Manufacturing Information Solutions

[CNCJimmy7]

Discussion on RS232 Common Mode Ground Noise, Grounding, and Isolation.

Grounding Problems.

You must do whatever is required by code relative to grounding. That is a must for safety.

Will this eliminate error or equipment damage problems in a directly connected RS232 link? It may not, and in many cases won't. Relative to data errors we hear people say a good ground is required. Do most of them know what a good ground is? No. So what is a good ground? Do these people have a quantitative value to define a good ground? If so why not specify the value? If they specify a value, then can they show how this will prevent data errors? What is ground? Is it earth ground or something else? How do you measure ground resistance? Not with a DC multimeter. I won't go into the reasons here.

Often it is suggested that an 8 foot ground rod adjacent to each CNC machine and connected with a heavy wire to the machine will provide a good ground. What is a heavy wire? What is material and diameter of ground rod? How deep is ground rod driven? What is resistance to ground? These are not defined.

The NEC ( National Electrical Code ) does not allow equipment to be solely grounded by a ground rod at the machine, and isolation of the machine from the Equipment Grounding Conductor. See comment on page 178, Article 250 of NEC2002 Handbook. The reason is that you can not get a low impedance thru a ground rod. NEC requires that you ground the machine thru the "Equipment Grounding Conductor" ( this is the safety ground wire run from the main breaker box ). Adding a ground rod at the machine may and probably does nothing for you and may actually create problems. Search the internet to see discussions on problems of adding ground rods at machines.

I believe that in most cases that an added ground rod at a CNC machine is next to useless, and probably a problem. The NEC (National Electrical Code) specifies that a maximum of 25 ohms is allowed for a ground rod system to meet code, this most likely requires multiple rods. This implies that a lot of installations may have fairly high ground resistance. If you look at the NEC2002 Handbook you will find comments like the one on page 207 -- "Also assume that the power ground has a resistance of 10 ohms, a very low value in most circumstances ( a single ground rod in average soil has a resistance to ground in the neighborhood of 40 ohms)". A source on the Internet discusses 5 to 10 ohms. Measurements I have made on a clay soil are in the range of 5 to 20 ohms.

If a ground rod alone is not satisfactory to protect a machine, then why are these required at the main breakerbox (service entrance). One reason is under normal conditions to keep earth ground and the electrical system neutral at close to the same potential. Second reason is that when lightning strikes you want the lightning energy to dissipate in the ground where there is a large mass to absorb the energy rather than the energy going into your building and setting it on fire. Suppose the ground resistance is 10 ohms and the lightning current is 1000 amps, then the service entrance and everything in your building tied together by the Equipment Grounding Conductor will rise to 10,000 volts relative to some place down in the earth. If there are no other paths to ground in the building except the one at the service entrance, then no lightning current flows in the building. But if you have one or more machines also tied to their own rods, then a current will flow in the building.

This is an over simplification, but illustrates the general concept. You should not use my comments here to make your decisions, but use this information to help you search and study the subject. The web site reference below with a sexy lightning rod is a good illustration of different points of view. Who is correct I do not know.

What does a given amount of ground resistance mean? Consider a supplemental ground rod at a machine with a ground resistance of 10 ohms back to the nuetral point of your main breaker box. This also means a ground rod array at the breaker box. The ground rod array at the service is the primary ground rod and the one absolutely required by the NEC. In this experiment said supplemental ground rod is connected only to earth ground. Apply 120 v to said supplemental ground rod from a hot line (not neutral or ground) in that breaker box with #12 wire which is fused by a 20 amp breaker. We will use a 50 ft piece of #12 copper wire which has a resistance of 0.08 ohms to connect from the breaker to said supplemental ground rod. The current that flows is I = E/R, or I = 120/10.08 = 11.9 amperes. This will produce a voltage on said supplemental ground rod of E = I * R = 11.9 * 10 = 119 volts relative to neutral and safety ground at the breaker box. This won't ever blow the 20 amp breaker. The failure to blow the breaker and the high voltage produced are the reason that the NEC does not allow this as a safety ground means. Do not run this experiment, but rather use an ac voltage in the range of 20 to 25 volts.

Suppose you could get the ground resistance down to 1 ohm ( I suggest for most applications a difficult task. Would require many rods. May be virtually impossible to do. ). Now use the same #12 wire and do the same experiment. Then I = 120/1.08 = 111 amperes. The voltage on said supplemental ground rod will now be E = 111 * 1.0 = 111 volts. A little less, but not much. In either case serious damage would occur to RS232 components. This current will blow the breaker, but that won't protect the RS232 components because no breaker is fast enough.

Now consider the safety ground wire, NEC's name for it is "Equipment Grounding Conductor", required from the breaker box to the machine by the NEC. This safety ground wire will be of equal size to any power wires leading to the machine. In any practical case these will be #14 or larger wire. Note: wire diameter increases as the wire number decreases. You would have a very small machine if the wire was #14. But 50 feet of #14 copper wire has a resistance of 0.126 ohms. This is 1/79 the resistance of a 10 ohm ground rod, or 1/8 the resistance of a 1 ohm ground rod. If the wire diameter is larger then the ratio is even greater. So typically any safety ground wire is going to have a resistance very much lower than the ground rod resistance. Thus, this supports my statement that the ground rod is virtually useless for noise reduction.

