The Texas Buck driver series, Q8 / Skyray King 2S/4S buck driver RELEASED!

Selecting parts is a fun part, especially if you want to be pretty systematic about it.

A little brushing up. This a reasonable reference for some of the basics, primarily for duty factor, and ripple vs frequency and inductance.

http://rohmfs.rohm.com/en/products/databook/applinote/ic/power/switching_regulator/inductor_calculation_appli-e.pdf

But I'll cliffs notes. (I don't like how casually it treats many of the linear approximations and averages, but ok)

It derives a detailed form of the duty factor including voltages losses in switches and diodes, but then I like the Ti approach of just calling it

V0/(eff*Vin).

The ripple current vs frequency and inductance is useful (and maybe helpful for choosing the inductor size and switching frequency) but also it's included in equation 10 here, a better document anyway:

http://www.ti.com/lit/an/slva390/slva390.pdf

Useful to go through all that, but equation 11 summarizes it. But better to look at 7 and 8 instead. 7 is the mosfet loss and 8 is the freewheel diode loss except they use a mosfet there too. For the diode I guess you have to look at the loss as Vf*I instead of I^2R, simple enough change. So that's losses for both of those components.

Oh and equation 4 gives the inductor loss, which is almost stupidly just I_out^2*R_L. So that's easy to figure too.

However you get to equation 12 and they say blah "+ other losses"

Oops, some nastiness there to figure out, especially switching losses, that looks fun:

http://www.allaboutcircuits.com/technical-articles/switching-losses-effects-on-semiconductors/

Oh, why was Ti using a Mosfet instead of a diode?:

So there's that too, but I don't think it's compatible with your IC.

Then there is discontinuous current mode to worry about at very low power probably.

But that's all just introductory for basic bucks. This 3409 looks pretty fancy (aside from using a diode I guess), with variable frequency, off time control etc. I'm just starting to look at its actual docs. I prefer their math actually. They don't make the approximations so casually, but then they don't start with -LdI/dt either, so that's too bad, unless you only want answers.

Anyway, the obvious things are obvious, keep Rdson low for the switch, diode Vf low, inductance high, frequency high (but not too high, see that link about switching losses, last figure), capacitor ESR matters significantly less, and here already are a few simple guidelines to put numbers on some of it. I think it's really not so bad.

Now this is what's bad (as you know):

11.1 Layout Guidelines The performance of any switching converter depends as much upon the layout of the PCB as the component selection. Following a few simple guidelines will maximimize noise rejection and minimize the generation of EMI within the circuit.

Discontinuous currents are the most likely to generate EMI, therefore take care when routing these paths. The main path for discontinuous current in the LM3409/09HV buck converter contains the input capacitor (CIN), the recirculating diode (D1), the P-channel MOSFET (Q1), and the sense resistor (RSNS). This loop should be kept as small as possible and the connections between all three components should be short and thick to minimize parasitic inductance. In particular, the switch node (where L1, D1 and Q1 connect) should be just large enough to connect the components without excessive heating from the current it carries.

The IADJ, COFF, CSN and CSP pins are all high-impedance control inputs which couple external noise easily, therefore the loops containing these high impedance nodes should be minimized. The most sensitive loop contains the sense resistor (RSNS) which should be placed as close as possible to the CSN and CSP pins to maximize noise rejection. The off-time capacitor (COFF) should be placed close to the COFF and GND pins for the same reason. Finally, if an external resistor (REXT) is used to bias the IADJ pin, it should be placed close to the IADJ and GND pins, also.

I should say that while I have a good 101 understanding of parts selection, picking them from scratch is something I prefer not to do. I tend to get bogged down in the details looking for the “best” possible option.

I spent spent some time looking at parts earlier, I think this mosfet would work ok, it is basically the p-channel version of the Sir800

If that works it makes things simpler since I already have the pads and models for it.

For the diode I found this http://www.digikey.com/scripts/DkSearch/dksus.dll?Detail&itemSeq=208576672&uq=636112551362876866

Only rated for 10A though but I think me might need 2 of them anyways.

The biggest things I need right now are footprints, I started looking at a very basic layout and it is gonna be a tight fit even n a 46mm board, it might have to be a 4 layer PCB.

So without the footprints I can’t do much of the layout.

The real issue I am running into are inductors, it is hard to find them with current ratings of 10-15A and enough inductance to work properly + be small enough to actually fit.

The spreadsheet calculator says that with a .75mhz switching frequency it needs at least a 25uh inductor. It could go up more but mentioned that above 1mzh can be difficult in the data sheet. The smallest I am seeing in that size are around 15x15mm, thats a lot of board space. They are only rated for 9A as well.

Humorous to me all the technical jargon coming from “Flintrock”. Yabba dabba Dooooooo! ;

(sorry, haven’t had coffee yet, a bit delirious from a good nights sleep)

Yep, I paid close attention to that section and it is part of why board space is at a premium and I think that a 4 layer PCB might be a must. To lay it out properly will need a fair amount of board space in order to keep things separated.

