values for the stepper I tried different values which I found here in the forum but X and Y aren't moving. Only the value is changing in the display.
Furthermore I have now only one input for all homes and limits. And when engaging the machine I get an error message that the limit switches are triggered. I have checked the limit switches and they are working properly.

Is the stepconf working properly with 2.8 when having two motors on the Y axis? I read somewhere that this doesn't work.

May it be that the China BOB is dead?How to check this?

For over a month the machine is down and I need it for some jobs.
Your help will be highly appreciated.
Kind regards,
Jean

Upload the hal and ini files.
China BOB might be dead, but it is for sure to slow.

Morning,

I attached what I got by using the stepconf.

The ChinaBOB is very limited for sure but it did its job well over the last 4 years. And the speed of the steppers were higher than my machine was able to handle due to vibrations in the TR20x8 spindles.

Machine:
1 rev = 8mm
400steps/rev
2 limit switches per axis, no separat home.
Driver: DQ860MA
Motor: WT86STH118-6004A

Greetings,

Hello

For my plasma cutter I am using a HP Compaq DC7600 Convertible Minitower /09F0h.
It has the pentium D CPU 3,4GHz with 4Gb RAM and solid state hdd
SMI Fix is not recommended for the given version of chip set.

I have the following images running on it:
Debian 10 Buster LinuxCNC 2.9
Debian 7 Weezy LinuxCNC 2.7.15
Debian Stretch 13 LinuxCNC 2.8.0

Without playing in the around in the bios with exception of the energy saving settings, I got following latency results:

Buster without pushing the CPU 53000 / 34000 otherwise 55500 / 104700
Weezy with Office and Firefox starts 12500 / 28050
Stretch with Office and Firefox starts 28900 / 49050 Stretch has a strange behaviour, with 6 GB Ram the Latency goes down a few 1000 but only after being in operation for 10 to 15 Minutes. And it is quicker when energy saving is on in the bios.


Weezy is the best solution for this PC, unfortunately I am unable to get the twin motors working with 2.7.15. I will have to try 2.8 on this version if I don't find a solution.

The only LinuxCNC system I was able to get a least a starting hal and ini is 2.8. The stretch version is has poor latency values and is much slower to start. Some counters are running in the startup procedure. It seems to be an issue with the solid state disc.


The communication is over cheap DB25 China Bob wich served me well over the last 4 years. My key issue now is that the ALL HOMES and LMITES is not working. The pin signal has no effect. To check if it is not the Input pin itself, I tried different pins. Somehow the Input is not considered in the software.
With exception of the twin motors all motors are working correctly now in the stepconf.

Any ideas why the limit switch doesn't t work for homing?

How do you have your limits wired up? Are they ordinary micro switches or are they prox switches? Do you have all of the voltage sources connected to the BOB? Most of those cheap Chinese BOBs require 12v be connected to the board to supply the power for the inputs.

Have you looked at the parallel port's input pins with HalMeter? Can you see the input pin change when you activate the switch?

All limit switches are parallel connected and are normally open. Yes, they are mirco switches. The BOB is connected to a USB plug with 5V and with a 5V power supply. (This should protect to PC side as it is not directly connected to the CNC.) As mentioned it worked for for 4 years, only with one input pin per axis, thus Voltage should not be the issue.


I will check tomorrow morning how to use the HalMeter. I don't think that the signal is read, as I get no error message when switching the limit switch while the machine is on. I have connected also a new switch directly to the board with 20cm cables to check the function.