eem2am
Banned
Hello,
I am wondering about how an offline current-mode heating, half-bridge fluorescent ballast driver “should” actually best work.
I mean one with an inductor in series with the lamp
..and with a resonant capacitor across the lamp.
..also including a series DC blocking capacitor
as on page one of this........
******************
http://www.irf.com/product-info/datasheets/data/ir2520d.pdf
Is it true that………..
1. It should start by oscillating at twice the LC resonant frequency to warm up the lamp filaments.
2. The frequency should then be gradually reduced towards a “target” frequency which is 10% BELOW the LC resonant frequency
3. When the lamp has been struck, the ballast controller should immediately detect this somehow….
And if the frequency at lamp “strike” is below the LC resonant frequency, then the frequency should be increased to the LC resonant frequency.
4. The ballast controller should then engage in Zero-Voltage-Switching detection of the (lower) half-bridge mosfet.
5. If non Zero-voltage –switching is detected, then the half-bridge frequency should be incrementally raised until zero-voltage switching occurs.
6. The controller should keep its frequency regulated to the lowest frequency above the LC resonant frequency which ensures zero voltage switching.
Unfortunately, the above “ideal” modus operandi gives some problems…………..
Due to the fact that the resonant capacitor is likely to have a tolerance of +-20%, we will not know what is the ACTUAL resonant frequency of one of a mass produced batch of ballasted lamps.
-This is made worse by the tolerance on the ballast inductor –and also by the tolerance on the lamps impedance.
Basically, the regulation is centred around adjusting the frequency to get zero voltage switching.
The higher the frequency, the lower the power and the lower the light output.
Unfortunately, the regulation cannot also be centred around setting the ACTUAL lamp current.
-Therefore, in any fluorescent ballast, we CANNOT be sure just exactly what the power and light output will “actually” be.
-The power achieved could easily be 20% below or above what we want.
So the customer may well have a lamp that’s too dim.
How do we go about solving this?
I am wondering about how an offline current-mode heating, half-bridge fluorescent ballast driver “should” actually best work.
I mean one with an inductor in series with the lamp
..and with a resonant capacitor across the lamp.
..also including a series DC blocking capacitor
as on page one of this........
******************
http://www.irf.com/product-info/datasheets/data/ir2520d.pdf
Is it true that………..
1. It should start by oscillating at twice the LC resonant frequency to warm up the lamp filaments.
2. The frequency should then be gradually reduced towards a “target” frequency which is 10% BELOW the LC resonant frequency
3. When the lamp has been struck, the ballast controller should immediately detect this somehow….
And if the frequency at lamp “strike” is below the LC resonant frequency, then the frequency should be increased to the LC resonant frequency.
4. The ballast controller should then engage in Zero-Voltage-Switching detection of the (lower) half-bridge mosfet.
5. If non Zero-voltage –switching is detected, then the half-bridge frequency should be incrementally raised until zero-voltage switching occurs.
6. The controller should keep its frequency regulated to the lowest frequency above the LC resonant frequency which ensures zero voltage switching.
Unfortunately, the above “ideal” modus operandi gives some problems…………..
Due to the fact that the resonant capacitor is likely to have a tolerance of +-20%, we will not know what is the ACTUAL resonant frequency of one of a mass produced batch of ballasted lamps.
-This is made worse by the tolerance on the ballast inductor –and also by the tolerance on the lamps impedance.
Basically, the regulation is centred around adjusting the frequency to get zero voltage switching.
The higher the frequency, the lower the power and the lower the light output.
Unfortunately, the regulation cannot also be centred around setting the ACTUAL lamp current.
-Therefore, in any fluorescent ballast, we CANNOT be sure just exactly what the power and light output will “actually” be.
-The power achieved could easily be 20% below or above what we want.
So the customer may well have a lamp that’s too dim.
How do we go about solving this?