[Toinvd]

Your stepper motors and spindle have a DC supply, However, they are switched on and of in a high rate, and that's causing trouble.
Switching on and of these high currents are the cause for a lot of interference which you better prevent going into the controller.
My advise is USE shielded cables to all motors which have switched power.

[JohnMiller]

Hi all,
following your discussion about cables and shielding I feel that shielding allone is only one side of the coin. Along a bad design any shielding may get the problems worse. I confess to be a newbie in mattes of CNC but an experienced engineer. Therefore I want to get you intersted in a proper desing of system ground and connections. Most builders are not aware of this hidden source of peace or war. Whle tearing down a Chinese TB6600 board searching for enhancements I observed enormous lag of knowledge at PCB side and connecting leads.
The notions being presented here are of general use in any electric system. But as the shared contribution is specifically dedicated to CNC builders those facts are explained for CNC mills.

The usual notion is that the GND signal is the base of any circuit. The name itself suggests the "GND" to be stable and rock solid. This expectatin is not true. Circuits like motor drivers dealing with power and/or steep edges on power lines suffer on permanent "earth quakes" on power lines - including GND as well. It is only a question of time when your dishes leave the shelf and get broken.
It is a matter of fact that most of cheap motor drivers suffer on bad design but on bad GND system as well.
Another surprize for you will be that such circuits own different connected ground systems and they shall be sorted smartly in order to to prevent "broken dishes".

Please regard in advance that not only + lines suffer on voltage drop! If you measure at distance a drop from e.g. 24V down to 20V the usual notion is that those 4V get lost at + line. That is not true. Given the + and GND lines are of same length and same gauge there are 2V voltage drop at + line and 2V voltage increase at GND line!!!!
If you transmit a 5V signal from the far end by an extra line to the PSU end - it will measure 7V compared to the GND level there. If you ponder on these facts you easily can imagine ANY strange effect.
But if you imagine at example above that the GND line is very, very fat and very very short, the GND is reliable and rock solid. If we measure 4V voltage drop we are sure it originates from + line only (or 99% :-) Exactliy this condition is what we want and urgently need to have in our circuits.
The notions above relate to DC in order to get first understanding. I do not go in detail but at any form af changing voltage and current a wealth of dynamic behaviour surfaces being very difficult to detect the faster the worse.
The effects mentioned above apply to any form of GND system and the only question is what safe margin do we have before failures. The art of design is to get different currents tamed and guide them to races where they can not disturbe other "parties".

Attachment 261600

In order to get a rough idea of waht you will learn, here we can imagine the ground systems as a tree along trunk and branches.

1.
The PSU is the trunk and there shall be a central GND point at GND terminal of PSU. It is good practice to distribute power including GND from there in star topology of similar wire length (+ anf GND twisted) to the drivers or other power consumers. Do not connect your motor drivers in chain topology! Therefore such GND points are called "star point" as well.

2.
Any connection to protective ground from mains shall be done at GND of PSU. The metal housing shall have exactly ONE connection to protective GND when delivered. Connect a fat wire from protective GND to DC GND. This will be the preferred and only connection where our system will get a reference to Earth. Else the system will float anywhere und react unpredictively.

3.
If you want to connect your mill to protective GND do it by an extra fat wire to preferrably the same GND point at PSU GND.
Of course if the mill has some mains driven appliances added you may comply to regulations and safety and connect the mill to protectve GND - but please use same mains phase (same socket or socket near by).

4.
What about motor housings? Imagine your complete electric circuit to be inside a protective glove. The entry point is the Main DC PSU. Conversely the motor housings are outside and will be naturally bolted (= connected) to the mill mechanics while the windings are isolated and belong to the inside of the glove. Same at switches or other add-ons. Do measurements in order to guarantee thet there is no direct connection. Any connection will act as a closed loop, acting as an antenna. Unpredictable currents will occure and lift or drop your GND system like menitoned above. Please note: Any autonomous loop at GND systems is prohibited. All branches shall be connected with exactly ONE path to the trunk (PSU GND point)

Now let's detect more inside the glove.