There is another consideration and that is inductance of the ground path. Inductance provides opposition to the flow of transient or ac current. A large area conductor will have a lower inductance (lower impedance), than a small diameter. This favors the earth ground, but the resistance at the interface of the ground rod to the earth, and the bulk resistance of the earth are so high that this may not matter relative to a copper wire.

One Internet reference discusses use of 0000 copper wire ( 0.46" in diameter ) and connected to the I-beam structure of the building. This is the kind of grounding that is needed to be effective in reducing noise. Of course the I-beam structure of the building needs all the beams to be bonded to each other. Welding would be a good means. Further aluminum should not be used anywhere in safety ground circuits.

Briefly some problems with supplemental ground rods:

(1) A conductor ( copper, brass, steel, carbon, etc. ) in an electrolyte produces a voltage. If there is a closed circuit and a voltage is present, then there is a current. If a current is unidirectional (one way), and there is metal in an electrolyte, then metal will move (electro plating). This in time causes disintegation of something, and therefore higher resistances.

(2) If you have a good low impedance "Equipment Grounding Conductor", like maybe 0.01 ohms, then by comparison the 10 ohm ground rod has little effect. However, if lightning strikes with a 1000 amp current, then a large current ( magnitude will depend upon the relative resistances of the various supplemental ground rods and the primary ground array ) may flow to the supplemental ground rod(s). Although the voltage drop within the "Equipment Grounding Conductor"s maybe small (1000 amps thru .01 ohm is 10 volts), the current may thru it's magnetic field induce damaging currents in near by elements. If there are no supplemental ground rods, then no current flows in the "Equipment Grounding Conductor". Everything is relative to the magnitudes of resistance and current.


Alternatively it is a lot easier and better to use whatever is required by code for grounding, this is for safety purposes, and then use an optically isolated, or equivalent, communication link for RS232 which will do a better job of noise problem elimination. For example the Betatronics I232 system. The isolation approach can tolerate large noise voltages (1000 volts), whereas the direct connection of RS232 ports requires ground impedances very much lower than required by NEC to keep peak ground voltages (less than 1 volt) low enough to avoid data errors or component damage.

Some internet references:

Some are funny to read. You will see some generalities without specifics. You will find the discussions to be confusing.

http://www.copper.org/applications/electrical/pq/casestudy/grounding.html
....... heavy ground wires, 0000 copper, and building steel structure. This produces
....... a low impedance safety ground network.
http://www.harvardrepeater.org/news/lightning.html
....... about ham radio lightning protection
http://www.electrical-contractor.net/ESF/Lightning_News/Standard_780_Removal.htm
....... discusses sexy lightning rod with radioactive material
http://www.steel.org/infrastructure/pdfs/Donohoe.PDF
....... some ground resistance values for sand and clay
http://www.mmsonline.com/articles/0302bp4.html
....... no quantitative values -- duplicate of first ref but different journal
http://www.ecmweb.com/mag/electric_cnc_machine_tool/
....... usefull discussion on ground rod at machine. However, this article is not clearly
....... wrtitten. The article should have been longer with very precise definitions. There is
....... inadequate info on what CNC OEMs are specifying for ground requirements. In other words
....... are the OEMs requiring only a ground rod and prohibiting connection to power system
....... safety ground wire. If OEMs are spefifying this, then this violates NEC specs.

A search using .... ground rod resistance earth .... on www.google.com produced the following useful sites

http://www.ees-group.co.uk/downloads/LEC-Green-%20Grounding%20Manual.pdf
....... good on theory, modification of soil with salts ( NaCl, table salt ), and multi-rod spacing
http://www.eng-tips.com/viewthread.cfm?qid=21381&page=8
....... interesting, to an extent surface area should make a difference, one person claims 1 ohm
http://www.hubbellpowersystems.com/powertest/tips_news/pdfs_best/05-2002.pdf
....... some theory, values, errosion of rod, galvanized
http://www.www.mikeholt.com/codeforum/ultimatebb.php?ubb=get_topic;f=18;t=004764
....... several different contributers, a question on worms getting fried
http://www.msha.gov/S&HINFO/TECHRPT/GROUND/GROUNDRE.HTM
....... gov report, tough to get as low as 5 ohms, good discussion on how to
....... measure rod to ground resistance


My conclusion from all this is that:

It is on average tough to get down to 5 ohms resistance to earth ground thru a single ground rod. This in my opinion is nowhere low enough to do much for reduction of ground noise. Thus, do not use or expect any great benefit from supplementary ground rods at CNC machines.

