Advanced Search

For the super nerdy, you can read all about it in the ISO standards for Geometric Accuracy by looking up CSN ISO 230-1(Linear Accuracy) and CSN 230-7(Rotary Accuracy). Or simply save the money it costs and wait for me to tell you about the ISO's and how to actually test them. There are other accuracy tests but they are expensive to test and as long as the above is satisfied, you can forego them and still have a solid .025mm total accuracy tolerance as long as your backlash is small. Usually within .075mm/per axis is ideal. Then, a compensation can be be properly made in the control to account for that .075mm/per axis.

A possible scenario to locate the mechanics of your 2 motors would be to do the following.

Home the 2 motors with zero offsets. Then measure your square determine the difference. Say 3mm out of square. Choose an appropriate motor to adjust. Place an indicator on that chosen motors frame to read the travel. Release the motor coupling and rotate the ball screw amount required to square the 2 sides +-.150mm or .007~.008" then lock the coupling. Rehome the tandem axis. Then measure and adjust the home offset parameters to complete the setup until your square is within .005~.025mm based on your acceptable tolerance. Then, to ensure that the reversal timing is aligned, once the homes are "squared" move the axis until at least 1 home switch activates. Then move the inactive switch until it makes at the same place. This will eliminate racking at reversal which will over time cause wear and issues that are preventable.

I only use compression type motor couplings. Keyed couplings cannot be used for the above alignment for obvious reasons.

The same would apply for rack and pinion designs as far as the connection type using compression couplings and the procedure listed above.

Remember that .001" out of square on the gantry axis will be exponential as the perpendicular axis(if X is your tandem axis then Y would be the P axis) axis moves from it's reference position. Meaning that over a 10" travel the tool tip will not be .001" out of position or CL of expected location, it will the rise of the triangle out. Here possibly .0025 or so. At 20", even more. Ideally you want the tandem square to be zero in perpendicularity to the P axis. This will also help with concentricity when interpolation of holes is performed.

I may do a post on using ISO standards of machine tool squareness and how the ISO describes the measurement of the parallelism, perpendicularity, and total square of the full cartesian coordinate system for both 3+2 and true 5x machines soon. I think it would be helpful for those unlike me who have 30+ years of nerding on CNC at the highest levels to understand how to get the most out your equipment and save themselves from many many headaches. Give me a week or 2 to put the documents and graphics together. After following the course, you will be able to have a machine that is aligned to ISO standards. It may sound trivial to do so, but doing so saves businesses annually well over $10B in maintenance and production loss costs. As you already know, the parts for CNC are not cheap. And time is money.

I was asked recently why I would just freely give away trade secrets. The answer is clear. The world has slowly stop making things and economic growth is showing that net effect. I am hoping that in sharing my experience, a flower will open up and a new manufacturing business will contribute to the equation driving economic growth in market economies. This benefits every single living being. Any notion otherwise, would be ill advised. Lol. I am a charitable guy, but I am also a strong capitalist.

Kinda nerdy, but I hope that helps.

Hi, I am really struggling to install LinuxCNC. I am currently running Ubuntu, but this will change to Debian next week. For installing to Ubuntu am I able to download LinuxCNC directly to my laptop, or will I need to download the image to a USB stick and boot it all from there (and is it the same on Debian)? I have been trying to install it directly to my laptop, but am having no success. Any help/advice would be greatly appreciated.

you can also try to run RIO on you Spartan6 board, it's not sooo hard to port it.

EDIT: the supported Arty-a7-35t uses also the vivado toolchain,

in the best case only :

 cp ./riocore/boards/Arty-a7-35t.json ./riocore/boards/Spartan6.json

and edit the default pins, type, clock and package values

then you can create a new config with    "boardcfg": "Spartan6",

 

Yeah the olimex headers aren't fun at all and not a lot on .1 pitch headers.

I haven't played with mine for bit as I went down the path of mesa firmware on a Spartan6 dev board, but since the Rpi5 doesn't have a linuxcnc SPI driver yet I'm a bit bummed.

Ok not knowning your exact board, I found a Debian image for the beagleplay, www.beagleboard.org/boards/beagleplay copied the rootfs part to a directory chrooted in and was able to install Linuxcnc via apt.

Depending on the design the motor end effect torque can be substantial and will break stuff. A 10mm pitch ball screw with a 2.9nm(750W) servo can generate near 1500nm or force. That is a lot. Most major industrial CNC machines have at least 5000nm of end effect force. Therefore, using tandem axis control must be set up very accurately.

