## Dual-needle pyVCP meter

By popular demand, a quick hack that modifies the pyVCP meter widget to have two independent needles. It's used inside the <meter> tag by specifying <halpin2>"my2ndpin"</halpin2> and hooking up something to that pin. If <halpin2> is not used meter works as before, showing only one needle.

There's an XML file for this test-panel, a short HAL-file that hooks up the pins, and a shell script to run it all here: pyvcp_dual-needle-test

The modifications to linuxcnc source required are in lib/python/pyvcp_widgets.py: 0002-dual-needle-meter-use-with-halpin2-meter2-halpin2.patch
NOTE: This is a quick hack to make it work - don't take my code/patch too seriously...

## Real-Time Tuning

I tried a number of things that are supposed to improve real-time performance, as described in this forum post.

But not much changed. This series of jitter-histograms shows little or no changes:

The things I tried are roughly

1. measure first latency histogram 0.png
2. uninstall the package irqbalance using synaptic. reboot.
3. measure 1.png
4. in /etc/default/grub modify GRUB_CMDLINE_LINUX_DEFAULT="isolcpus=1 acpi_irq_nobalance noirqbalance"  (Aside: why are the files in /etc/grub.d/ made so incredibly hard to read? Someone should re-write them in Python!). Run sudo update-grub. reboot.
5. measure 2.png
7. Add set-irq-affinity and watchirqs to /usr/local/sbin. reboot
8. measure 3.png
9. Try to tweak BIOS settings. Turn off power-saving features, etc.
10. measure 4.png

The output of watchirqs looks like this:

The scripts mentioned above: irqstuff

## Temperature PID control - Part Deux

Update: this version of the component may compile on 10.04LTS without errors/warnings: frequency2temperature.comp (thanks to jepler!)

There's been some interest in my 2-wire temperature PID control from 2010. It uses one parallel port pin for a PWM-heater, and another connected to a 555-timer for temperature measurement. I didn't document the circuits very well, but they should be simple to reproduce for someone with an electronics background.

Here's the HAL setup once again:

The idea is to count the 555 output-frequency with an encoder, compare this to a set-point value from the user, and use a pid component to drive a pwm-generator that drives the heater.

Now it might be nicer to set the temperature in degrees C instead of a frequency. I've hacked together a new component called frequency2temperature that can be inserted after the encoder. This obviously required the thermistor B-parameters as well as the 555-astable circuit component values as input (these are hard-coded constants in frequency2temperature.comp) . Like this:

I didn't have the actual circuits and extruder at hand when coding this. So instead I made a simulated extruder (sim_extruder) component and generated simulated 555-output. Like this:

This also requires a conversion in the reverse direction called temperature2frequency. A stepgen is then used to generate a pulse-train (simulating the 555-output).

• The INI and HAL files for the simulated extruder, based on the default axis_mm config: simextruder
• frequency2temperature component:  frequency2temperature.comp (install with: "comp --install frequency2temperature.comp")
• temperature2frequency component: temperature2frequency.comp (only for simulated setup, not required if you have actual hardware)
• sim_extruder component: sim_extruder.comp (only for simulated setup, not required if you have actual hardware)

"heartyGFX" has made some progress on this. He has a proper circuit diagram for the PWM-heater and 555-astable. His circuits look much nicer than mine!

The diagrams above were drawn with Inkscape in SVG format: temp_pid_control_svg_diagrams

## Why Real-Time?

Why bother with these real-time kernels and APIs at all? Isn't timing on a modern PC good enough? Look at this:

This histogram shows latency-numbers from the same 1ms thread test run compiled without (red) and with (green) real-time enabled. All the green real-time data clusters around zero +/- 20us. Without real-time enabled the event we are expecting to happen every 1 ms might happen almost 1 ms too early, or up to 3 ms late. With real-time the timing is mostly consistent to better than 1% (10 us) with a worst-case jitter of 2% (20 us).

## Latency Histogram

This shows a latency-histogram for a 1 ms thread running on Xenomai on my recently acquired ITX-board. Note how badly the histogram is approximated by a normal distribution (Gaussians look like parabolas with logarithmic y-scale!). See also Michael's recent RPi data and  Kent's Athlon/P4 data.