5.
The power arrives at the driver. Every driver sits inside a finger of the protecting glove. Directly at the terminals we (hopefully) find a low ESR electrolitic capacitor - preferrably acompanied by a 100nF ceramic capacitor. Both act as thort time battery and block non DC current from PSU and power lines. Severe and short timed AC activities happen now locally only.

We recall that we entered with each driver board into one finger of the glove and now we travel along such a finger.

6.
+ and GND shall be traced shortest way and with fat traces (much surface) to the power GND of the driver chip. If you have the choice to have GND or + short - decide for GND! in order to pertain GND stability.
BTW: The capacitor SHALL be connected on the direct poer / GND path. Many poor designs connect it at side stubs. In this case you are left with problems and the capacitor lives a comfortavble and undisturbed life. Every fraction of an inch counts?

7.
Where power lines arrive at the driver chip it is good practice to add a 100nF carammic capacitor again. It supports the task of the capacitor mentioned before. But we shall take in account that capacitors will act whithin a certain distance only. It depends on travel speed of electical signals (about 0.8 x speed of light) and frequencies to be blocked. I do not want to go deep in to the matter but please note that we need here an additional capacitor (at TB6600 one for each channel of course). If SMD is difficult to add, then add a wired cap across the correspondig IC pins.

8.
Continuing along the finger of the glove we exit now the motor output pins and arrive at the motor windings - well isloated from the outside world.
Paird wires shall be twisted (e.g. two wires for a motor winding or 3 wires of a switch). If the bulider decides for a cable shield: The shield MUST be unconneced at far end and shall be connected at driver IC GND - as close to the chip as possible. It shall not be common with next signal ground available - bad practice!.

Here is the end of one finger.
But this is not the end of the story. Differently from a real glove we anticipate that fingers (maybe aliens ones :-) ) can be forked (like brances of tree) but still well isolated from the outside world.

9.
Now we have to regard the control circuitry of the driver. We enter here one side branch of the finger. Drivers usually have different GNDs for: POWER / logic and analog low voltage (3.3V or 5V). (TB6600 -> SGND). This pin is connected iternally to power GND but is in fact the start of the side branch. It is essential to omit ANY current from power circuit to flow here. Such a design does not care of teh PGND moves or not. The control logic will do its job.

10.
Notions above are true IF we have NO electrical connection to e.g. the controlling PC. Therefore galvanic separation is essential (usually opto couplers).
If this condition is not true we earn several implications at same time:
A: Given we use an Arduino with grbl we have 3 GND loops between all drivers. Any driver current is allowed to path signal grounds as it likes.
B: Additional ground loop via protective GND (PSU and PC with USB) via drvier GND and signal GND . Of course a lap top PC might have an islolated PSU but still condition A: is true.


Note:

* Please regard that some notions presented above are simplified in order to explain basic principles of GND design. No measure has advantages only. The designer needs to know exactly what he does and why, what impacts are. Therefore different approaches are possible and can be successful.
* If you want to check your build for proper GND design please identfy every GND wire or plane at far end (finger tip of the glove). If you travel back: only ONE (exactly ONE) path to PSU GND ist accepted. Any multipath choice may be detrimental and explain your problems.
* The plurality of requirements presented is not easy to fulfill. Any use of a two layer PCB is a drawback and requirements need to be negotiated thoroughly. Some thoroughly crafted free hand designs on bread boards might be superiour compared to poor 2 layer PCBs. There is no real reason to not design with 4 layer PCBs. Usually the inner layers are dedicated to GND and + while the outer layers are dedicated to components and their signals. A GND plane is a very stable GND and these inner layers often replace the necessity of 100nF at every chip (please still use them!).
Additonally the GND connection of any leg will enable the GND current to take the direct and shortest way to the respective GND (e.g. SGND). Nevertheless keep the power area and signal area seprated and connect the planes at power pins of the driver.