Hello,
I wish to drive 12 350mA , White LEDs.
(LED V(f) = 3.3V +/-20%. )
I want to sometimes have these LEDs on all the time, and other times I will flash them on/off at ~4Hz to make them noticeable.
My Vin is 10 to 30V (…that is , it’s the battery voltage of a car or lorry, so it could be anywhere between 10 to 30V)
Please assist me with choice of topology?
I have the following choices of topologies to do this…….
1…..Buck converter with 6 parallel strings of 2-LEDs-in-series
(-also including three current mirror circuits, each with two BJTs to match the currents up…..these BJTs will comprise emitter degeneration resistors)
Unfortunately, for worst case led Vf’s , one of the current mirror BJTs would be dissipating 2W.
2….Current Mode SEPIC converter in Discontinuous Conduction Mode.
(all LEDs in one series string)
(using UCC28C42)
UCC28C42 datasheet:-
http://www.ti.com/lit/ds/symlink/ucc28c42.pdf
(-needs to be DCM because the duty cycle can sometimes be very high and CCM would need too much slope compensation, which degrades the FET source current sense)
…one unfortunate thing about this solution is the need for an external opamp to amplify the led string current sense resistor voltage, since the Feedback reference voltage inside the UCC28C42 is 2.5V, and thus otherwise you’d get too much dissipation in the current sense resistor.
-Fortunately SEPIC inductors off-the-shelf, cost little more than single coil inductors.
3….Voltage Mode SEPIC converter.
(all LEDs in one series string)
(using UCC35705)
http://www.ti.com/lit/ds/symlink/ucc35705.pdf
(-this would have feedback compensation for the CCM case when Vin=10V and Vout = 48V, however, it would fall into DCM when Vin was 30V)
…..one unfortunate thing here is the need for an external opamp to drive the Feedback pin of the UCC35705.
4…..Two separate simple Hysteretic Buck-Boost converters
-Each driving a series string of 6 LEDs.
Using ZXLD 1370)
http://www.diodes.com/datasheets/ZXLD1370.pdf
…..The problem here is that the ZXLD1370 is hysteretic, and the LED current setting relies on fixing the peak and trough of the inductor current….and though this accurately sets the current in buck mode, it is inaccurate when in buckboost mode, because of the led Vf variation, which makes hysteretic methods inaccurate for buck-boost mode)
5…..Two separate SEPIC converters each driving 6 series LEDs
(using TPS61500)
http://www.ti.com/lit/ds/symlink/tps61500.pdf
The TPS61500 is cheap, and needs no external opamps as its feedback reference voltage is 200mV.
Also, it has an on-chip FET.
It would need slope compensation, but not that much , since the maximum duty cycle would be just 0.6
So which topology do i choose?
I wish to drive 12 350mA , White LEDs.
(LED V(f) = 3.3V +/-20%. )
I want to sometimes have these LEDs on all the time, and other times I will flash them on/off at ~4Hz to make them noticeable.
My Vin is 10 to 30V (…that is , it’s the battery voltage of a car or lorry, so it could be anywhere between 10 to 30V)
Please assist me with choice of topology?
I have the following choices of topologies to do this…….
1…..Buck converter with 6 parallel strings of 2-LEDs-in-series
(-also including three current mirror circuits, each with two BJTs to match the currents up…..these BJTs will comprise emitter degeneration resistors)
Unfortunately, for worst case led Vf’s , one of the current mirror BJTs would be dissipating 2W.
2….Current Mode SEPIC converter in Discontinuous Conduction Mode.
(all LEDs in one series string)
(using UCC28C42)
UCC28C42 datasheet:-
http://www.ti.com/lit/ds/symlink/ucc28c42.pdf
(-needs to be DCM because the duty cycle can sometimes be very high and CCM would need too much slope compensation, which degrades the FET source current sense)
…one unfortunate thing about this solution is the need for an external opamp to amplify the led string current sense resistor voltage, since the Feedback reference voltage inside the UCC28C42 is 2.5V, and thus otherwise you’d get too much dissipation in the current sense resistor.
-Fortunately SEPIC inductors off-the-shelf, cost little more than single coil inductors.
3….Voltage Mode SEPIC converter.
(all LEDs in one series string)
(using UCC35705)
http://www.ti.com/lit/ds/symlink/ucc35705.pdf
(-this would have feedback compensation for the CCM case when Vin=10V and Vout = 48V, however, it would fall into DCM when Vin was 30V)
…..one unfortunate thing here is the need for an external opamp to drive the Feedback pin of the UCC35705.
4…..Two separate simple Hysteretic Buck-Boost converters
-Each driving a series string of 6 LEDs.
Using ZXLD 1370)
http://www.diodes.com/datasheets/ZXLD1370.pdf
…..The problem here is that the ZXLD1370 is hysteretic, and the LED current setting relies on fixing the peak and trough of the inductor current….and though this accurately sets the current in buck mode, it is inaccurate when in buckboost mode, because of the led Vf variation, which makes hysteretic methods inaccurate for buck-boost mode)
5…..Two separate SEPIC converters each driving 6 series LEDs
(using TPS61500)
http://www.ti.com/lit/ds/symlink/tps61500.pdf
The TPS61500 is cheap, and needs no external opamps as its feedback reference voltage is 200mV.
Also, it has an on-chip FET.
It would need slope compensation, but not that much , since the maximum duty cycle would be just 0.6
So which topology do i choose?
