Vfd cable, shielding, star earth

Thread Starter

robinscottlawrie

Joined Feb 15, 2018
35
Hi all, i have a question re. Grounding/shielding a diy Cnc machine.

I hope this is the appropriate sub-forum

Ive been reading all I can get my hands on, particularly regarding the cable from the vfd to the spindle. Apparently its a very noisy link with lots of mhz level noise, which often causes problems for the stepper motors and limit switches and various other sensitive equipment in the vicinity. Both the earth connection and shield of the cable need treating in specific ways. (shield connected at both ends, braid straps used instead of cable etc..)

My issue is as follows:

Im following guidance on a star-earth setup, and both the vfd cable ground and shield (360 degree connection) are extended to the star earth block using 6mm tinned copper braid straps inside some thin rubber tubing.

My concern is what happens next.. From the star earth point i connect to mains using a standard 16 amp kettle lead, so from braided straps i go to normal 16awg cable.. My limited understanding sees this as a "choke point" for the high frequency stuff, if that makes sense.

Given the stepper cable shields and all other bits go to the same star earth, i imagine i need "the path of least resistance" to be to mains ground..? Is my setup problematic?

I could potentially run another braided strap from the star earth, or even just from vfd cable shield (seperate it entirely) a few metres to the house boiler, and strap it to a water or gas pipe (copper, and the gas pipe goes straight outside and underground)

Would that be better? Should i leave mains earth connected too?

As you can see all this shielding stuff goes over my head, and there is *tons* of contradictory info online, from people who maybe dont know more than me!

Any advice from the pro's most appreciated.

Thanks!
 

Ya’akov

Joined Jan 27, 2019
9,069
It would be far better to connect your earth to wherever the electrical system is bonded to earth rather than to the CPC in a receptacle.
 

MaxHeadRoom

Joined Jul 18, 2013
28,617
I spent the last 40yrs in CNC retrofitting and design, and with regards to VFD's, in the case of HF versions, i.e. 24krpm then it is advisable to use the specialized shield cable which includes and earth gnd conductor.
For the standard style 3ph motor I generally used flexible metalic conduit with TEW/MTW conductors twisted along the length with the GND conductor ran alongside in the conduit.
I would set up a star point GND plate where the service ground was also connected.
There is also a publication by Siemens on the practice of equi-potential bonding of the machine in order to prevent ground loops.
I believe I posted a copy here before, a search may find it.
 

Thread Starter

robinscottlawrie

Joined Feb 15, 2018
35
thanks very much for the replies! Yaakov.. im in a rented apartment on the first floor of a big building.. i would imagine the closest i could get would be to run a cable down to the breaker panel in the room below, but i guess i would just be mirroring the cable that im already connected to. this is why i suggested maybe grounding to the heating system since those fat copper pipes which run down to the garden probably get close to "true earth" if such a thing existed..

MaxHeadRoom, the VFD is one of those chinese Huanyang jobbies, bread and butter of the diy cnc world. spindle is a 1.5kw 24000 rpm watercooled one.. i contacted IGUS to get advice on the vfd cable, i purchased one of their "chainflex" shielded ones based on their suggestions. its not one of those fancy ones with 3 earth connections, but i read that doesnt really matter on a smaller spindle?

i have a star ground plate as described already, everything is routed there.. my issue is what to connect the star to.. as i see it i have a few options, either the heating pipes, and/or the mains earth connection in the local plug, or maybe another cable routed to the breaker panel (its a sub breaker panel, theres another in the garden where the main electric meter is, but its 20 metres from the house and 1 floor down. where the actual ground connection for the house is made ive no idea, its a 150 year old stone villa and its not ours.

given these constraints, what would be the best solution..?

ive tried reading up on equi-potential bonding, but what i read i didnt entirely understand, apart from the concept of connecting all parts of the machine together with decent cable/ braid and routing that to the star ground. ill search for the siemens PDF, possibly ive not seen it.
 

Ya’akov

Joined Jan 27, 2019
9,069
You need a low impedance path to earth. The quality of the CPC connection, what it is shared with, and the route it takes are all unknown. You could try it, but if you get poor results it is the first place to look.
 
