I've been thinking about making this ballast https://www.eleccircuit.com/fluorescent-light-by-6v-power/ to make my own emergency setup however I'm not sure about all the connections. How would the fluorescent tube actually be installed, 1 pin connection on each side and what is the GND at the bottom right of the first setup?

Also what do I do to make the 12v kick in when no power is detected?

I would rather go with something along this or these lines.

The first one is really a minimalistic ballast. Transistor would be 2SB772, add about 10..15Ohm in series with the feedback winding to limit BE reverse current during lamp startup, the secondary would have about 66 turns.


The second can use e.g. 2SD882, add about 10Ohm in series with C2 (limitting the current when E-B is driven into breakdown during lamp startup).
For a 4W lamp the secondary (6-1) will have just about 60 turns, the primary (1-2) about 16 turns for 6V feed.
The fluorescent pin labeled as A3 I would disconnect from the transformer (don't make the tap there), so the lamp operates in the rectificationn mode (way more stable; with these short and rather thick lamp that is not a problem). Then the top side of the resistors can be connected to the A3, to serve as a missing lamp lockout (the transistor bias will be off when lamp removed).

Why do you prefer the other methods?

I just realized that why do we need this complex stuff when all is needed is 36vdc, when power goes out, quick short and the tubes on. Now how long it'll last, i have no idea. At lower voltages like 36vdc, there's no mercury migration.

First I personally see maintaining a 36V DC power as a more complicated thing than one of these ballasts operating from way more convenient 6 or 12V.
Second Even at DC, the discharge needs a ballast, something that would regulate the current through the lamp. At DC it could ge either a switching regulator (the single transistor inverters are practically a flyback type of converters), or some current regulator or a resistor (an incandescent,...). So for the resistive ballasted F4T5 you would need more than a 36V source. The 36V is just the arc drop, you need some extra room for the ballasting (a resistor,...; at least 18V for the end-of-discharge battery voltage).
And last the F4T5 is already rather inefficient as it is, adding big ballast losses and you are barely on par with an incandescent.


The mercury migration is not that much tied to the voltage, but more to the lamp shape and its burning position (so the electrolysis is opposed by the distillation and gravity). But most of those single transistor ballasts feed the lamp by a DC as well, the rectifier is the lamp alone.

I've noticed that with a mercury starved f20t12, the mercury was always at the low end (it was burned vertical).

Well I'll look into eventually making those circuits one day. What is the difference in the ones you provided and lantern ballasts?

Which lantern ballast?
Generally these low power battery powered fluorescent circuits are pretty much all the same. The beauty of these circuits is, they pretty well respond to the ionization inside of the lamp, so maintain it (so the power) quite stable, even when being so simple.

Like fl450 lantern ballasts, they're 6v ballasts but they're go as low as like 4v

Generally two circuit types are used:
The single transistor flyback and then a two transistor Royer oscillator type.
The second (more modern) is completely insensitive to transistor parameters (the power is just dictated by the transformer inductance, turns ratio snd the circuit capacitances), operates the lamp at AC but is not able to compensate for battery voltage in any way and making it other than cold start means added complexity.
But the varying power is usually not that big deal, the lamps tend to be quite tolerant.

The flyback means a single transistor, a transformer with one primary, one feedback, one secondary with a tap (~3V from the end that is negative when the transistor is off), two capacitors and two resistors. The result is a kind of RS circuit, which either stsrts the lamp very quick (full battery / max voltage), or it first warms the cathode (moderate to lower limit), or in case of too low voltage, it is just warming the electrode but does not start the discharge. The result is, the lamp would practically never run with cold cathode for longer than a second or so.

The Royer oscillator just "transform" the battery voltage into a secondary and then feed the lamp via a capacitive reactance. So when the battery gets low, it may continue to feed the lamp at low current, so it won't warm up its electrodes. Plus as the lamp transitions from low current cold electrodes to high current hot electrode mode, the ballast draws way greater power for a brief time. But when the battery is not able to deliver that (and sags its voltage), the lamp does not turn into the hot electrode mode, so the electrodes suffer a lot from the sputtering.

In more industrial applications, I've seen fixtures with 4 bulbs, a standard ballast and emergency ballast, what type would that be?

Either of them. Really both are used...

Interesting, in warehouses with hid highbays, how would emergency lighting work?

Mostly separate fixtures turning ON only in case the power fails. Plus halogen lamps designed to cover the HID restrike time.
Plus there the AC power has a backup from generators, so unless the cause is something really inside of the building (fire,...), the power gets restored there within minutes...

I remember at places like Costco, there were some lights that now that I think about it were halogen colored because the bulb was dead, is that what you mean?

Could be. It was a small halogen capsule behind the main bulb... And yes, the halogen lights up whenever the main bulb is off, so even when it is e.g. dead. And of course, the power has to be present.

Oh wow I never knew that till now!