Fluorecent choke for power supply

I used a 250W mercury vapour ballast (choke) for a CLC filter once and it worked fine even though it was not designed for DC. The henry value can be calculated by knowing the details of the lamp it drives.

Low loss types have a lower DC resistance and more inductance, for comparison @240V 36W, a standard ballast has a 40 ohm DC resistance and a low loss type has a 23 ohm DC resistance.

80 ohms seems like a lot of DC resistance even if the ballast were designed for a 277V supply it should have a similar or slightly lower DC resistance than the 240V 50Hz ballasts I measured.
 

Elvee

Member
2006-09-08 2:04 pm
DC will reduce the inductance, as these chokes won't have an air gap. You need to measure it with DC present, or just build a PSU and measure the output. What you can't do is measure the inductance at zero DC and then simulate a circuit assuming it retains that inductance.
They have distributed high reluctance sections, because they have to operate as power inductors, and are not allowed to saturate significantly within their operating range.
The fact that they operate in AC changes nothing.

The parameter to take into consideration for saturation is the instantaneous current.

I just measured a 15W ballast (for 230V). The (cold) DC resistance is 56 ohm, and the inductance is 1.56H (under small signal; at the nominal current, it is normally higher).
The maximum rms current is 0.33A --> 468mA instantaneous.
It could be used up to that current in DC, except it will overheat.

1.5H is not much, but it can be useful tough, and it could be parallel-tuned for 100Hz.
 

Elvee

Member
2006-09-08 2:04 pm
Does it mean we can parallel the choke and ideally it should behave as follow..

50Ohms ( 1.5H ) || 50 Ohms (1.5H) = 25Ohms (3H)

and the max allowable current will double 0.33 A x 2 = 0.66A ?

No, I mean you can parallel it with a 1.69µF capacitor to increase the rejection at 100Hz (~30dB), provided you already have a conventional filter to remove higher harmonics.
 
1.5H is not much, but it can be useful tough, and it could be parallel-tuned for 100Hz.
actually,is enough
i did simulations of same capacity used in multiple CLC filter, and compared to single choke 10H vs two 1H
/hammond has low current types at hi inductance, i was curious what is better- if one heavy or two smaller chokes/

believe or not, the two choke was better, and this has decreasing ripple with increasing current. (second one starts kickin in)
but single type had increasing ripple with

36W lamp choke /i have no LCmeter at home/ seems has ~0,94H 2,7µF paratune
(according to lamp datasheets [email protected],43A)
18W lamp should have ~1,5H,
58W 0,57H
difference between currents are not high, therefore 18W seems best

also tried primary of 230 transformer as choke, but it failed-no filtering at all(obviously), but these puppies works for me
 

alka161

Member
2009-03-22 2:22 pm
I just measured a 15W ballast (for 230V). The (cold) DC resistance is 56 ohm, and the inductance is 1.56H (under small signal; at the nominal current, it is normally higher).
The maximum rms current is 0.33A --> 468mA instantaneous.

Strange. I measured the my choke by LCR meter, this is the result:
DC resistant of ~ 59 Ohms and the inductance is 115 mH
UFOlder.ru ?????????? ??????? ??????, ???????? ??????, ????? ???????.
[IMGDEAD]http://ufolder.ru/files/34947817[/IMGDEAD]
 

Elvee

Member
2006-09-08 2:04 pm
Strange. I measured the my choke by LCR meter, this is the result:
DC resistant of ~ 59 Ohms and the inductance is 115 mH
UFOlder.ru ?????????? ??????? ??????, ???????? ??????, ????? ???????.
[IMGDEAD]http://ufolder.ru/files/34947817[/IMGDEAD]
Then, either your LCR meter does not work, or your ballast is damaged:

A fluorescent tube drops around ~100V; the reactance of your ballast is 0.115*6.28*50=36Ω, which results in a total impedance of 69Ω.
The current would be 130/69=1.88A, thus a power of ~188W in the tube: not very realistic
 
Those chokes can be very usable in a PS. As allready mentioned, they are up to about 1,5H. If needed, you can use similar units in parallel/serial connections. I had the best results with "old" steel encased types (Philips) that have low dc-losses, low strayfields and no audible noise. If you want to know the exact inductance you must measure them in the actual circuit with the actual dc current applied (LCR meter readings may vary wildly and are most often worse than a educated guess).