The general rules are:

RFI (Radio Frequency Interference) is reduced by shielding
Shield at one end only - typically the source of the signal
EMI (Electromagnetic Interference) is reduced by twisting.
So, you see a lot of twisted pair shielded cable

EMI occures when cables are run next to one another. Power lines in parallel with signal lines without separation.
They can cross, but not run parallel.

So, when I was running signals from a 100mV bridge (Hydrogen sensor) in a building, they were kept separated from 24 VAC and 24 VDC signals.

Obviously, it doesn't work well with CATV.

There is TRIAX Cable which has two shields. The outer shield is earth. The inner shield (Guard) is at the same potential as the center conductor. The center conductor can carry low currents e.g. Picoamps. The guard prevents leakage. Cables such as these generate tiny currents when they move an when they are flexed, so the cables I used had graphite to lubercate the shields. It's called the triboelectri effect and piezoelctric effect. Most people don;t have to deal with that, I did.

Star grounds
Keep dirty, clean, and high current grounds separate until one point where they connect to earth.

EQUIPOTENTIAL Bonding is yet another method.

Ferrite beads can be used to stop high frequency signals.
Ceramic capcacitors close to the source can help RFI as well.

There is shielded power cable. I used it once for heaters in a RF Plasma system. I designed using it. Didn't want 13/56 Mhz near thermocouples.
 

MaxHeadRoom

Joined Jul 18, 2013
28,617
MaxHeadRoom, the VFD is one of those chinese Huanyang jobbies, bread and butter of the diy cnc world. spindle is a 1.5kw 24000 rpm watercooled one.. i contacted IGUS to get advice on the vfd cable, i purchased one of their "chainflex" shelded ones based on their suggestions. its not one of those fancy ones with 3 earth connections, but i read that doesnt really matter on a smaller spindle?
Those spindles have a 4 pin connector 3ph and GND, but invariably the ground pin is not actually connected internally, it requires lifting the 4 pin socket and supplying the earth GND to the frame of the motor.
Also you may already be aware that those spindles cannot operate below 6krpm, without burn-out, therefore the min RPM should be set by parameter.
Multi-core cables that emit EMI have to be twisted in order for self cancellation.
The star GND point is a termination point of some kind where all the GND referenced conductors meet and connect at this point.
 

Thread Starter

robinscottlawrie

Joined Feb 15, 2018
35
This is something i dont quite get, it seems equipotential bonding and star grounding are mutually exclusive, if the parts of the machine are bonded together, then to the star point, and the spindle is also connected to the star point via ground cable, its no longer a star but a loop? Unless (silly idea) spindle is grounded and isolated from machine, and all machine parts have seperate cables to star point (dozens of connections!)

I get the idea but i have problems visualising a practical version.

Regarding low impedance to earth, this was my original question.. Obviously for the higher freq. Stuff, a normal ground lead wont be ideal.. Ive got an emi filter on the vfd input, and a large ferrite ring on the 3 phase output to the spindle.. None of this tested yet, im still recieving packages and building/designing as i go, and trying to follow "best practise" whete possible within my constraints.

So a nice simple question: is there any risk/ illegality/ unadvisability to having both the cpc connection and a braid to the pipework both connected to the star point? Two better than one?
 

Thread Starter

robinscottlawrie

Joined Feb 15, 2018
35
Those spindles have a 4 pin connector 3ph and GND, but invariably the ground pin is not actually connected internally, it requires lifting the 4 pin socket and supplying the earth GND to the frame of the motor.
Also you may already be aware that those spindles cannot operate below 6krpm, without burn-out, therefore the min RPM should be set by parameter.
Multi-core cables that emit EMI have to be twisted in order for self cancellation.
The star GND point is a termination point of some kind where all the GND referenced conductors meet and connect at this point.
Already added the missing gnd link in the spindle, cable is twisted internally, and min rpm is already set to 6k rpm :)
 

MaxHeadRoom

Joined Jul 18, 2013
28,617
The principle of both is that as well as a star point GND, all metalic parts of the machine be bonded (equi-potentially), this in order to prevent any ground loops (i.e. any different points of GND being at a different potential.
 

Ya’akov

Joined Jan 27, 2019
9,069
So a nice simple question: is there any risk/ illegality/ unadvisability to having both the cpc connection and a braid to the pipework both connected to the star point? Two better than one?
Two is not better. One good path is what you want. The CPC is an unknown, if you have a low impedance part with your own connection, the only thing the CPC could do is add noise from some other source your conductor would have to eliminate.
 

