LM3410 as buck-boost

I am not sure the exact circuitry they had in mind, but I'm guessing it is similar to the high side current sense used in the LM3429. It would just be a rail-to-rail differential amplifier with the gain adjusted to give 190mV at the FB pin when the current through the resistor is correct. But in reality it would be better, and more accurate, to use the LM3410 in SEPIC. It simply changes the inductor to a coupled inductor and adds a capacitor. But that is less complex than adding a diff amp and you will not be adding more errors.

The SEPIC design is tempting, although I am very space constrained and adding a second inductor would be difficult. I also need to handle a relatively high input voltage (20V) and LED current (1A), which eliminates the LM3410 and leaves relatively few options (I realize this would be an issue for my buck-boost as well, but I was only using the LM3410 as an example as it is the part they used in the article). I feel that being able to do this with one inductor and a commonly available boost regulator would be the smallest, most flexible design.

For the high-side sensing, it looks like I could use a current sense amplifier like the INA199:
www.ti.com/.../INA199

I figure I put the amplifier across my high-side sense resistor, connect the output to the FB pin of my boost regulator, and adjust the gain and resistor to give me my desired LED current.

I'm hoping that, before going too far down that path, I can get some feedback about that. It does look like that's what's inside the LM3429, so perhaps it will be just fine if the amplifier is outside the regulator instead.

If you can provide the full input voltage range, LED stack voltage, and full LED current plus any other requirements I would be happy to give you a recommendation. But based on an input up to 20V, 1A of LED current, and you want a single inductor buck-boost you should check the LM3421, LM3424, LM3429, or the TPS92691. They are all capable of high-side sensing and most of them likely have a buck-boost application similar to what you need in the datasheet or as an EVM.

For the sake of argument, let's say I wanted to further explore the idea of converting a high-side current sense for use with the LM3410.  I looked closer at the INA199 current sense amplifier and my concern is that the bandwidth would be a problem - a max of 80KHz, and the LM3410 will have a switching speed around 500KHz.

An alternative is a current monitor, like the INA138.  This seems to do exactly what I want, and with a bandwidth of 800KHz.  How do you think the LM3410 (or other low-side sensing regulator) would work with the INA138 providing the low-side feedback signal?

This is interesting. I would think that:
- you'd want to copy the current sense voltage as exactly as possible. Low side and high side resistors would see the same current, so if the low-side circuit from the "typical application" saw the switching current, you'd want to pass that on from the high side as well.

- I thought that many of these converters DO work by reacting to the switching - that they monitor the current ramping up while the switch is closed, and make decisions based on that. Or perhaps that is just the hysteretic controllers, and the LM3410 is based more on an average current?

Hysteretic type converters do have other considerations, but the LM3410 is a constant frequency current mode converter. It also uses an output capacitor so the FB pin is really just regulating to a mostly DC signal of 190mV. So you do not want all of that current ripple imposed upon it.