Dear Power by Linear,
I'm having trouble with a buck regulator based on the LT3763 control IC. I'm using it as a voltage regulator, with an input of 54.0VDC, an output of 46.0VDC, and a maximum output current of 10A. When I apply input power, the output voltage reaches 46.0VDC like it should, but then it immediately begins to drift downwards, and eventually settles at around 12-15VDC. All of my testing is being done with no load at the output. The switching frequency is 160 kHz. I've checked many different waveforms and compared the circuit against an LTspice simulation that's as accurate to real life as I can make it. I found that the BG pin doesn't pulse - not even during soft-start. In LTspice, even with no-load, the BG pin pulses periodically to refresh the bootstrap capacitor for the high-side FET drive, but since the BG never pulses, eventually the boot cap in my real circuit discharges, and so then the output voltage drifts downwards.
Ch.1 = Vout, Ch.3 = BG pin, Ch.4 = inductor current
Ch.1 = Vout, Ch.3 = TG pin (WRT GND), Ch.4 = inductor current
These captures of the startup look nearly identical to the LTspice, except that in LTspice the BG pin pulses during soft-start and then gives minimum pulses for bootstrap refresh each switching cycle.
The LT3763 datasheet states that the BG drive is disabled when inductor current is less than 1/10th of its nominal value, so naturally I expect this circuit to be in DCM at no-load, but I can't understand why BG doesn't pulse during soft-start, or to refresh the boot cap.
I'm going to try increasing the frequency to the stated minimum of 200 kHz. But what else could disable the BG drive?
Can you try changing MOSFET's as suggested by the data sheet on page 15.
The recommend maximum threshold voltage of the MOSFET used should be less than 2V.
IPA028N08N3 has a typical threshold voltage of 2.8V
I replaced the MOSFETs on two of my four prototypes with the TI CSD18535KCS, which is specified for operation with a Vgs of 4.5V, but I didn't observe any change in the BG pulsing. The switching node does rise and fall faster, but BG still doesn't pulse.
I should also note that I use the exact same PCB for two other applications: 48V to 24V out and 48V in to 12V, with the same MOSFETs, and in those two circuits the BG pin does pulse during softstart and during steady state when the output current crosses a given threshold.
I'm pretty sure that BG should pulse, even if the threshold of the gate it's trying to drive is very high, no? It might not successfully enhance the channel of the FET, but it should at least apply 5V. My system only shows switching noise. I have carefully confirmed with a multimeter that the all the connections are soldered. I get a short from the actual pin of the LT3763 to the gate pin of the low-side FET.
Can you try increasing the INTVcc cap to 22uF?
During your testing have you checked if the IC is indicating a FAULT ?
I just tried this in the lab. I took the stock LT3763 demonstration circuit DC1831A and programmed the output voltage to 45V. I also increased the voltage rating of COUT and I increased the inductance proportionally to the increase in VOUT. BG seemed to work properly according to the scope capture below. Can you use DC1831A to try and troubleshoot your circuit?
Conditions: LT3763EFE, mod DC1831A, 54VIN, 45VOUT at no load, VCOUT IL VBG startup_0.bmp