I once again apologize in advance for my very lengthy post, but I am curious.

I am aware that for an inductive ballast, the OCV of the circuit must be around double-ish of the arc voltage of the lamp for the lamp to stay lit after it strikes. For example, the OCV for American 55V HPS lamps should be at least around 110 volts to keep the arc lit (once struck by an ignitor of course).

I realize that once the lamp extinguishes after each half-cycle, the inductive flyback voltage from the choke is what re-ignites the lamp. This means that the peak OCV does not have to be as high as the ignition voltage of the discharge, because (with the help of an ignitor at the beginning) the discharge will re-strike after each cycle with the flyback voltage from the choke ballast.

I have made the following assumption regarding resistive ballasting:

Obviously, resistive ballasts don’t give flyback pulses after each half-cycle, so (I would assume) the peak OCV of a resistive ballast has to be at least the striking voltage of the discharge for the arc to remain lit. For example, if you had an American 55V HPS lamp, the resistive ballast OCV would have to be several kV to keep the arc alive, instead of just 110V for an inductive ballast. This would also eliminate the need for an ignitor, since the OCV has to be the striking voltage of the lamp. Having a ballasted 5kV output that drops down to 55V during use would be incredibly inefficient, dissipating many times more than the lamp wattage in just plain old resistive heat.

I have the following questions:

1 Am I right?
Is my assumption correct about the necessity for a significantly higher OCV with resistive ballasting?

2 What’s the ratio?
I know inductive ballasts require approximately a 2:1 ratio of OCV to running voltage, but how about resistive ballasts?

3 Capacitive Ballasting?
I would assume that this would apply to capacitive ballasting as well due to the lack of flyback, but is this correct? If it isn’t the same rules as resistive, then how so?

Yep resistive ballast generally follows the same 2:1 OCV to lamp voltage rule, more or less. See for example 160W blended HPM lamp is 80W burner and 80W filament, the same with 250W one. And, you are right there is a problem with ignition and re-ignition. For example, it is hard to start even a short T8 18W tube on resistive ballast, even a single tube on 230V mains, if the circuit does not contain any extra inductance. Also, tungsten-mercury blended lamps are known to be notorious for drop-outs. You can see lamps having a hard time re-ignition and long current pause on resistive ballasts here: https://www.lighting-gallery.net/gallery/thumbnails.php?album=7753

As for capacitive ballast, capacitor in theory can carry-over some charge from previous half-period which means some voltage is added to line peak voltage to simplify re-ignition. But, pure capacitive ballasts are never used at mains frequency in any serious applications. Negative resistance of the discharge combined with a capacitor is causing strong oscillations and high current peaks leading to quick electrode wear.

Pure capacitive ballast is used at high frequency but then a dynamic balance sets in the discharge and re-ignition is no more an issue.

Note some excess of OCV over lamp voltage is necessary not only to facilitate re-ignition, bul also for general working point stabilization, to compensate negative resistance behavior of the discharge.  That way this 2:1 approximate rule is followed not only on 50/60Hz but also on DC and HF.
   

Of course 2:1 rule is not always followed with resistive ballast, especially when energy saving is non-issue. Tatra T3 trams have two fluorescent tubes (said to be 25W, not sure, may be 20W actually) in series on 600V DC circuit.

Yep resistive ballast generally follows the same 2:1 OCV to lamp voltage rule, more or less. See for example 160W blended HPM lamp is 80W burner and 80W filament, the same with 250W one. And, you are right there is a problem with ignition and re-ignition. For example, it is hard to start even a short T8 18W tube on resistive ballast, even a single tube on 230V mains, if the circuit does not contain any extra inductance. Also, tungsten-mercury blended lamps are known to be notorious for drop-outs. You can see lamps having a hard time re-ignition and long current pause on resistive ballasts here: https://www.lighting-gallery.net/gallery/thumbnails.php?album=7753

As for capacitive ballast, capacitor in theory can carry-over some charge from previous half-period which means some voltage is added to line peak voltage to simplify re-ignition. But, pure capacitive ballasts are never used at mains frequency in any serious applications. Negative resistance of the discharge combined with a capacitor is causing strong oscillations and high current peaks leading to quick electrode wear.