Thus, use heavy copper wire or bus bars for a ground system. I think you need to look at dc resistance more in the range of 0.01 to 0.001 ohms for this ground system. Inductance of the ground system will probably be dominate for high frequency noise ( 10 kH to 100 kH ). Let's make a wild guess on inductance at 30 microhenrys ( inductive reactance = 2 * Pi * f * L ), at 10 kHz this is 3.77 ohms, and at 100 kHz it is 37.7 ohms. On one of our machines I have measured 0.2 amp rms safety ground current under very light load (engraving). Thus you can see that inductance can be very important. The reason for wanting low dc resistance is to keep voltages low when someone shorts a hot wire to the machine and causes hundreds of amps of short circuit ground current.

Or use isolated means, like optoisolators, to communicate RS232 to the CNC and use the minimum safety ground required by code.
===========================================

Why is a DIRECT RS232 to RS232 connection risky?

1. Direct connection means you directly run several copper wires from one machine RS232 connector to another (for example a computer to CNC). For safety reasons all CNCs and computers and other devices have their chassis or enclosure tied to a safety electrical ground. The exception is things that are labeled "double insulated" for example many electric drills. Laptop computer internal common is not usually tied to ac power ground and is isolated by a "wall" transformer or equivalent, or using it in battery mode.

2. Assumptions for the following discussion. Your 120, 240, 480 vac main power source is from a transformer. At your main circuit breaker box the "transformer neutral", "earth grounding system", all "branch circuit neutral" wires, and all "safety ground" wires are connected together.

In any branch circuit from this breaker box the "safety ground", "netural", and "hot" wires are of equal diameter. This is to provide equal resistances in the following example, but also is required by code. If the safety ground wire is smaller the problems get even worse (old installations).

I believe that in general for many different types of earth ground that the electrical resistance between two 8' ground rods, even at short spacing, will be much greater than the resistance of even a #14 safety ground wire. #14 copper wire is 0.0025 ohms per foot or 0.126 ohms for 50 feet. Thus, for the following discussion a grounding rod path between machines would not change the results. This comment does not imply that you should or should not use ground rods at each machine. This must be determined from local and national codes. How you ground equipment is not what this discussion is about. The National Electrical Code allows a maximum ground resistance of about 20 ohms. Even getting ground resistance down to 1 ohm does not have much effect in comparision to 0.126 ohms.

3. Consider one CNC machine and one computer. We will space them 5 feet apart with a 5 foot long RS232 cable. This is to illustrate that short RS232 cable length will not protect against the described fault. Each is fed from the main circuit breaker box with its own branch circuit. For example to the computer a 120 vac circuit using #12 Romex with ground. And to the CNC a separate 3 phase cable with ground wire ( possibly #8 or #6 wires ).

Our voltage reference point will be at the "safety ground" point of the circuit breaker box. The "neutral" and "earth ground" of the circuit breaker box are at this same voltage.

At the CNC the "electronics common" is connected to the "enclosure" of the CNC and thus to the CNC "safety ground" wire from the circuit breaker box. In turn this means the RS232 connector common is connected to the CNC "safety ground" wire. Note any current in the safety ground wire produces a voltage drop I x R. This voltage drop determines the voltage difference between the breaker box ground and the CNC cabinet.

At the computer its electronics common is connected to its incoming safety ground wire and enclosure.

With no fault conditions and no erroneous ground currents the voltage at the CNC common, computer common, and breaker box common will be the same.

Wires of equal diameter, length, and material have equal electrical resistance. The midpoint of two equal resistors in series has a voltage 1/2 of that applied to the series string.

If I have 120 vac at the breaker box applied to the branch circuit to the computer then under no load at the computer I have 120 vac. Assuming a perfect voltage source at the circuit breaker box, then independent of the load current the voltage will remain 120 vac at the breaker box.

If at the computer end I place a "DEAD SHORT" (zero resistance, maybe a screw driver) between "hot" and computer "safety ground" (hot and safety ground wires are of equal resistance), then 1/2 of the 120 vac = 60 vac rms or 85 v peak is applied to the computer case and computer electronics common relative to the circuit breaker box common and also the CNC common. Thus, 85 v peak is applied between the computer common and the CNC common. This will blowout RS232 components and maybe much more at both CNC and computer.

If you do the shorting at the CNC instead with the 240 or 480 lines then the fault voltage will be even higher. Note: HAAS puts 100 ohm resistors in series with each of the pins 2, 3, 4, 5, and 7 at the RS232 25 pin connector. Pin 1 goes to machine chassis. Pins 6, 8, and 20 are connected together. 18 v bidirectional transient limiters are connected between the 100 ohm resistors and pin 1. The transient limiters put a substantial capacitive load on each signal line.

Note that it takes time for the circuit breaker to trip and this time constant is long in relation to the time to destroy semiconductor ICs. Also note that the RS232 cable length really had nothing to do with this analysis. However, the shorter the RS232 cable the higher the current is thru the electronic components and the greater damage.

Even if there is no short circuit to safety ground there may be leakage currents that cause current in the ground wire and thus a voltage. These leakage currents may be large enough to produce data errors.

The problems may be worse if the communicating RS232 equipment is operating from different main transformers and breaker boxes.