This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

LM5104: Duty cycle accuracy and propagation delays at high frequency operation

Part Number: LM5104

I've designed a current regulator around LM5104, intending to operate at around 500kHz (period = 2us).

I knew there would be some minimum functional duty cycle and that it would depend on the switching frequency and various delays, but the limitations I found left me wondering how people possibly design circuits using this or similar gate drivers which operate at 1MHz or even higher.

I imported the LM5104 spice model into LTspice and it matches my actual measurements fairly well:

Assume my Rload is about 10ohms, Rgate is about 10 ohms or less, and that my half bridge supply voltage is 36V.
I've reduced the Rdt resistor to the lowest value recommended in the datasheet, 5k.
Period of the PWM input = 2us (500kHz operation).

Let's run the sim at a few duty cycles and see how the propagation delays and effective duty differs:

50% duty cycle (1us on time).

VPWM input (green), LO gate drive output (red), HO gate drive output (blue), VH half bridge voltage (green), and high side FET gate-source voltage (magenta).

Now let's run it at 25% (0.5us on time)

And 10% (0.2us on time) (zoomed in on horizontal time scale):

And now let's compare some of the timing parameters and duty cycle error:

It makes sense, with fixed propagation delays that low duty cycles will eventually become very inaccurate especially at higher frequencies.

I expected to need to confine my control loop to operation within some bounds, and am now assuming I'll use 10-90% or so (bounded on the upper side by the need to charge my boot capacitor).

My question is this: How exactly are designers making 1MHz and higher switching regulators with these types of gate drivers? Comparable devices from TI and other suppliers all have propagation delays on the order of tens to hundreds of nanoseconds for some of their parameters.

Is it just assumed you'll bound your control loop to smaller and smaller permitted operating ranges, say 30-70%, and then your control loop will simply move around in that space as needed to regulate your output?