Are these buck converter waveforms OK? (LTspice simulation)

[exscape]

I'm working on a high power (~60 W) white LED driver. I might even be done with it (PCB layout remains, but nothing else) -- except I noticed some worrying things in my LTspice simulations.
I've never worked with DC/DC converters prior, so all I know is what I've learned while working on this for the past few weeks.

I'm using the LT3755-2 LED driver controller.

Here's the schematic, and the waveforms when dimmed low (in DCM, I assume):


Gate voltage and inductor current:


Diode current (positive = forward conduction)


MOSFET Ids


The reverse current spikes through the diode are just under 4 ns long. Are they real, or a simulation artifact? If real, are they a problem?

When the dimming voltage is higher (load current between about 0.4 and 4.5 A), the waveforms look roughly like this (screens are at 100% dimming):

Load and inductor current:


Diode current:


MOSFET Ids:


My worries are:
1) The negative inductor current when ringing (in DCM)
2) Negative current spikes through the diode when switching (DCM and CCM)
3) Negative current through the MOSFET while ringing (in DCM)

Are all of these OK? If not, how should I work to solve them?
This is a hobbyist project, so EMI isn't a huge issue; however, I'm doing this to learn, so I'd of course prefer the design to be at least acceptable from an EMI standpoint.

[T3sl4co1l]

The controller seems to be doing what it's supposed to.

If you don't want such sharp edges, try some gate resistance.  I don't get why LT never shows them; their FAEs are pretty hopeless when it comes to EMI as well.

I don't get why the thing needs two current senses.  You already know one from the other.  Not a very smart controller design on their part.

The critical layout loop is between R2, Q1, D1 and C5. Cram them together as close as possible, connected with as much trace width as possible (usually with polygons/pours).

BTW, look up the capacitance drop vs. voltage for any large ceramic caps you're using.

Tim

[Benta]

Looking at the data sheet, it doesn't seem that LT intended the part to be used the way you are...

[exscape]

Looking at the data sheet, it doesn't seem that LT intended the part to be used the way you are...

Hmm, in what way? Do you mean using only analog dimming, or something else?
The circuit is based on the "Buck mode 1.4 A LED driver" circuit on page 20, only simplified as I'm not interested in using PWM dimming.

[Wimberleytech]

I don't get why the thing needs two current senses.  You already know one from the other.  Not a very smart controller design on their part.

The controller can be configured in multiple ways.  In at least one configuration the two current-sense paths are independent.