Why are European HPS lamps 100v when US HPS lamps are 55v? Do they have a different chemical mix in them? If so, do the european 100v HPS lamps have a slightly different color? In some videos of 100v HPS lamps, they look slightly more golden, but that could just be the camera.

First it have to do with ballast loss on 120 volt realm
  It less ballast lost on 120 volt mains and half the maiins
Voltage to keep arc stable across the ballast to the lamp

 S56 HPS lamp will run on MH 150 watt gear.
    While 35 to 150 watt lamps are design around 52 to 55 arc volts.  s56 is 100 arc volt

I doubt the material type is any different. The difference are :

Arc length and maybe pressure - to make the intended arc voltage

Thermal design (arc tube thickness, electrode mass etc) to reach the correct working pressure, arc temperature, excess amalgam temperature, electrode temperature and such

If there is indeed difference in pressure, then it will affect the color



The reason for the difference is efficiency :

More significant : Simple choke ballast is more efficient than HX or CWA (unless they are WAY bigger size)

Less significant : Higher arc voltage makes more efficient lamp



On 120V, it makes sense to make a bit less efficient lamp, but which can suffice with a way more efficient ballast, so get better overall efficiency. On 240V, the voltage is high enough to allow efficient lamp and efficient ballast

What does not make sense to me is, why in the US the 100V lamp is not widely used for streetlighting, where 240V, 277V and 480V are readily available ?

For 230V mains markets the main design goal is to operate with all lamps with a most efficient ballast. That means keeping the ballast just an inductor and design the lamps with arc voltage as high as possible, so the current as low as possible for the given power level.
Because the arc on an inductive ballast needs twice as high OCV to be stable, it means the arc voltage should not exceed half of the mains. That condition should be then maintained over the complete lamp life.
So because MV's have the arc voltage stable over life, they are designed at that maximum arc voltage - so in the 100..140V range (fatter arcs tend to be more stable by itself, so allow going a bit higher with the voltage, so the 100V is for the 50W, 140V for the 400W lamps).
Now because HPS tend to increase the arc voltage over their life, the initial (rated) one should be kept lower, so it keeps sufficient marhin for the aging before reaching the 100..120V limit, hence the 70..90V arc (again, thinner arc for lower wattage needs more OCV margin, so lower arc voltage, fatter of the higher power models suffice with less margin, so allows higher arc voltage design)

In the 120V areas the MV spec's were just taken from the European originals (the GB Osira,...). One of the reasons is the 120V won't start the simple probe start lamp even when designed for 60V arc, so either more complex lamps (preheat,...) or ballasts become necessary anyway. So the choice was to keep the compatibility and use step up transformers.

The HPS came later and there it became clear, for most wattages the step up transformer function would be required anyway, so the lamps were just optimized for the maximum efficacy for the given wattage. That has lead to the wide range of HPC arc voltages and their incompatibility with anything else.
The separate cases were the power levels of 150W and below. As the HPS needs the extra ignitor anyway, with that became possible to start the lamps even on 120V OCV, so the question of ballast losses came back to the main topic. And it popped up, for the 150W the overall efficacy was about the same for both optimized lamp on less efficient ballast vs 55V lamp designed to suffice with just the series reactor with the high efficiency (and low cost) of that simple reactor ballast. For lower wattages the series reactor just won on the system efficacy and cost.

Now why to not use the higher voltages? They are used, but not everywhere. And mainly for the lower wattages there was quite significant market to suffice with just the ordinary "socket" voltage, so the 120V in the US, so the 55V arc lamps.
That requirement is the same for both sides of the pond, but in Europe that is 220..240V, hence the 70V arc lamps.

One BIG disadvantage of the simple reactor ballast even if cap powerf factor compensated--the STARTING current of the system could be up to 3 times the run current.So feed cables,breakers,and fuses have to be sized for this.CWA,Ferroresonant transformer ballasted fixtures have a LOW start current-less than the run current.So cables,breakers can be at smaller ratings.

One BIG disadvantage of the simple reactor ballast even if cap powerf factor compensated--the STARTING current of the system could be up to 3 times the run current.So feed cables,breakers,and fuses have to be sized for this.CWA,Ferroresonant transformer ballasted fixtures have a LOW start current-less than the run current.So cables,breakers can be at smaller ratings.

The inrush (few mains periods) current is about the same, the startup current is then low, if correct power factor correction is used.

It is higher-esp on large systems-have even measured it.The current drops as the lamps warm up.It is in the NEC codes.That is a big disadvantage of reactors.That is why some lighting systems no longer use them.And CWA,isolated transformer systems offer a more stable lighting systems.Reactors were more sensitive to voltage variations that are tolerated by the other ballast types.Of course-now its electronic ballasts and LEDS.The reactors are low cost-but this can be offset by the larger cabling cost.They work OK on voltages of like 240V on up.

The lamp current tend to be larger, but (at least how the 230V ballasts tend to behave) it is just an inductive reactance, so the input current is either the same or lower than with the lamp at full power.
But that depend on the exact ballast design - when it is designed to strongly saturate, the current may be higher.

In fact the HPF HX ballasts are exactly the same principle as series reactors, just maybe with larger margin towards saturation (it does not cost that much extra, compare to a plain series reactor)...

But what may be problematic (with 230V designs it is the mode with the largest mains current) is the high uncompensated capacitive current when the lamp extinguishes. It may boost the voltage at the end of a long line ("Ferranti" effect), cause some resonant problems with other ballasts in series (missing Neutral failure), so that is, why many 3-phase installations here use lamps connected in an "Y", but power factor correction capacitors in "Delta". That way no single "broken wire" fault may cause any dangerous resonance effects or so... And even double fault (broken two phase wires) means there is no more than 230V across an eventual series LC with two lamps in the circuit.


The advantage of CWA is, the lamp current remains rather constant, better matching the power factor compensation (the gap in the main magnetic circuit), so indeed the input current would be low.

With the resonance-saturation stabilized ballast the input current really remains low even when the lamp current is high at startup (improving lamp thermal stability), but this ballast type has the highest losses of all magnetic ballasts I know (assume equivalent OCV and size; the losses could always be reduced by sacrificing cost, weight and size)...

277 and 480 are not generally available on residential streets. Most residential street lights are owned by the local utility companies, and just tapped into the 120/240 lines serving the houses.
They could use 240 just as easily as 120, for streets, but to use it in a house requires special wiring, and someone isn't going to want a special circuit to put a yardblaster on his garage, so you'd have to have two series of lamps anyway.
Freeway lighting does tend to use 240 or even 480 due to the distances from the transformer.
I do find it interesting that the Europeans use tubular lamps for lower wattages, and electronic transformers for HPS.