Hi to all
In my industrial application I would be able to DIM the light only ON and OFF, but with variable frequency.
In some machine I need to make photos slowly (10Hz) syncronized with LED light,
while in another machine I have two cameras working at about 2000 fps synchronized with 2 LED strings that must be mutually exclusive turned ON.
In other words for me it would be sufficient to drive the DIM pin with a 50% duty cicle square wave with frequency from DC to 10kHz.
Finally (much less critical) I would be able to configure the output current in small steps covering the whole range 0-1.5A.
In my board I've two strange behaviours, they happens in two other boards, including the Digikey evaluation board, so they seem due to the LM3406HV.
After testing I have chosen the followings as best tradeoff between ripple current, switching frequency and dimming "precision" (see below):
Ron = 200K
L = 100uH
No output capacitor
Issue #1, very important:
Especially when working with high power leds driven nearly at they maximum level, it's desiderable to use the output capacitor for reducing the ripple current.
Initially I chose Cout=1uF 100V X7R ceramic capacitor, it helps in both configurations: in parallel with leds and between LED+ and GND (like Digikey evaluation board).
But both configurations have the same problem: when dimming the output capacitor keep the output voltage high when DIM=0, and in the next rising edge of DIM pin
the LM3406 fail to start immediately. Depending on the output voltage (time elapsed from last turn off) the LM3406 try 0, 1 or 2 times to start.
What it's very strange is that it seem it disable the output mosfet for 100us due to a Peak Current detection as explained in the datasheet, but sometimes this happen after a single shot of the mosfet, and calculation shows that the inductor current shouldn't be more than 100mA at this time.
What's happening?
Since there may be a random delay up to 200us after DIM=ON, and this happens always when using an output capacitor, at the moment I'm working in a sub-optimal configuration with a bigger output ripple.
The following two screenshot show two typical traces when using Cout from LED+ and LED- or from LED+ and GND.
green trace = external DIM command, very similar to DIM pin
yellow trace = output voltage (LED+)
red trace = Vsw voltage, Vin=48V
cyan trace = Vsns voltage
Vin=48V Vout=20V Iout=750mA Cout=1uF to LED- Dim=50Hz
Vin=48V Vout=20V Iout=450mA Cout=1uF to GND Dim=50Hz
Issue #2, less important:
With low output current (<150mA) the voltage ripple measured at the CS is very high, in some case more that 1Vpp and sometime with Vsns going negative, too.
Vin=48V Vout=20V Iout=050mA Vsns=624mV
yellow trace = LED+ output voltage, AC coupled, about 20V ± 1V
red trace = Vsw voltage, Vin=48V
cyan trace = Vsns voltage, well outside OVP/OCP limits
Page 10 of datasheet report: "The CS pin includes an output over-voltage/over-current comparator that will disable the power MOSFET whenever VSNS exceeds 300 mV."
I see the mosfet switching with VSNS=500mV and more, why? It seem that OVP/OCP doens't work. Can the board operate safely/correctly in this case?
Thanks in advance,
Andrea