Would this even work? I have a grill ignitor i want to use to light the lamp... would the grill ignitor be sufficient?? or could i hook a traditional ignitor up to such a setup?

Or perhaps if i cant do it with an HPS lamp... how would i do it with a 175W mercury vapour lamp? I'd imagine that being more simple but maybe not...

HPS may not like the lack of reignition impulse with Incandescent lamp. I'd try 4 F20T12 chokes in parallel instead (for 120V supply, 55V lamp)

Well i mean my whole idea was to avoid any conventional ballasts... if HPS would not like it... how would a 175W MV lamp do?? or perhaps a 6W Fluorescent lamp .. i kinda want to ballast something with incandescent lamps... i find the idea interesting... ballasting a Mercury vapour lamp would be cool...

In a 120V area even the MV would not work. You would need 240V supply and a 200W incandescent lamp, of course 240V rated.

For a F6T5 or F8T5 you may have a better luck: Around 15..18W incandescent may work with manual preheat.
Similar  an F15T8 can work with about 25W incandescent, talking about 120V ratings...

In a 120V area even the MV would not work. You would need 240V supply and a 200W incandescent lamp, of course 240V rated.

For a F6T5 or F8T5 you may have a better luck: Around 15..18W incandescent may work with manual preheat.
Similar  an F15T8 can work with about 25W incandescent, talking about 120V ratings...

Well, nowadays i can boost the voltage as high as i'll ever need, but i still do prefer to work in my native 120V most times, so yeah i will try the fluro thing... maybe i'll try the merc thing some other day... dont have any 240V rated bulbs at the moment

MV lamps have their own ignition, which is starting electrode by via a starting resistor, HPS lamps need to have external starting or "ignitior" to start the lamp.

MV lamps have their own ignition, which is starting electrode by via a starting resistor, HPS lamps need to have external starting or "ignitior" to start the lamp.

I'm aware...


So.. I tried the fluro thing, it works, however with a 15W lamp it seems to be banding rapidly, I upped it to a 40W lamp and it seemed OK, then a 60W lamp and it started sputtering.. so I'm guessing i must remain between the 15W to 40W range?? is 18W like you mentioned earlier the ideal wattage Medved?

So.. I tried the fluro thing, it works, however with a 15W lamp it seems to be banding rapidly, I upped it to a 40W lamp and it seemed OK, then a 60W lamp and it started sputtering.. so I'm guessing i must remain between the 15W to 40W range?? is 18W like you mentioned earlier the ideal wattage Medved?

The exact ballast lamp specification should be based on some calculation guess, but then it should be corrected based on real world measurement. There are effects, which are very hard to cover by the calculations.

What I gave were very quick, but still safe (should not lead to quick blow up due to an overcurrent) guesses - I took the rated arc current and take an incandescent leading at full voltage to 50% higher than the fluorescent rated operating current (usually the minimum specified preheating current), as you may see below, such quick guess could be still quite far off.
But lets calculate it more accurately, yet still neglecting errors like reignition delay:
The exact calculation is a bit complex, I will try to explain it:
You need to reach the current the fluorescent is rated for. So for an F6T5 or F8T5 that is about 0.155A, for the F15T8 it is 0.22A.
In the circuit you have a 120Vrms (so 170V peak) sinewave voltage source (the mains), a square wave voltage source (the fluorescent; 57V for F8T5, 43V for F6T5 and 75V for the F15T5; the actual voltages may differ a bit, but it won't be too far away) and a resistor.
The voltage across the resistor is, what remains from the mains sinewave when the fluorescent voltage is subtracted from the mains.

