Would placing a halogen capsule electrically parallel to the arc tube of a MH or MV lamp make a good quartz restrike? Why has this never tried in a commercial lamp?


And where external quartz restrikes were used, why didn't they just wire them parellel to the socket instead of using using relaying which I've seen fail btw?   

Would placing a halogen capsule electrically parallel to the arc tube of a MH or MV lamp make a good quartz restrike? Why has this never tried in a commercial lamp?


And where external quartz restrikes were used, why didn't they just wire them parellel to the socket instead of using using relaying which I've seen fail btw?   

Because it wont work properly. The resistance of the incandescent would interfere with the arc, making it way less stable (with most technologies to the point of really not working at all).
Plus the filament would consume power about 30% of its rating without generating any significant light output.

Electricity will follow the path of least resistance... Which in this case would be the halogen lamp, so the arc tube will never receive enough power to fire up.

Almost, electricity takes all paths. However, a big enough filament will pull down the voltage to the point it may not strike. So I can understand now why it was never tried.

The point here is, the discharge resistance changes with passing current, while after current zero cross it is almost and before first ignition it is infinite. That are the points, when the discharge has to be (re-)ignited, what means the lamp needs some minimum voltage for that and that has to be delivered by the ballast (either the initial HV pulse or the reignition spike).
The ignitor has rather limited current capability (in mA), while it is required to generate few kV. That means MOhms are becoming a significant load for it.
Now the ballast is  designed with exactly the assumption the lamp resistance is very high, so with many designs the thing counts on the internal ballast capacitance (few 100's pF till nF) to form an LC that lets the voltage to swing above the mains at the moment of reignition, so create an extra voltage spike for that. But because the capacitance is so low, it needs really high impedance loading to maintain the Q so the effect is still there. The assumption is, if the ionisation is low so the effect is helpful, the discharge impedance is high (in MOhms), so is the Q.  And the discharge has lower impedance (the ionisation from the previous half cycle havent decayed so much), there is enough ionization so the effect is not needed.

That means virtually any incandescent filament becomes virtually a short circuit compare to the impedances at play during these impedance sensitive moments.

Ok- but what about a probe start lamp? MV lamp?

Ok- but what about a probe start lamp? MV lamp?

The zero cross restrike is still rather sensitive part, because the high pressure makes all the voltages high.
With MV the probe aids that, but that would allow really only very low power backup: Fraction of the MV rating, so low it would be of no use.
The MHs have their probe deactivated when hot, so rely to the ballast OCV and swing back only.

Plus there is still the problem of the backup lamp eating substancial power all the time.

The current sensing relay is nothing complex, while it allows unlimited backup power in case the input voltage matches the incandescent rating and mainly is not messing up the ballast vs discharge interface.

And by the way the easiest way for full backup is to wire just two (or even three, if you are expecting double cutouts like with generator backup; with pulse start lamps) discharge lamps in parallel on a common ballast output. The discharge characteristics ensure obly one of them will light, preventing the others from igniting, so keeping them cold. Once one of them extinguishes, another cold one takes over, as long as there is at least one of them cold enough for ignition.
However this could be problematic with probe start lamps, as the operating voltage of one lamp may cause some glow disharge on the cold backup one, so some sputtering of the electrodes there.
With pulse start the cold ones need far higher voltage for ignition from the igniter action, but because the voltage is limited by the burning lamp, the ignitor wont generate anything.

Makes sense now that I think about. A 230 volt lamp across a 100 volt arc tube will still dissipate just under 1/4 the wattage. So either you will have a very dim quartz re-strike, or one that raises the wattage of the lamp noticeably.


Assuming you have a halogen tube ballast inside the envelope but the starting electrode and its main electrode are still "across the line", would this help in significantly in re-lighting the arc after the zero crossing with the re-strike filament pulling the voltage down across both main electrodes? 

Makes sense now that I think about. A 230 volt lamp across a 100 volt arc tube will still dissipate just under 1/4 the wattage. So either you will have a very dim quartz re-strike, or one that raises the wattage of the lamp noticeably.

Yes, it will dissipate 1/4 of its rated power, that is a lot. A $1 NC current control relay is way cheaper even for few 10'sW ratings...

Assuming you have a halogen tube ballast inside the envelope but the starting electrode and its main electrode are still "across the line", would this help in significantly in re-lighting the arc after the zero crossing with the re-strike filament pulling the voltage down across both main electrodes? 

The auxiliary helps to restrike the arc, but the OCV is way stronger factor.
In fact for restrike you need voltage high enough for the avalanche charge multipication to be supercritical (so the charge multiplication is faster than decay via recombination).
In a homogenous field environment this will be exactly the arc voltage. But because the field between electrodes is not homogenous, the voltage becomes slightly  dependent on the currents, lower currents usually yielding higher voltages.
Plus even if supercritical, you need time to reach the desired conductivity to reach the nominal current. This time is then longer, if you have less criticality margin.
The auxiliary there just set the base ionization, just a starting point, which still needs heavy multiplication, so some time.
So yer, witrh the auxiliary it will be a bit better, but if the voltage becomes too low, it won't be enough...