Need help for troubleshooting a rustic CNC 650x2000x215

Hi,

My father and me we have built a rustic CNC and it does some good work.
I will attach to this message examples of works and machine pictures. As you will see I had not discovered this forum before building it and we have certainly done some theoretical mistakes.
The second will certainly be better and cheaper.

We use an Arduino-grbl-UGS system. The parameters will be accessible at the end of this post.

Basically, the CNC works but sometimes it stops.
On the screen OFFLINE (black) appear instead of RUN (green) and the CNC stops moving.
The strange thing is that the little orange logo showing connection is still orange when this happens like if we were still connected. And it's impossible to disconnect/reconnect using the logo.
We have to reconnect using the setup wizard procedure and a homing.
It takes a lot of time and as it happens totally randomly (sometimes after 5 seconds sometimes after 12 hours straight without problems we have to constantly look at it whereas we could do other things during the time it works.

We have suspected many different problems :
-connection problems - solved by replacing connectors, using weld instead of bolts.
-electromagnetic problems - solved by replacing the cable with shielded cables for the spindles and the motors. And we manually shied the USB cable waiting for a good one from amazon. We also tried to make a metallic shield in front of the computer with no effect.
-static electricity - solved by connecting pretty much everything to earth we even separate the dust collector from the CNC to see if it changed something.

We certainly also have tried other things I forgot but now we start to be out of ideas and we look to the internet to find the solution.
Sadly I haven't been able to find a post with this type of disconnection.

Did some of you already have this problem?
Could you provide me some idea or link where I could find a solution?

I will be happy to share more information/photos/data if necessary.

Looks like PC-based CNC common problem, not so reliable as NC-based CNC.

Earth wiring is critical when you are using a USB connection. You need a star earth wiring. Make a good earth connection at the power outlet. Connect using a thick wire (separate wire for each connection):

• The housing of the PC
• The housing of the CNC
• The housing of the CNC router
• The housing of the CNC power supply

Connect all equipment to the same power outlet

I have over 19 years of experience in solving these kind of problems. Nevertheless I am unable to run my (bigger) lathe on a USB connection. I use WiFi instead.

...check for high AC Ripple on all DC power supplies. Use a volt meter on the lowest AC scale to check for excess.
happy hunting.

Thank you for your answers.
We are still trying to find what the problem is but a find a new symptom.

The problem occurs only when I use the biggest tools. A 4 flutes Ø40mm running at 4000 rotation/min.
When I use small tools Ø20 2 flutes at 6000 rotation per minute or Ø6 and Ø3 2 flutes at 24 000 rotation per minute I never get any disconnection.
Whereas the power consumption of the spindle measured at the command of the spindle is nearly the same when working 220W.

Could it be possible that a high torque for the same power generates more electrical or electromagnetic problems?

Quote:

---

Originally Posted by machinehop5

...check for high AC Ripple on all DC power supplies. Use a volt meter on the lowest AC scale to check for excess.
happy hunting.

---

I will check that, thank you.

Quote:

---

Originally Posted by hfjbuis

I have over 19 years of experience in solving these kind of problems. Nevertheless I am unable to run my (bigger) lathe on a USB connection. I use WiFi instead.

---

I think this will be the next improvement, a wifi arduino. thank you for the idea it will allow the pc to be at a higher distance from the CNC and maybe remove some problems.

Quote:

---

Originally Posted by Osife

The problem occurs only when I use the biggest tools. A 4 flutes Ø40mm running at 4000 rotation/min.
When I use small tools Ø20 2 flutes at 6000 rotation per minute or Ø6 and Ø3 2 flutes at 24 000 rotation per minute I never get any disconnection.

---

I seems the distortion is influenced by the spindle speed and your setup is more sensitive for low RPM (frequency). The servo motor on my small lathe generates more distortion on higher RPM.

There is a (free) virtual serial to WiFi program that I use some times for testing. It makes a virtual serial port on the PC that can be connected to a WiFi port (telnet). I use ESP32 boards and Arduino Uno and Mega boards from RobotDyn that also have an onboard ESP8266 WiFi chip.
https://www.hw-group.com/software/hw-vsp3-virtual-serial-port

You can also make a WiFi to serial bridge if you connect the Serial port from the controller to a ESP32 or ESP8266 board. There are a lot of examples for this on the internet.

Hi, a little update,

Since 3 days using only small tools (<Ø20mm) it runs without any interruption 12 hours a day.
So I will not make any test or change until I haven't finished my order.
But I have done a little change that seems to have a great effect.

The Chinese Arduino has been replaced by a true Arduino (probably also chinese).
And now I can even run with the hard limits. Whereas before that we frequently had false hard limit detection and need to suppress them in the parameters.
I'm not 100% sure if it's a correlation or a causation but two days without false detections is something never seen.

The original Arduino Uno is recommended for 12 V but rated from 7..20V. When I replaced my original Arduino Uno board for a Chines clone, it give up after a few days. After inspection I found that the voltage regulator was fried.
So there probably is a difference between the original Uno and the Chinese clones.

I also had the problem that limit switches generated false triggers. I solved it by enabling the software debounce. Remove the two slashes at the start of the line

File config.h, line 472 (in my version)
// #define ENABLE_SOFTWARE_DEBOUNCE // Default disabled. Uncomment to enable.

change it to:
#define ENABLE_SOFTWARE_DEBOUNCE // Default disabled. Uncomment to enable.

Thank you for this advice.
You seem to be a big expert on this topic. Our next project is to build a CNC for painting our blades.
Have you already done something like that? Or seen someone doing it?
We have spent so much time solving all problems and doing mistakes with our first CNC that now I'm will take as much advice as I can.

And for info one more day without any disconnection or false alarm. The arduino card seems a lot better at filtering parasites.

Quote:

---

Our next project is to build a CNC for painting our blades. Have you already done something like that?

---

I have no experience making a CNC painter and do not know anyone that has build such a CNC.

It can be quit difficult to simulate a wrist moving in all directions! If the movement can be made using a "standard" 3 axis router, I think it is doable. For me I would first try to simulate the CNC painter by hand moving at a constant speed without interrupting the paint spray.

Great to hear there is progress in solving the current problems!

Quote:

---

Originally Posted by hfjbuis

I have no experience making a CNC painter and do not know anyone that has build such a CNC.

It can be quit difficult to simulate a wrist moving in all directions! If the movement can be made using a "standard" 3 axis router, I think it is doable. For me I would first try to simulate the CNC painter by hand moving at a constant speed without interrupting the paint spray.

Great to hear there is progress in solving the current problems!

---

In fact when we paint (but we are bad painters) we only do parallel lines. The very simple form of our blades allows it.
I think that if the blade can rotate we may even have only 2 axes. 1 For the tool and 1 for the blade.
Some tries are needed but it could be a huge improvement.

Quote:

---

I think that if the blade can rotate we may even have only 2 axes.

---

If you can use 2 paint guns and make the spray wide enough, you only need 1 axis