Advanced Search

Hi everyone,
I’m building EgeCobot, a custom 6-axis collaborative robot arm powered by LinuxCNC and EtherCAT.
I’m writing my own custom kinematics component for LinuxCNC to handle trajectory planning and joint control for this 6-DOF setup.
Here is the first video introducing the project and mechanical layoutlanned Tech Stack:

* 6-DOF articulated cobot design
* Custom LinuxCNC kinematics module
* EtherCAT fieldbus for real-time joint control

I’d love to hear your feedback and suggestions on EtherCAT setup and HAL integration!
Best regards,
Emre

snowgoer540 wrote:

I wonder what the cause is. Part of me thinks people have moved on to laser since it's become more affordable.

Yes, exactly, here the government payed 15K Euro to china for everyone working with metal that applied for the grant to buy a laser and they added the rest, so basically killed off my business completely.
Not the only business, so i was not worried ... till the PC market went bananas with prices skyrocketing, so no more PC builds for now, and in the near future by the looks of it.
3D printing business was not that active, but occasionally it would bring nice revenue from people with bitcoin miners that needed cooling adapters...
I am ranting again ... sorry.

tommylight wrote:

Sorry Snowwy, i should have also mentioned that on all my other machines (probably 20? Plasma's) positive limit is at 0, and the example above was for very old PlasmaC, not QtPlasmaC.
On "my Red" machine (my refers to the one i use often, not the rest of them), i still use old PlasmaC and new QtPlasmaC, as my clients have both types in use so i can not afford to forget how to use them, technical support being the main reason, and testing new versions being the other reason, as you can notice if you watch my videos on youtube, different PC's and laptops but always the same red machine.
I am also sorry for not helping more, but as i mentioned several times by now on the forum, work has plummeted badly, been probably over 5 months without turning the machines on.

No apology is necessary!

Sorry the work is plummeting. I do think there are WAY fewer people making plasma tables as compared to when this part of the project was in its prime. I wonder what the cause is. Part of me thinks people have moved on to laser since it's become more affordable.

Sorry Snowwy, i should have also mentioned that on all my other machines (probably 20? Plasma's) positive limit is at 0, and the example above was for very old PlasmaC, not QtPlasmaC.
On "my Red" machine (my refers to the one i use often, not the rest of them), i still use old PlasmaC and new QtPlasmaC, as my clients have both types in use so i can not afford to forget how to use them, technical support being the main reason, and testing new versions being the other reason, as you can notice if you watch my videos on youtube, different PC's and laptops but always the same red machine.
I am also sorry for not helping more, but as i mentioned several times by now on the forum, work has plummeted badly, been probably over 5 months without turning the machines on.

A couple of photos would help as I don't understand what you have done
Normally you would use 0-10 volt from the VFD itself.
Your VFD will have several terminals on it that need their function set, I have X1,X2,X3,X4 that can be programmed for specific functions
A 5v relay either on your BOB  or an external one will trigger forward and reverse. (I used X1,X2)
The VFD should also have a relay internally. You can program this to be "Frequency reached" or similar and send that back to LinuxCNC for your at speed signal.
The only catch I  had was I had to invert the pwm pin otherwise it ran the wrong way (eg S0 gave 10 volts). Not sure how Mesa does it. I recently published some videos on my YouTube @MrRodW channel about wiring in a BOB and how to set up the spindle.

I’ve recently spent a lot of time visualizing the workpiece in Vismach and optimizing toolpaths.
If anyone uses Vismach and would like to test this out, please let me know.

 

it's not that, zeroing out an axis in G54 offset (the default) wont do anything. the config isn't right. having a positive travel distance tells it there is 7.5 inches above wherever "home" is. the limit switch is at the top of the travel looking at the pictures. The first move is to go up to z max minus 0.196". which instead of being -.196" it's trying to go to 7.304.

This is all in G53 to be clear.

You can try something PhillC mentioned quite a while back when i was using non standard Z axis limits on my machine:
-jog Z axis down till the tip of the torch is just under the material/plate
-touch off Z axis to 0 there
do some tests, see if it works properly.
It did work for me, and most probably it is still like that.

The step/dir outputs should continue toggling indefinitely.

At this point I would ask for a RMA# from Mesa and return both the 7I76U and the FPGA card for
analysis/repair/replacement.

I dont know much about your setup, but I am guessing that Positive Travel Distance should be 0. It's the distance above the origin that the z axis can travel. So if you homed it in the middle of travel your setup above would allow you to jog 7.5 inches up and 7.5 inches down. I'm guessing your home switch is at the top, and therefore when it finds home it should call that 0 and there is no travel above it. Make sense?

From the docs:

POSITIVE TRAVEL DISTANCE
This is the distance the Axis travels from the Origin to the positive travel distance or the total travel minus the negative travel distance. You would set this to zero if the origin is positioned at the positive limit. The will always be zero or a positive number.

setp hm2_7i92.0.gpio.019.is_output 1

Since the both the pump and fans are meant to be used in PC's they are rated for 12 volts so I added a separate Meanwell 12V/2A power supply for them.

The power supply is constantly powered but I switch the output side using a solid state relay so that the cooling only runs when the spindle is on.

However the pump proved to be a bit noisy and way too powerful give the small diameter hose I'm running. So added a 7.5Ω resistor in series with the pump to bring the voltage down to 8V which made it much quieter and brought the flow down to a more suitable level.



With that done I tackled the cable chain along with all the wires and hoses.

 

 

With that in order I turned my attention to the stainless cover I had laser cut and CNC bent a while back. I had to manually make a cut out for the connector of the spindle servo which I had forgotten to take into account. But I think it came out okay anyway.

 

Fitted it with a rubber strip to cover the sharp edges.

 

Here is how the machine looks with the cover installed.

 

 

 

Thanks for the feedback dbtayl. Don't worry about clutter, forum threads like this are all about dialog and sharing experiences.

For some reason the final bits always take longer than expected to get sorted. I've managed to get the spindle liquid cooling installed and to somewhat button up the build for now.

The first challenge with the spindle cooling was the location of the inlet which is located at the back of the spindle flange.



I didn't want any hoses down there as the hot chips when milling steel could easily damage them. So I did a little workaround using an angled fitting and a bit of pipe originally meant for a grease based central lubrication system.



 

 

Next up I tackled the the cheap PC water cooling radiator I had purchased of Amazon. I 3D printed some wall mounts for it along with two fan grills for the Noctua fans.

 

Here it is mounted on the wall together with the pump/reservoir unit which I also purchased of Amazon. The yellow color is from the special type of antifreeze I use together with deionized water.

 

Hello,

I think I’ve found the reason for the reduced speed at F1000.
It’s the "MAX_FEED_OVERRIDE" setting in the .ini file.
When MAX_FEED_OVERRIDE is set to a value greater than 2.0, the speed is the same as it is for F5000.
Is this behavior intentional?


Gruß
Rüdiger

 

And just because i measured a lot, are all outputs Pull-Up's?
Do i need an Pulldown-resistor to make the card work? 

Strange behavior with the attachments, the files somehow have a date added..   The Bitfile is attached her..