Buzzing change over

Should also add that coil voltage ok and as best as could be managed without complete dismantling, inners cleaned and blown out.

Are you brave enough to thump the side of the enclosure with the side of a clenched fist?

That was the first thing out of the tool bag Andy! 
The CO is relatively large but the current drawn through it is is largely due to LED lighting load which, incidentally, was installed as replacement for fluorescents and coincides with the buzzing. Total load is only around 30A per phase with intermittent increase to 50 odd when restaurant is going. Facility is a large indoor bowling arena. 
A tear down would be a difficult operation as big pressure on the facility every day of the week. No signs of thermal problems.

Is there a sensing coil in series with the load ? If the waveform is the classic LED diode input one of  drawing almost no current at all except at the crests of the waves that can lead to a much sharper 'raspy' buzz (think heavy metal bass guitar waveform ) from inductors asked to pass that waveform  when compared to the gentle hippy trance hummmm of a sine wave. 
Sometimes seen on MCBs and RCDs also.
If it gets gentler when a few 3kW heaters are added in parallel that is all it is. If so it is not sinister, but it may indicate a poor waveform and a flag for possible power factor billing anomalies.. Longer term some low pass filtering may be worth considering but probably not a big issue.
Mike.

Love it Mike, a “gentle hippy trance hummmm”! Somehow I don’t quite see you with the long hair, sunglasses and odd looking smoking implements, at least not during your scout master days….there again…

I still help lead a group these days you know, I'm not totally past it yet (grinning not cross).
Mind you the last winter camp I did a few weeks ago  (116 youngsters hiking about in the new forest for a weekend doing various activities at different bases) my back did let me know in the morning getting up and out of my tent that it was not quite as springy as in years past.Perhaps I need a taller hike tent, or at least one with a popup section as a  'stand up here while getting dressed' annexe.
And I gave up smoking in the 1980s, though I certainly did have noticeably longer hair, and it was not grey and thin on top as it is now.... 
But you know the happy transformer sound I mean anyway.
Mike.

It sems to have at least one contactor, it may be worn pole faces, in industry when too noisy, we used to dismantle and file the mating surfaces as flat as possible till replaced; if left, there will be an increase in heat at the coil with eventual burnout.

Jaymack  

Ah yes that is a really good point - perhaps I am too complacent  in my previous advice - if there is a contactor not quite 'sitting down' so there is an air gap in the magnetic path, it  will draw a lot more current as the inductance of the solenoid, rather than the wire resistance, limits the current  once the armature has moved into the end stops. The coil has a much higher inductance (impedance to AC ) once it has a complete loop of iron core, compared to one with a mm or so missing..
With an AC supply this is a nice way of giving a huge pull-in current to do things with a satisfying bang, and yet have a much lower holding current one when the contacts have shut. I've never had to do the contact filing thing but I have seen a stray grain of sand in the gap cause a coil to overheat, which I guess is the same problem.
M.

The iron cores used to have what what known as a shading ring fitted to them to cancel out circulating currents - has this shaken loose and fallen off? If so, it can result in buzzy coils as described above.

Weren't the copper shading rings not designed to create a "remanent" circulating current at each armature face, to maintain the magnetic attraction during zero cycles?

Jaymack