The usual latency-test numbers people report is the maximum latency, a measure of how far out to the left or right the most distant single data point lies. The histrogram can probably be used to extract many more numbers, but for real-time critical applications like cnc-machine control the maximum latency is probably an OK figure of merit.

The latency numbers are recorded with a simple HAL component:lhisto.comp

The instantaneous latency-number is then put in a FIFO by the real-time component sampler and written to a text-file using halsampler. I'm setting this up with the following HAL commands (put this in a file myfile.halrun and run with "halrun -f myfile.halrun")

net latency shisto.latency sampler.0.pin.0
start

The numbers can now be plotted with matplotlib. I'm using the following script:

 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 import numpy as np import matplotlib.pyplot as plt import matplotlib.mlab as mlab # load data from file x = np.loadtxt('latencysamples.txt' ) x=x/1e3 # convert to microseconds   fig = plt.figure() ax = fig.add_subplot(111) nbins = len(x)/1000 n, bins, patches = ax.hist(x, nbins, facecolor='green', alpha=0.5, log=True) bincenters = 0.5*(bins[1:]+bins[:-1]) # from matlplotlib example code mu = np.mean(x) sigma = np.std(x) area = np.trapz(n,bincenters) # scale normpdf to have the same area as the dataset y = area * mlab.normpdf( bincenters, mu, sigma) l = ax.plot(bincenters, y, 'r--', linewidth=1)# add a 'best fit' line for the normal PDF   ax.set_xlabel('Latency ( $\mathrm{ \mu s }$ ) ') ax.set_ylabel('Counts') ax.set_title('Latency Histogram\n 12.04LTS + 3.2.21-xenomai+') ax.set_ylim(1e-1, 10*max(n)) ax.grid(True) plt.show()

## LinuxCNC on Ubuntu 12.04LTS

Recent developments has made it possible to run LinuxCNC on the latest LTS release of Ubuntu. This is experimental work, so not recommended for controlling a real machine just yet. The main obstacle for moving LinuxCNC from 10.04LTS to a more recent distribution has been the RTAI real-time kernel, which has not been kept up-to-date with development of the normal Linux kernel. Fortunately there are alternatives such as Xenomai or RT_PREEMPT.

Here is a step-by-step description of the install/build process, if you want to experiment with this.

1. Download and install a normal 32-bit 12.04LTS Ubuntu (ubuntu-12.04.1-desktop-i386.iso). Note that the 64-bit version is not supported for the steps that follow further down. I could not get Ubuntu's startup-disk-creator to work, so I used unetbootin to write the ISO-file to a USB-stick.
2. It's possible to compile the xenomai-kernel from scratch, along with the runtime etc., but I used pre-compiled deb-packages by Michael Haberler from here: http://static.mah.priv.at/public/xenomai-debs/
3. Install the xenomai kernel:
sudo dpkg -i linux-image-3.2.21-xenomai+_0.1_i386.deb
4. make sure it will show up as a GRUB-entry when booting:
sudo update-initramfs -c -k 3.2.21-xenomai+
sudo update-grub
5. reboot. uname -r should now show: 3.2.21-xenomai+
6. now install the xenomai runtime:
sudo dpkg -i libxenomai1_2.6.1_i386.deb
sudo dpkg -i libxenomai-dev_2.6.1_i386.deb
sudo dpkg -i xenomai-runtime_2.6.1_i386.deb

This installs the xenomai system on top of which a recently available version of LinuxCNC can be built. There are probably many ways to now obtain the tools/dependencies that are required. I used the following:

1. sudo apt-get install synaptic
sudo apt-get install git
2. Now using synaptic, install the following packages (I found these are required for a minimal linuxcnc build):
build-essential
autoconf
libpth-dev
libglib2.0-dev
libgtk2.0-dev
tcl-dev
tk-dev
bwidget
python-tk
python-dev
libgl1-mesa-dev
libglu1-mesa-dev
libxmu-dev
3. Get Michael's version of LinuxCNC that can be compiled for Xenomai:
git clone git://git.mah.priv.at/emc2-dev emc2-dev
cd emc2-dev
git branch --track rtos origin/rtos-integration-preview1
git checkout rtos
4. Configure and build for Xenomai:
cd src