Please be careful in knowing that a F40T12 ballast, a European 80W mercury vapor ballast, a North American 75W H43 mercury vapor ballast, or a North American 100W H38 mercury vapor ballast WILL QUICKLY OVERHEAT AND CATCH FIRE IN A MATTER OF MINUTES WITH THE ND20 low pressure sodium lamp due to the lamp’s extremely low arc voltage drop. THESE BALLAST SUGGESTIONS ARE EXTREMELY DANGEROUS AND MUST BE AVOIDED AT ALL COSTS.

Thanks for the info!

These are all interesting ideas, but do any of these result in significant ballast losses or could be used in security lighting?

Hello,

My first try would be a leak transformer for an 90W SOX lamps.

Regarding the ballast of an 80W fluorescent, I suspect that RRK means the soviet 8ft fluorescent tubes and balasts?

I certainly think that these ballast are overloaded at such a low lamp voltage drop, however id need the exact ballast data to tell anything about how quick in would catastrofically fail. However I personally and this is my opinion you shold be fine with runnig that for short times / testing purposes.

Soviet/British 80W 102V 865mA 5ft tubes. Ballast sure runs hot but OK for a reasonable time. About the same as running in a circuit with a stuck starter. One may set up a little fan to keep the temperature low enough BTW.

These are all interesting ideas, but do any of these result in significant ballast losses or could be used in security lighting?

No easy way with off-the shelf ballasts. But I doubt these Chinese lamps (and all spectral lamps also) have enough lifetime to be considered for general lighting. More of scientific toys with some hundreds hours (at best) of life.

Not to mention they have abysmal efficacy due to the large straight tube and no IR reflecting layer, at around 20-25 lm/w for this 20w lamp, might as well go halogen.

These Chinese lamps are rather short lived as there glass chemistry of the discharge tube is not good leading to the glass beeig attacked by the sodium vapour. I Think I was told that you could be happy if you get 100hrs out of one.

Regarding the ballast, i disagree with the stance of @WorldwideHIDCollectorUSA. At least for the leak transformer. Se, every Transformer is able to limit it schort circuit current, due to its stray field inductance and resistance of the winding. However The resistance of the winding is way smaller then Impedance caused.
As a site note in The short circuit current in transmission systems is limited only by the impedance coming from the stray field as the winding resistance is overs of magnitude smaller then the impedance caused by the stray field.
This is very similar with leak transformers. They have a very big impedance between Primary and secondary due to the much more present stray field. This leads to leak transformers having very high open circuit voltages. For example the open circuit voltage of a 90W SOX leak transformer is about 450-500V while the lamp voltage is only 112V or so.
The current of a SOX lamp is about 950mA.
this gives the Leak transformer an impedance Z=(R^2+jwL^2)^(0.5)=430Ω.
If you no decide to run the ballast with this lamp (ie to get down to U_L≈15V) you get an current of I_T=(U_oc - U_L)/Z=1.13A

So the current is about 20% more. 
P_cu=I_T^2*R --> Power losses in the windings are about 40% Higher. This will lead to more heat development. However we also Know the losses of such leak transformers: Osram specifies the losses in a leak transformer with 23W. These however can again be divide in two parts: Magnetising losses and current losses. Magentising losses are constant and makes something of around 10% of the rated power of such a small transformer. So only around 13W remain as current losses when operated at nominal. When running this spectral lamp these losses would rise to around 18W, so 5W more. Regarding the fact that the maximum tempreture of the windings of electrical machines is around 130°C and that my leak transformer only gets handwork when running an SOX 90W I can sure say that yes that ballast will run hotter However it will stay well clear of the 130°C winding temperature.

I am also rather suspicious of your claims regarding the other ballast, but to be honest i cannot be bothered to calculate every single one of them. If that interest you get yourself a good book about electrical machines and do it yourself. If you have questions feel free to ask.
However what @RRK said sounds reasonable.