So you just "cut" a part of the mains sinewave (e.g. 75V for the F15T8; lets use this lamp as an example here) from the zero, so what remains are just the peaks above the 75V and zero in the meantime.
For our example we get nearly half sine pulses with 95V amplitude, but with gaps corresponding to 1/6 of the mains periode, so 2/3 of the time we have the half-sines, 1/3 we have zero.
You then calculate the rms voltage from that: Vincrms = 95V*0.707*sqrt(2/3)=55V (for our example). That is then the rms voltage the incandescent will see. You find somewhere a graph (e.g. here) depicting what the incandescent resistance does when the voltage falls such low (about 75% to normal resistance).
At the same time you calculate the real resistance the ballast have to have, in order to pass the required power to the fluorescent.
Because the fluorescent act as a constant voltage source of 75V, the dissipated power corresponds to a "rectified average" current (Pfluoro=75V*Iavg = 75V * average(abs(i(t))) )
The current (i(t)) are the 95V pulses, so the averaged value is 95*2/Pi()*2/3/R = 15W/75V
And the R is the resistor we want (=200Ohm).
And for a 120V incandescent we know this is 75% of the nominal resistance (at 120V), so the corresponding incandescent will have to have 270Ohm (I'm rounding a bit, but to stay within few percent tolerance) at 120V.
And given power and Ohms law equations we get (120V)^2/270Ohm=53W.
Because the rms current will be higher than the tube is rated for (the pulses have higher crest factor than the sinewave), better to underpower the system a bit, so go for lower wattage, so e.g. 40W incandescent.
You may see, the initial guess based only on the preheating current could be really quite off.
And even this neglects some effects, so the result should be adjusted to keep both preheat, as well as operating currents in check.

Ive tried ballasting an HPS lamp with incandescent (using a plasma ball for ignition) and can confirm it doesnt work, the main arc strikes but cuts out in a second, ive tried many combinations of different incandescent and HPS wattages and had no luck, the 250w HPS with a 500w quartz halogen lamp lit for about 5 seconds then cut out, that was the longest runtime I was able to get out of all the combinations.

Also, you CAN light a series incandescent ballasted mercury lamp on 120v, ive done it many times with many different MV wattages. My mains is typically 122-123v, which is pretty common. Some brands of MV lamps take a few switching cycles to fire up, but most start up right away. And I am not using any extra ignition sources, just the MV lamp with incandescent in series on 120v.
 
The most stable combination I achieved (assuming runs indefinitely, or at least 12 hours before I switched it off) was a standard 175w MV (Sylvania) on a 300w and 150w quartz halogen. so 450w of incandescent for 175w MV

What makes this difficult or impossible is the relatively high re-ignition voltage of the HPS arc tube on each half-cycle of the mains.  Each time the current reverses the arc goes out momentarily, and a higher than normal voltage is needed to re-ignite it.  With a mercury lamp the reignition voltage peak is quite small, however HPS lamps are wall-stabilised discharges that are rapidly cooled by the wall during the current reversal, and they have a higher reignition voltage peak.

When operating an HPS lamp on a pure resistive / tungsten ballast, the reignition voltage peak occurs at a point when the mains voltage sinus is very low - therefore the voltage available from the mains is not sufficient to keep the arc burning and that's why it goes out quickly.  However when operating on an inductive / magnetic ballast, there is a phase shift between the voltage and current waveforms of the lamp.  Very conveniently indeed, this shift is such that the reignition voltage peak occurs at a point when greater voltage is available from the supply.

There are ways however to run an HPS lamp on a tungsten ballast.  In the simplest case the mains voltage should be increased - and for some lamp types it is then possible to keep the arc burning.  However this results in very inefficient lamps, because when the HPS arc voltage is kept at the same level as normal and the mains voltage is raised, the tungsten ballast has to drop the large difference in voltage.  Since the current through both the arc and discharge is basically constant, but the volt drop across the filament must be very large, much more power is therefore dissipated in the filament than the arc.  Since the filament is not efficient, this results in very low efficacy lamps.  This is the main reason that self-ballasted HPS lamps were never commercially developed.

Another technique explored in the 1980s was to make an HPS arc tube with a very high mercury content.  This adjusted the voltage waveform favourably to allow stable operation, and allowed the proportion of power dissipated in the filament to be reduced so as to attain a reasonable lamp efficacy.  Philips developed 100W self-ballasted HPS lamps for use on 220V mains and these had an efficacy of about 30lm/W - considerably better than the 10lm/W of the 100W self-ballasted mercury lamp.  The colour rendering was also better at about 70.  However adding more mercury to an HPS discharge causes a sharp increase in red radiation, and the resulting light colour was not very attractive, even with the tungsten ballast.

