I have been using this light fixture for quite some time with a 150W HPS lamp. However, the ballast is probably a 100W ballast. What happens to an HPS lamp after running it like this for a while?

My guess is that the life is decreased, because as the sodium escapes the arc tube, the voltage of the lamp rises to a point where the fixture cannot sustain an arc. I imagine that the voltage is lower with smaller lamps, so the voltage would more quickly go outside of the acceptable range.

That link is not working for me...

I fixed it 

It is OK to run HPS lamp at 2/3 power, and factory made adjustable ballasts do exactly this (and even slightly more). I have one DALI controlled to play) Probably no lifetime penalty at this level, may be even some life gain with arctube running cooler, less corrosion, less sodium migration.  But it is usually agreed to run the lamp at full power for some 30 min to filly stabilize and then dim.

Of course you will get poor CRI with more monochromatic light.

Exactly. I don't know the exact way to dim them, but i also have a DALI compatible Philips driver. It claims up to 40% reduction in energy cost, so you can probably under-drive a SON by 1/3rd without negative effects, though i am unsure if you can do that both with switching drivers as well as with conventional inductive ballasts.

The only question is, what type of 150W HPS we are talking about. If the low voltage S55 type, all above is perfectly valid.
But if the lamp is the S56, the rated working current of it is way smaller than the rated current of the S54 100W HPS. So when cold, the higher current of the S54 ballast may overload, so potentially damage the electrodes of the S56 lamp. But the higher current (if it ignites - with just 120V OCV igniting an S56 is not guaranteed at all) will stay only few seconds when the arctube is still cold and has low voltage drop.
Once the tube starts warming up even a bit, the steeply rising arc voltage will either extinguish the arc in the lamp completely or at least cause it to reduce the current very significantly, as the 120V OCV of the S54 reactor ballast can not support any significant current in a normally 90V S56 arc (if the arc stays burning, the reducing current will stabilize at a temperature leading to no more than 70V, above that the current will start to get reduced very steeply). So in that condition the electrodes will be safe again, only the ignitor may be firing continuously to support (keep reigniting) the arc.

But with an S55 lamp on an S54 ballast there is another problem: Because the S56 lamp will be underpowered, it will have lower than normal voltage drop, which will lead to currents higher than the S54 ballast is designed for, so the ballast may overheat...

Lcubed3 definitely talks about low voltage lamps.

"But with an S55 lamp on an S54 ballast there is another problem: Because the S56 lamp will be underpowered, it will have lower than normal voltage drop, which will lead to currents higher than the S54 ballast is designed for, so the ballast may overheat..."

An old wives' tale, never supported by any practical observation!


That way, if you run a mercury or sodium lamp that leaked or have a broken outer bulb, ballast will overheat and die too. BUT I never have seen or even heard this ever happened in practice.

 

This is the exact lamp I am using. It says S55 in the specifications. I have not noticed it cycling.

I remember the ballast running at about 150° at 70° ambient temperature.

The CRI doesn't seem to be so bad, in fact it's a little higher than the 35W HPS wall-pack I have nearby.

The fixture had various hints for 150W lamps, for instance the ignitor said "150W" on it.