Pure capacitive ballast is used at high frequency but then a dynamic balance sets in the discharge and re-ignition is no more an issue.

Note some excess of OCV over lamp voltage is necessary not only to facilitate re-ignition, bul also for general working point stabilization, to compensate negative resistance behavior of the discharge.  That way this 2:1 approximate rule is followed not only on 50/60Hz but also on DC and HF.
 

Wait, so resistive ballasts don’t need to have an OCV that is higher than the lamp’s striking voltage? The requirements are the same for inductive ballasts?

I didn’t even think about the possible leftover charge in the capacitor helping ignition, but that definitely makes sense.

Yep resistive ballast generally follows the same 2:1 OCV to lamp voltage rule, more or less. See for example 160W blended HPM lamp is 80W burner and 80W filament, the same with 250W one. And, you are right there is a problem with ignition and re-ignition. For example, it is hard to start even a short T8 18W tube on resistive ballast, even a single tube on 230V mains, if the circuit does not contain any extra inductance. Also, tungsten-mercury blended lamps are known to be notorious for drop-outs. You can see lamps having a hard time re-ignition and long current pause on resistive ballasts here: https://www.lighting-gallery.net/gallery/thumbnails.php?album=7753

As for capacitive ballast, capacitor in theory can carry-over some charge from previous half-period which means some voltage is added to line peak voltage to simplify re-ignition. But, pure capacitive ballasts are never used at mains frequency in any serious applications. Negative resistance of the discharge combined with a capacitor is causing strong oscillations and high current peaks leading to quick electrode wear.

Pure capacitive ballast is used at high frequency but then a dynamic balance sets in the discharge and re-ignition is no more an issue.

Note some excess of OCV over lamp voltage is necessary not only to facilitate re-ignition, bul also for general working point stabilization, to compensate negative resistance behavior of the discharge.  That way this 2:1 approximate rule is followed not only on 50/60Hz but also on DC and HF.
 

actually in European SBMV lamps the filament often has a higher proportional wattage then the arc tube for improved colour and stability, often SBMV lamps will have arc tubes of lower arc voltage to *try* and get some stability in there, simple maths kinda tells you this a 240V 160W SBMV lamp draws 0.66A from the mains, if the arc tube was 80W that would mean it would have an arc voltage of about 120V! which obviously is not the case, IIRC a 160W SBMV has a 55W arc tube and a 100W Mercury lamp arc tube is only just 30W-35W, by far I think the smallast "regular" Quartz HPMV lamp in that regard

I just thought i'd mention this since I have seen this "a SBMV is 50:50" myth get repeated a lot of times in my time on LG!

(American high wattage SBMV do have a more even balance, but again only in the high wattage lamps)



@Multisubject indeed the re-ignition voltage of a discharge lamp is rarely going to be as high as its cold ignition voltage, and if you know what your doing you can have some fun with that for example I once ballasted a 400W metal halide lamp with a 1500W GLS lamp, quite happily! the main reason you dont see self-ballasted HPS lamps is the re-ignition voltage of those is just that bit higher then say a normal quartz metal halide or mercury lamp, so its quite hard to get a lamp that works stabily *and* efficently, as mentioned above even Self ballast mercury lamps have to have a proportionally lower arc voltage for stability reasons, so in a HPS lamp it would have to be *even* lower and thus more power wasted in the ballasting filament so efficiency goes down, one of the things that is on my list to try is to try ballast one of my low voltage Japanese 40W SON or American SON lamps by way of resistive ballast, I think those will have a low enough arc voltage not to cycle but we will see!