There is a caveat to my prior statements in regard to home to index in the drive with a tandem drive. That is the index pulse alignment. This is where things can get harry quick. If the index pulses are 180 degrees out from each other, then you home to the index for each motor you will be out of square by the pitch of your driven axis. A 10mm pitch ball screw would be out by at least 5mm when final home is complete. If your machine is real stiff this could break a bolt, warp a ball screw, etc.

Even an ABS encoder has a one per rotation pulse aka index pulse. On an incremental encoder this is the Z pulse. On an ABS serial encoder this is the turn data marker. They do the same thing but have different meanings. Regardless, it would be wise to know where the 2 motors are in relation to their respective index pulse locations. Every situation is different in how to find that pulse sync location. But, it is well worth the time and effort. Usually, a few degrees from exact sync is acceptable.

Hope that helps...

The new boards is not for the Olimex iCE40HX8K-EVB,
but RIO already works with it.

the only thing i hate about the olimex board are the connectors, they can't be plugged on other boards.
but i once thought about building a small adapter board to use the olimex with 4-5 external bob's.


EDIT: eu.robotshop.com/de/products/devantech-icewerx-ice40-hx8k-fpga

Oh BTW the link isn't working

Is this the correct one ?
eu.robotshop.com/de/products/devantech-icewerx-ice40-hx8k-fpga

That is pretty awesome.

Damn so I cant use the Olimex iCE40HX8K-EVB ?

So I will have to get a new toy ?
Will the others get jealous ?

sys/io.h has the functions for ioport access on the x86 arch.

Can you use the debian arm64 packages ?

From what I've read the best way to do this is not to cross compile but to compile in an emulated environment via qemu.

I've done for the Rpi4 by copying theRPi4 file system to directory then chrooting into it.

Here's some info about the procedure
wiki.debian.org/QemuUserEmulation

If you have a link to a filesystem image i can have a look for you.

Since I use more and more the Servo42/57 drivers and play around with robots and many axes,
I thought I'd make a new board so that I don't have so much cable clutter
and everything is relatively compact.

 

ok, 8 axles is not really necessary, but my Parol6 has 6 axles + a servo gripper,
if I put a stepper gripper on it, I'm already at 7, so the 8th doesn't bother me any more

* 8 x opto inputs (5/24V can be installed in 2 groups)
* 8 x outputs (5/24V can be installed in 2 groups) / can also be used as PWM/RC servo output
* 8 x stepper axes
* 3 x encoder inputs with index pulse (3.3Volt level)
* 2 x I2C ports (3.3Volt level)
* 1 x modbus-rtu
* 1 x WLED output + one ws2812b onboard
* 1 x PMOD socket
* Current and voltage monitoring via ADC's
* Size: 100x100mm
* FPGA: eu.robotshop.com/de/products/devantech-icewerx-ice40-hx8k-fpga (plugged on)
* Ethernet: wt32-eth01 (plugged on)

I'm not quite finished yet, maybe you or I can think of something else that might be missing.

I chose the FPGA because it is fast and relatively cheap to get in Germany.
The Ethernet board is also cheap and extremely stable.

The board can also be used for milling machines and lathes,
Initially, I wanted to put a standard NEMA17 driver socket on it,
but that would have been too big.
Maybe I'll make an add-on board with space for the drivers.

For my next trick: O-1 tool steel (soft).



This is the first proper "sample part" I am going to video to market the new machine. Just waiting on my 4mm end mills to come today or tomorrow I hope.

I expect only one real issue with the fine surfacing on the steel and that is the lack of jerk control and stepper tuning. It will probably leave marks at transitions is my expectation.

All banned.

Customs will seize them if they find them and you can genuinely go to jail. There's no messing around here.

Also, for clarity these are machines I will sell to customers. So no used items for rando chinese inverters, even if it was legal.

Hey, everyone!
I am installing LinuxCNC on an ARM64 board (the core module is TI AM6254), and the board uses the premeet-rt kernel. I know I need to use cross-compilation, but I don't know how to proceed. My operation sequence is as follows:

./configure --prefix=~/Desktop/linuxcnc/linuxcnc-build/ --build=i686-pc-linux --host=arm-linux  --enable-static --with-realtime=uspace --enable-non-distributable=yes --with-libmodbus --with-libusb-1.0 ARCH=arm64 CROSS_COMPILE=aarch64-linux-gnu
make 
make installThe terminal prompts me that there is no. in rtapi/uspace_rtapi_app.cc
How can I should do? Looking for ur reply.