Yeah, in the end, it may be fun to see if we can estimate the losses, but there probably aren't that many choices of components really that come close, and you just pick the best from the ones that can fit, if any and hope for the best.

I found a couple of small buck converters on amazon that looked pretty nice. Maybe a couple were upward of 10 A. They were on the right size scale, single board, but not the right shape. Maybe it's worth seeing what they use? Then again, they might not really handle what they say they handle, or not well. I doubt they use any magic component that you don't see on digikey. Another possibility is your standards are too high. Maybe you can squeeze by with less than you think?

And you checked the Coilcraft line-up on the inductors? Arrow Industries is a good place to look as well as DigiKey and Mouser, probably not telling you anything…

WIthout trying hard (first click) I found a mosfet with lower capacitance, but higher Rdson. I don't know, maybe you can push harder on switching frequencies with lower capacitance? Might require running numbers to see what's a good tradeoff.

"In practice, switching frequencies higher than 1MHz may be difficult to obtain due to gate drive limitations, high input voltage, and thermal considerations."

Ooh, "thermal considerations" sounds like power loss to me.

Well a flintrock makes a little more light than a Flinstone at least.

Very possible, I was looking for low resistance when I searched as the calculator was showing something like 30w of power would be dissipated by the default MOSFET at our drive currents.

Thermal considerations are something to think about but I was more worried about the gate drive limitations myself. Thermal can be dealt with, gate drive can not.

Like I said, picking parts from scratch is NOT my strong suite, I am simply trying to find footprints at this point.

This is the closest inductor I could find to our specs: SRP1770TA-220M Bourns Inc. | Inductors, Coils, Chokes | DigiKey

Still only rated for 12A but I figure we could push it a bit to ~15A since the saturation current is 18A.

The issue is it is huge at 17mm, it is also pricey at 3.50 each.

Seriously though, maybe the first component to check is the driver IC.

http://www.ti.com/lit/an/slyt358/slyt358.pdf

"A synchronous converter with an integrated low-side MOSFET offers benefits such as reduced size, lower parts count, and easier design. "

I don't know why.

Look at that figure 3. Those moonlight efficiencies in the synchronous bucks look a bunch better. Synchronous means the diode is replaced with a switch. There seem to be drawbacks at high curent though and that's probably not a good trade. I haven't had time to dig into it.

I was just about to suggest searching on price, highest to lowest. I'd aim this project at quality.

I wonder if the tradeoff from low to high current is Vf*I vs I^2*Rds That would make sense. At low current I^2Rds wins. At high current, Vf*I wins. So.. why not put a switch and a diode in parallel (other than space and price obviously), then you get both? A smart controller would turn the switch off at high current too.

If you have a suggestion for a better buck IC I am all ears, like I said finding parts from scratch is not my strong suite.

Here is a VERY early first draft of the PCB, just put the components onto a 46mm pcb to see how fitment looked. All the small components are 0603 and I am sure some will need to be enlarged so as you can see it fits but there is not a lot of room to spare once you start running traces.

Once we nail down at least the footprints for the component list I will start routing traces.

Not sure what y’all are finding wrong with the Vishay SIR800DP but it’s proven to be extremely hardy, Richard used a single one of these to run 12 XHP-50’s at 34,000 lumens. Just sayin.

I’m using one 17mm FET driver with Zener mod to run 4 of the 9V MT-G2’s at 15,000 lumens. Also running 3 9V MT-G2’s in a BTU Shocker at around 10,000 lumens. Same driver, again, for 3 XHP-70’s at 14,455 lumens in another BTU Shocker. Pretty sure this MOSFET works, tried and true.

The sir800 is great for a driver that switches the ground but you can’t use an n-channel mosfet to switch the positive side, that is where the P-channel comes into play.

The one I linked to above is the P-channel version of the sir800 because we know that one works good. It has really low resistance which is great but the capacitance is higher then some which lowers the max frequency that you can run the buck driver at. This is turn means you have to use larger parts.

It is a trade off both ways, just a matter of figuring out the best options.

Ok, I could not leave it alone, first run at running traces and not tweaked at all but the circuit itself should be good.

Lots of copper in the ground pour, I tried to get it between most traces to minimize interference.

Not yet, spare time is tightening up. But a bunch of those thoughts came from Ti docs, so I guess they make something, but don't know the specs. Given enough time, I'm sure I'll take an interest to look (and understand the high current trade offs). At the rate you're going it may be irrelevant, which is great.

http://intl-outdoor.com/ld4b-24a-17mm-buck-driver-3v16v-p-817.html

hard to see exactly what's going on there. It's "only" 2.4A. But it seems the inductor is lofted over some of the components, and they made use of the hole in the middle. That might even be a nice driver.