Still another technique was to employ a capacitive device inside the lamp to cause a similar phase shift between the voltage and current waveforms as that attained by a magnetic ballast - thereby shifting the reignition peak further towards the peak of the mains sinus, and allowing the ratio of power in the arc vs the filament to be further increased.  However the available electronics of the time suffered short life and reliability problems in the hot environment of the lamp outer jacket.  Such lamps were also not commercialised.

One obvious difficulty of such lamps is that the HPS arc also needs a high voltage pulse to strike the discharge.  The glowbottle type starters used inside some HPS lamps cannot be used with a tungsten ballast, because in that case there is no inductive high voltage kick provided by the purely resistive ballast, as would be the case with a magnetic ballast.  Therefore, self-ballasted HPS lamps should really be made using Penning-Start arc tubes having a neon-argon filling.  However these have the drawback that the thermal conductivity of neon is much greater than the usual xenon gas filling.  This causes still greater cooling of the arc between each half-cycle of the mains, and neon-filled arc tubes therefore have an even greater reignition voltage which exacerbates the above problems.

Many years ago before I understood these effects I also tried to make a self-ballasted HPS lamp.  The best combination I found was to use a 110W Penning-start mercury retrofit arc tube together with incandescent lamps of about 200-300W 220-250V rating.  This combination is right on the edge of being able to operate stably and sometimes if the mains voltage is less than about 230V it will extinguish - its best to run on a variac at slightly higher mains voltages.  I did later make some lamps with modified Sylvania SHX 110W arc tubes to have a moderately increased mercury ratio and slightly lower arc voltage, that would function stably when sealed into a single outer bulb together with a tungsten ballast filament.  The efficacy was in no way high enough to make them interesting for general lighting, but they were partly interesting for some special applications that required the sodium spectrum plus shortwave infrared radiation from the filament.  If I can find one of these I will make a photo to upload here!

For the mercury, quite strong reignition aid comes from the starting probes, where the distance is short, so starts retriggering at way lower voltage. Then once the voltage rises, the gas is already preionized, so it then does not need that much to turn back into a full power, fat arc... It may be even so effective the reignition overshoot completely disapears, so allow the operation even at 120VAC (the voltage is about 50% of the time above the 120V, so still long enough to provide good power through the resistive ballast).
With HPS you do not have such aid, so one more reason why the operation is so problematic.

And regarding the sapacitive ballast: I dont believe it would be any more stable than the resistive. The reason is the impedance at higher harmonics, in other words at the instant the arc disappear. With an inductor the reignition voltage is available immediately after the currenr disappears. Because of so short time, the plasma ionization does not have much time to decay, so it is able to immediately build the fat arc, without the need of the extra acceleration voltage.
With resistive you have to wait from a point, when the current disappears till the point, when the mains voltage exceeds the arc again.
With capacitor the situation is even worse, as although the mains voltage is nearly at its maximum, the voltage across the capacitor is at its maximum too and because it subtracts from the mains, there is nothing left for the arc. So after reaching zero current, you have to wait, till the mains drops by double the arc voltage, so at least the arc voltage becomes available again. And that makes longer gap than just the resistor. The only help is, at the moment of a next half cycle there is a charge pump effect available to restrike the arc, but that would mean really huge crest factor.
Another story is, when there is an inductor in series with the capacitor (CWA,...). There the voltage available for the lamp reignition becomes the capacitor voltage (boosted by the resonance above the ballast OCV), plus due to the inductor it becomes available at the instant of the current zero cross. So such ballast really brings the advantage of both charge pump action of the capacitor, as well as instant reignition (and extra voltage from the swing back after the zero cross), but it as well needs both the capacitor and inductor in the circut, so not usable for a selfballasted lamp either.

Another option would be using some form of an ionization maintaining circuit, which keeps at least some current in the arc stream even when there is no current available from the main power path. But I can not imagine, how to do that without using either a rectifier in the main power path (so supplying the arc by DC, like the GE's Halarc) or by a high frequency (so an electronic added to the circuit). Neither becomes practical in any way (if to use the electronic, it would become way easier to make the main ballast electronic and so save quite some power).

I tried ballasting a MV lamp.... I had to tap the lamp to get it to ignite... and when it did it was only a wispy arc that lasted only a second ... was using a clear regent 175W lamp and used a 150W and 300W lamp as you had suggested, I've measured my mains voltage and it is 120.5W rather than 123V.. perhaps that is the issue?