Thread Starter

robinscottlawrie

Joined Feb 15, 2018
35
so, going off on a slight tangent... equipotential bonding "in practise" (assuming i have even slightly understood the concept)

i have the star ground plate located in the control box, circa 2m from machine.

so: does equipotential bonding require a separate heavy gauge cable/strap from each axis/major component of the machine, all the way to the star-point? this would mean quite a few more heavy cables in my drag chains, and trailing to the control panel.

Or, is it sufficient to simply link all the axes together with straps/cables through the drag chains, then connect the machine to the star ground with ONE cable? (doesnt seem very "equi-potential" in this case, but a lot more practical ... i envision a machine that is more cabling than cnc, with every bolt and extrusion connected independently to the star point...)
 

Thread Starter

robinscottlawrie

Joined Feb 15, 2018
35
Local connections to the machine’s chassis Properly made makes everything on the machine look like a single device which is connected to the star ground.
And (sorry for potentially stupid questions) given the spindle is earthed, and bolted to the machine with a huge clamp, if i bond all the axes of the machine togehter well, and maybe run a strap from the z axis directly to the grounding bolt on the spindle, is that earth connection sufficient, or should i run a seperate earth cable from the frame ( causing potential ground loop?)
 

Ya’akov

Joined Jan 27, 2019
9,069
And (sorry for potentially stupid questions) given the spindle is earthed, and bolted to the machine with a huge clamp, if i bond all the axes of the machine togehter well, and maybe run a strap from the z axis directly to the grounding bolt on the spindle, is that earth connection sufficient, or should i run a seperate earth cable from the frame ( causing potential ground loop?)
I am losing track... from the frame to what?
 

Thread Starter

robinscottlawrie

Joined Feb 15, 2018
35
I am losing track... from the frame to what?
Sorry! So, i need to bond the various parts if the frame togehter, and then run a cable from there to the star ground.. There is already a grounding cable inside the vfd cable, which runs from the spindle (electrically connected to frame), to the star ground point (via the vfd ground connection)

My question was, do use the existing spindle ground cable to ground the machine, or add a second, dedicated cable from frame to star ground (meaning, if we consider the spindle ground, there would be two paths to the star ground point from the frame.
 

Ya’akov

Joined Jan 27, 2019
9,069
Sorry! So, i need to bond the various parts if the frame togehter, and then run a cable from there to the star ground.. There is already a grounding cable inside the vfd cable, which runs from the spindle (electrically connected to frame), to the star ground point (via the vfd ground connection)

My question was, do use the existing spindle ground cable to ground the machine, or add a second, dedicated cable from frame to star ground (meaning, if we consider the spindle ground, there would be two paths to the star ground point from the frame.
One ground to run away from the machine. The existing ground may be OK. You will need to test, of course.
 
Take a car. it's a huge mass of metal. Light bulbs used to return to the body of the car. So, that's equipotentiall to a point.

Now you stick an engine in there. Now there are bushings potentially insulating the engine (your CNC). Our STAR point is the negative battery terminal. From there, we run the rest of the car. We do have specific ground points on the vehicle that get used, so its not just any bolt. The heater blower motor ground is going to be chosen. A guy putting in a sound system is going to wan to use the negative battery terminal. The front door may be hinged, so it;s not the greatest connection to ground, We don't power stuff through the greasy door hinges. The window motor will have a ground conductor exiting the door. The switches will have their own.

There is the alternator as an enemy. See https://m.littelfuse.com/~/media/el...n_automotive_environment_application_note.pdf

Cars had an ignition system with long wires to each plug and that arc source, the distributer. Electrical things got quieter with no arcing. Quieter yet when you have a coil per cylinder. When you have an anlog signal, the reference is 1/2 the 5V power supply, It's NOT 2.5V. The 5V power supply might be 5.1V or 4.9V. The reference is 1/2 those numbers. You can't have a 5.00000V output and you can;t have a 0.00000V output. The MAP sensor isn;t referenced to the engine. It's referenced to the negative battery terminal. The electronics in the car don't want to see the effects of 200A flowing to the negative lead of the starter cable.

So you have RFI, EMI and you have low level DC signals and you have your neighbors. You don't want a whine from the spindle to get into the power line and bother the neighbor downstairs.
 
Top