Best regards,

Alex

It gets really bothersome when different collectors share different conflicting opinions and try to correct me on them and being defensive of certain pieces of information that they share.

Sometimes technology can be complicated and it is always important to remember that when someone tries to correct you on a technical thing, to not take it as an attack on your person. (...)'s complicated when you're working with power factors, real and imaginary components of impedance and all that stuff. Hell, even the DC resistance of a choke can't always simply be ignored. Personally i still don't have a thorough understanding about why arc voltage has an effect on heat production in a series choke ballast, when the current is otherwise fine. 
It's staggering how something as simple as a choke in series with a tube is actually so complicated. I know i've shared wrong information before...

Anyway, i'm still toying with the idea of using a variac wired up as a choke to be an universal ballast for any and all tubes. As long as the inductance is large enough you should be able to set it to the value that makes you reach the exact lamp current (rms) you need.

Ballast can get comicated,  fue to impedance and the current u trying to push thhiugh it and the line  voltage.   

  Here in US u can see affect of choke ballast edp the 20 watt
Preheat chokes u  the Advance LC1420 for example  it will run a 14 watt to 22 watt lamps.     14  watt lamp have lowest arc voltage of 30-40 volt.   That lamp will work but ballast runsbit on hot side.   While U take sanw choke and 17 watt T8 2 footer
  That lamp 70 arc volt and lamp only need 270 mA.  Putting it on that choke balladt because of higher arc voltage the ballast rubs cooler whille  ballast is shoving 330 mA into the 270 mA 17 wat tube.   It forcing 22 watts of poeer into the lamp and ballast is running cooler than it ever did before.


   What I am saying  here U hace such a extreame low arc voltsge and high current lamps.   U have figure out how to take up 215 V out of 230 volt.   Yhe result will usually end up in Heat
  So let say have North american HPS choke.    Arc volt drop is 55 volt  but u have 120  volt.    If u add another ballast in series
 U can drop lamp aex vvolt down to 20 -25 volt  also  u can had caocitorrs in seriee to help fine tune the two balladt  to match the lamp requirement.   Thiing  is it need expermentijg bit also  montior the ballast Temp to not over heat them where u start burning the varnish off the chike windings

Yes, choke ballasts will run hotter when you run extreme low voltage lamps. There two reasons:
1) higher current through the ballast, Pv~I^2.
2) Higher core losses, due to higher voltage drop over the ballast: Pµ~U^2.

Regarding the North American son lamp ballast i cannot say anything as I have no knowledge about it. Regarding the 80W FL ballast described by RRK I see no problem as these ballast need to withstand a stuck starter.

@Laurens I agree. Lamps and ballasts are something difficult due to their highly non linear nature and them being operated dynamically. For example point two above is not correct on detail, as the magnetization curve of Iron is non linear...However that is to my understanding not relevant in this case. There is likely no analytical solution for these problems bit only better or worse approximations. For most lamps these approximations are good enough and it can be fine tuned with a variac.

@funkybulb ignoring the DC resistance of a choke is in my opinion not a good idea as it brings too much error in the approximation. I agree that higher voltage drop across the ballast likely means more heat in the same, if it will kill it however is a different kettle of fish.

@WorldwideHIDCollectorUSA I don't get it. I explained one case. That took me 1.5 hour to write that post with the research. I wrote were you can find additional information. I can't see were i was defensive with certain knowledge. If you have direct questions i will do my best to answer them.
This is not a personal attack, i have no reason for that. It is a complicated technological topic. I understand that. However there is a reason it takes 5+ years to get an electrical engineering diploma (Or master of Science however you wonna call it). That is not something that can't be explained with a few posts on an internet forum, I wish it was, than we would't have a lack of engineers...

I am currently working on a step by step "instruction" to mach unsual lamp ballast combinations, however there again to what i elaborated before you will need basical electrical and mathematical understanding to do that and i am afraid that the basic stuff from 8th grad is just not enough.