I have an LED driver circuit that is not behaving the way I expect it to. More precisely, it’s not drawing as much current as I expect it to do. The design is not originally mine, so I don’t know the motivation behind all design choices, but I suspect the design was mostly copied from the evaluation board for the LED driver that is used in the design. The schematic of the circuit is included below. It uses an NCL30160 LED driver to drive 4 different LEDs with wavelengths of 365, 400, 480 and 520 nm. The selection between the different LEDs is done with a BTS5200-4EKA high side switch, and the intensity of the LEDs is controlled by PWM (PC_PWM in the schematic)
The LEDs are from a company called Lumixtar with the following part numbers:
The 365, 400 and 480 nm LEDs have a listed forward voltage of 3.4V, while the 520 nm one has a listed forward voltage of 5.5V. All LEDs can handle a current of 700 mA. The design choices made for the NCL30160 LED driver are the current sense resistor that controls the current (R12 in the schematic), the large inductor (L2), off-time setting resistor (R14), catch diode (D3) and input capacitor (C2). The design seems to work fine, except for the issue that I’m not getting the current consumption that I thought I would. The selection of current sense resistor has been made to 0.27 Ohm, which according to the datasheet for the NCL30160 should result in an output current of 200 mV / 0.27 Ohm = 0.74A. However, I’m only getting a maximum of 233 mA of current. I’ve measured the current draw from a PSU at different PWM duty cycles and for the different LEDs and put it in the table below:
So I pulled out the oscilloscope to take a look at what’s going on. Below you can see in yellow the voltage of the LED- net label, between the inductor and the LEDs in the schematic above, and in blue the current measurement from the PSU, measured with a 100mV/A current clamp. These measurements are with 30%, 60% and 90% PWM duty cycles. I’m not entirely sure what’s going on, since I haven’t worked with LED drivers before, but from looking at that image and reading the datasheet I would guess that it does sort of what it’s supposed to do? The current ramps up until it hits a threshold, at which the FET starts rapidly switching on and off, depending on the tON and tOFF of the driver.
My first theory was that the inductor selection was off, that with a larger inductor there would be less ripple and a higher average current, so I dug into the calculations for the inductor selection, which comes from the calculations for the tON and tOFF.
After not being able to apply the equations properly I did what most engineers would probably have done, I simply swapped the inductor for something different to see what happened. I swapped the 100µH, 300mΩ inductor for a 15µH, 100mΩ one I had lying around, and then measured the current consumption again, with the results below.
There’s some strange hysteresis effect going on between 90-100%. When I increase the duty cycle up from 90 -> 100% there’s a jump in current consumption and radical change in the switching frequency somewhere around 94-96%. Going back down it doesn’t “reset” until somewhere 92-94%.
The overall change in current consumption was negative, which I guess makes sense? The voltage regulation swings a lot more, which results in less overall current going through the LEDs. So, ignoring the consumption results ata 100% PWM the theory about larger inductor holds? So I then tried swapping the 100 µH inductor out for 2x 100µH inductors in series (I didn’t have any single SMD inductor at hand with a value larger than 100 µH, and stacking large components on PCBs is fun!) This didn’t have any impact on the current consumption at all pretty much unfortunately, the results were pretty much identical to the first table in this post.
I then tested the swap the last component I could imagine would affect the current consumption, the off-time setting resistor. I didn’t expect this to have a large effect on the current consumption since the datasheet specifically states “The off-time setting resistor (ROT) programs the NCL30160 with the initial time duration that the MOSFET is turned off when the switching operation begins”, which I interpreted as that it shouldn’t have an effect on the continual switching, only the initial switch cycle. I changed it from 240k to 120k, and measured current consumption again, with the results below.
This also didn’t change the overall current consumption, nor the shape of the wave forms, which I didn’t really expect it to.
So it’s at this point that I figured that I don’t really understand what’s going on here, and how the LED driver operates. I’m expecting the output current to be 0.74A continuously, based on the datasheet, but maybe that’s not the correct assumption? Maybe the LED driver is working exactly as it’s supposed to?