BuckBoost converter PCB review

[Alex_Baker]

Hello!

Here I have a full-bridge buck-boost converter that I designed. I have done a handful of buck converters but not one this complex, so I figured some feedback would be a good idea.
The converter is based on the Texas Instruments LM51770, most of the circuit design came from TIs online power designer tool.

The input to this converter is DC from a brushed motor running as a generator, the voltage will range between 8 and 25 volts, so I designed for 6 to 30 volts. The generator is only rated for 10 amps, so I put the LM51770's auxiliary current shunt on the input. The output is meant to be a fixed 13.5V for a system based on a 12V lead-acid battery.

I tried to keep this converter as inexpensive as possible without cutting corners on components, this is the reason for the two layer board. I mounted the mosfets on the bottom of the board so that I could bolt it flat to a heatsink.

The biggest differences from the TI power designer are the addition of J4 as well as Q1 and D1. J4 lets me adjust the operating mode slightly, and Q1+D1 is effectively a fault/power good indicator.

Feel free to ask many questions, thanks!

[boz]

I am kind of surprised TI do not stipulate gate resistors to control the ringing, 1.8Mhz is very high and the evaluation board has 3 ohm ones in place, wouldnt hurt to stick some in

[Alex_Baker]

The frequency programmming resistor is 78k, which should set Fsw around 400khz. Should gate resistors still be used at this speed? Also are the differential pair traces for mosfet gate drive needed at this speed? The datasheet makes a big deal about it.

[Magnethicc]

going to need to see those gate drives waveforms and EMI/EMC results.

* Signals under the inductor can radiate from the magnetic field from the inductor as well as from the current induced by them passing through the high frequency current loop of the converter.
* There is no solid ground plane which is too bad because it will do wonders on the noise in the system.
* I would swap places between the ceramics and the elko capacitors (at the input and output) since this will make the high frequency current loop smaller.
* Add some gate resistance to limit the current out of/into the IC at the switching times as well as transient decoupling through the Cdg/Cgs of the mosfets.
* Some reverse polarity would add to the robustness.
* you said you designed it to 30V but the MOSFET itself is 25V, there is no margin for voltage overshoots at all which most likely will be.
* Is L1 able to handle the current without saturating with enough margin?

[Alex_Baker]

Thanks for the feedback.

I changed mosfet parts at the last minute, I must have missed the voltage rating, my bad, the last one was 50V I think. Finding another in that package should be no problem.

I need to compare the cost of a 2 vs 4 layer pcb, if 4 layer is not too pricey I will definately design with that.

The inductor I selected is rated 30A without saturating, irrc. It is a shielded type inductor if this makes any difference to noise.

[boz]

The frequency programmming resistor is 78k, which should set Fsw around 400khz. Should gate resistors still be used at this speed? Also are the differential pair traces for mosfet gate drive needed at this speed? The datasheet makes a big deal about it.

there is generally no harm, you can always substitute 0ohm ones, and ringing is a real PITA

[Alex_Baker]

I should have checked pcb prices before the design, 4 layer is still very cheap from pcbway 

Also I will be adding gate resistors and updating the mosfet part numbers.