Oznog
Active Member
I'm trying to use the HV9901 PWM LED driver **broken link removed**
for a red Lamina LED that needs 2.1 amps @ 6.7v.
I'm using the prescribed Buck converter config, the Vin is 15v, more than twice the output so the duty is under 50% (they're unstable over 50%). I've got a 220uH toroid which has a 2.4 amp thermal rating and no Isat listed, a small but very low rds-on IRF7811 transistor, and a 0.12 ohm current sense resistor, and a 1N5821 3-amp Schottky. I wanted to run it at 100kHz.
So here's the thing. It wasn't working right despite the simplicity of the circuit, glitches in the gate signal and it looked like really high current spikes on the current sense.
Changing the diode to a 1N4935 1 amp fast 150nS silicon diode makes it run clear as day. To me, it looks like maybe the high speed switching performance of the Schottky was flawed and I don't understand why. It seems to me like it didn't shut off fast enough and in this config that will shoot current from the rail, which would make a spike through the current sense resistor, which would instantly be read as time to turn off the PWM output latch because it would think that's the inductor current. There was all kinds of noise going on on the current sense resistor voltage and the PWM chip's gate output. I thoroughly verified that the Vin, Vdd, and GND pins were not seeing a lot of switching noise. Added ceramic caps to be sure. No help.
I found they list part numbers on their demo board schematics. They chose a ultrafast recovery, non-Schottky diode.
I must say, all I know about Schottky is it's a low forward drop, a bit high on reverse leakage, and it's supposed to be "fast", usually they don't list a recovery time though I did see a site list the 1N5821 as "under 10nS". If it's so fast why was there this current spike which didn't happen in the 150nS silicon diode?
for a red Lamina LED that needs 2.1 amps @ 6.7v.
I'm using the prescribed Buck converter config, the Vin is 15v, more than twice the output so the duty is under 50% (they're unstable over 50%). I've got a 220uH toroid which has a 2.4 amp thermal rating and no Isat listed, a small but very low rds-on IRF7811 transistor, and a 0.12 ohm current sense resistor, and a 1N5821 3-amp Schottky. I wanted to run it at 100kHz.
So here's the thing. It wasn't working right despite the simplicity of the circuit, glitches in the gate signal and it looked like really high current spikes on the current sense.
Changing the diode to a 1N4935 1 amp fast 150nS silicon diode makes it run clear as day. To me, it looks like maybe the high speed switching performance of the Schottky was flawed and I don't understand why. It seems to me like it didn't shut off fast enough and in this config that will shoot current from the rail, which would make a spike through the current sense resistor, which would instantly be read as time to turn off the PWM output latch because it would think that's the inductor current. There was all kinds of noise going on on the current sense resistor voltage and the PWM chip's gate output. I thoroughly verified that the Vin, Vdd, and GND pins were not seeing a lot of switching noise. Added ceramic caps to be sure. No help.
I found they list part numbers on their demo board schematics. They chose a ultrafast recovery, non-Schottky diode.
I must say, all I know about Schottky is it's a low forward drop, a bit high on reverse leakage, and it's supposed to be "fast", usually they don't list a recovery time though I did see a site list the 1N5821 as "under 10nS". If it's so fast why was there this current spike which didn't happen in the 150nS silicon diode?
