Tom  I could not tell you exactly what freq the arc start will be on DCEN but definitely on the majority of older machines it will be the same as the constant high freq but will last only say .5-1sec.    The problem here in lies with the amount of current flowing thru a given conductor with this high frequency superimposed over it.....more current equals more wattage therefore more transmission power available thru the open air to be captured by other conductors like circut boards in CNC equipment.   Most CNC equipment built after 1980 or so (please dont make me pull out NEC code) is rated for a certain amount of shielding capability on the controls.  Modern stuff is well shielded against interference.  The problem specifically occurs when there is a reasonable wattage (think power) radio freq transmitter is nearby.  The power (wattage) is high enough to induce emf (electro motive force) in anything conductive nearby.  This in turn can induct  current in said conductors and cause failures.  Most electronic circuts built on the principals of the transistor only require .7 volts or less to function.  A high wattage radio transmitter can easily induce more voltage than .5 volts in a nearby conductor.  Therefore high freq ....it can shut it off or do any function inadvertently that can occur by the push of button on its controls any other device which functions on a solid state circuitry built on the premiss's of less then a volt unless its properly shielded by modern technology.  

A great example of this is the cheap cd boombox attached to my toolbox...it works fine most of the time but when I am playing a cd sometimes when welding aluminum (only aluminum) it will shut completely off an arc start....it laks quality shielding   Well imagine having two or more hours into a program on a cnc and the welder strikes an arc and its GONE....That is a reality Tom Cooper....on newer machines I do not see it happening.......but anything pre 1985  you betcha.....and even on modern machines if the shop electrical supply is improperly setup...a good backfeed could cause the same result.  Look in the CNC cabinets do you see coaxial wire and aluminum strips on the cabinets.

Now as far as making shielding to prevent such I say look at the common coax cable built from space technology......a conductor inside a shield which is also part of the signal.    The reason that shield works is the woven strands...geometry....opposite directions but still connected....ok you see the idea...it has to do with magnetic poles and how radiation can can conduct itself.    If the shielding is aligned 90 degrees against (each other' itself":molecularly" then it will cancel itself out in propagation (wattage) from the source.  It works the same essentially with an induced signal ......the cross conductors eat away the influence of the signal power wise.  So in other words a ferrous conductor woven in the traditional basket weave method makes a powerful shield against any high freq or other radio signal now matter what the wattage for all intents and purposes.

So in summery I can say without question the fears of your machinists are grounded in hard scientific fact.....check to make sure their controls are safe before venturing into their space.

I don't know what the frequency is, I guess it could be figured by a radio engineer, but this type of device generates a frequency over a pretty wide range due to it's nature. These devices fall under FCC jurisdiction.
As to the AC component, the voltage is rather high, but the amperage is very low, so the AC has nothing to do with the melting of the material, it just ionizes the shielding gas so it will conduct electricity and start/maintain the arc.
At the auto frame plant there was always a concern about welding causing unwanted power spikes in equipment, and nothing was to be welded while a crane was hooked to it. This was the rule and had nothing to do with HF. I don't know if the concern was warranted, but that was the rule, PERIOD.