https://www.lighting-gallery.net/gallery/displayimage.php?pos=-243157

Well Dez, you need to be a little bit careful in calculations, as mixed-light mercurys certainly do not contain any reactive electric components BUT still do not have power factor of 1! So simple Joule law calculation of P=U*I does not work. See current waveform here, captured by myself from a real lamp. https://www.lighting-gallery.net/gallery/displayimage.php?pos=-225954


If you look up datasheets from Osram HWL's you'll see 240W 160W lamp is rated at 0.8A and 225-235V regular HWL is rated at 1.23A which is really a bit above 1.15A of regular 125W burner, so 250W lamp may really have somewhat shortened burner compared to regular '125 HQL.

I'll try to measure some real RMS current of my ML lamps in the evening.

your not wrong that the power factor wont *actually* be unity, however Thorn give a running current of 0.65A for the 160W Self ballast mercury lamp with an arc tube voltage between 90V and 110V for the 240V version, and note the difference in arc tube specifications for a 160W SBMV arc tube (an overall length of 50mm with an 18mm electrode gap and 11.5mg-12.5mg of mercury) vs that of an 80W mercury lamp arc tube (56mm overall length, 21 mm electrode gap, 15-18mg mercury) details of which can be seen in this most interesting document that I have spent far too many hours studying over in minute detail

http://www.lamptech.co.uk/Documents/Specs/Thorn/Thorn%20-%20Specification%20Handbook%20-%20Group%209%20File%201%20-%20MB.pdf

it is rather interesting that Osram give an *actual* wattage for their HWL of 170W (and 175W in the Russian version?), because, the British Self ballasted *woods glass* mercury lamp has always been rated at 175W rather then 160W and I remember James commenting many years ago why that was, that this was because, although they where internally 160W lamps, the amount heat trapped inside the lamp by the woods glass envelope raised the wattage up

Interesting document really!

Yes, it shows that at least Thorn mixed lamps are designed with with mains to burner voltage (power) ratio of 2.44:1 (220-230V 160W) and 2.4:1 (220-230V 250W). Does it still qualify as "approximately 2:1", you decide

There is still some uncertainty, as this book mentions two burner voltage ranges for 220 and 240V ML lamps, but only single mechanical specification for each burner wattage.

An interesting fact is that 160W and 250W ML burners are notably different mechanically to their plain mercury counterparts from 80W and 125W lamps. While 500W ML in fact re-uses the exact mercury 250W burner.  I believe the explanation is the difference in fill pressure, giving the burners significantly different cooling modes. While 80 and 125W mercury lamps are filled to 20-40 Torr of nitrogen, mixed light lamps are specified as >500 torr. This pressure difference is generally absent in 250W mercury vs 500W ML lamps.

In fact, two counteracting factors work in mixed lamps when compared to plain mercury one. Filament sure gives some significant extra heat to the burner, radiant and convection, but at the same time, higher fill pressure significantly increases heat transfer from the burner to outer bulb.

I am skipping 100W ML as a weird type, rarely seen in the wild.

As for real measured parameters, I tried ML160 from GE and HWL260 from Osram.

GE ML160 230-240V

line voltage at that time: 225.6V
current: 0.708A
power 156.7W
power factor 0.98
CCT 3196
CRI 67
made in Turkey, which is unusual

Osram HWL250 235V


line voltage at that time: 224V
current: 1.081A (power monitor) 1.076A (RMS multimeter)
power:232.5W
power factor: 0.95
CRI: 55
CCT: 3512

date code f668 made in P.R.C.

Interesting, 160W GE lamp has significantly higher filament light to the burner ratio compared to 250W Osram. I have no more ML lamps to compare another. Both lamps notably run below their powers specs, likely because my line voltage at test time was below design voltage for both lamps. 

Another interesting fact is color temperature measured for 160W lamp just turned on is 2881K which certainly means the filament is overdriven compared to usual 2700K of plain filament lamp, though a little light from the burner may offset CCT at that time a bit, too.