V4 of my B17 17mm boost driver, success finally!

My firmware has a configurable max amp setting. It can be programmed when flashing the firmware or it can be entered by config menus with a bunch of clicking. Unfortunately the current accuracy is a little less with the HDR FET, it changes it’s resistance a little, but I think it will be good enough. If very low(moon) modes are not wanted, then the HDR FET doesn’t have to be mounted, and the high power sense resistor can be placed in the low/moon sense resistor’s position. Current control by firmware will be more accurate then. There is a possibility to limit it by resistors, but I don’t think that’s very flexible if the LED should be replaced with something newer later on.

I’ll build another TPS61288 based version and fully test it. If it has flickering like thefreeman reported, then I’ll for sure stick with the MP3431 as it does not have this issue. I do have a bit more functionality testing on the MP3431 too, but if that all checks out it’s ready to go.

:smiley:

Whenever i see these drivers all i can say is amazing and how much(?)

Oh sweet. That will be fantastic. When you say less accurate, are we talking more than +/- 0.1A? I love my moonlight modes.

Do you expect any clearance issues with retaining rings? I know some of the double-sided 7135 drivers are tough in a S2+

I’m thinking of a Convoy M2 with a SFT-70 and another with a sliced Nichia 144 R9050

Why not Nichia NV4B35AM? I would think it’s better than 144A and would be glad to learn why it’s not. :slight_smile:

When I list the BOM I will at least be able to provide material cost.

Possibly a bit more, I have to do more testing. I also have an idea on how to eliminate the issue while having the FET but so far it’s just an idea. I have to do some research, it could be far fetched.

I just checked with a S2 host, I haven’t opened one of those in ages. The retaining ring would have to be filed out. But I noticed that the pill depth has no chance of fitting the XAL7070 inductor, darn! A lower height one would have to be used, like the XAL7030. I couldn’t find any S2+ host in my drawer although I’m sure I have one or two somewhere.

I checked with a S3, the retaining ring on those is no issue at all, plenty of clearance, but same issue with driver depth. I haven’t done any testing at all with the XAL7030 but that’s something I will have to do as I see some of my target lights won’t fit the XAL7070.

No proof yet, but I am assuming the smaller package is going to be thermally limited. I also fear they’ve used newer silicone phosphor binders that burn at low temps like the e21a and 219F etc. I really hope I’m wrong and that they can be driven to 40W+ like the 144. For anything but a max power build, I’d prefer b35a for all the reasons I chose against it.

When I designed my driver I thought of using a mux for Vsense :

That way the mosfet is taken out of the total Rsense and doesn’t interfere with it, the gate driver could even be removed.

There are small ones in SC70 package, though I haven’t looked into it much, if possible smaller would be better. I don’t know if they can interfere with the very small Voltages.
It doesn’t need a separate control signal because it can be the same as the one for Q1.

The disadvantage is that it adds one more component (obviously), also now Q1 resistance is added to the circuit and thus incurs additional conduction losses, so maybe keeping the gate driver for a lower resistance would be better.

Yeah height clearance is a problem with many hosts.

The solution I used seems to work well, even though it reduces efficiency at very low levels it’ll still be more efficient than USM at low level. But if you want that crazy output then yeah the MP3431 is a better choice.

The mux idea is what I was thinking of. I’ve never used one before but I’ve just made a V5 design and managed to squeeze one in. The one I went for is the Texas Instruments SN74LVC1G3157 because it comes in a variety of packages, two of them very small: SON (DSF) 1.00mm × 1.00mm and X2SON (DTB) 0.80mm × 1.00mm.

How it works for these low voltages I don’t know, the datasheet list of features includes “rail-to-rail signal handling” so I assume it shouldn’t be an issue. I don’t know about noise though. I haven’t found anything in the datasheet that suggests this one is a bad choice but I’m all ears. Otherwise I guess we’ll find out.

Yeah this is worth testing, especially for high current (9A lol) where the sense resistor has to be very low for a low Vsense.

Nice work Mike. It’s great to see this boost topology getting perfected by several people at once! Next you guys can tackle linear and buck topologies. :slight_smile: But seriously, from what I am reading across the threads, it sounds like some of the techniques being employed will translate to the other topologies as well.

For buck or buck-boost it’s just copy-paste, linear I don’t see much point as it’s possible to do relatively small buck converters when using ICs with high frequency switching.

I think the leakage current might be a problem, it’s 1uA for this one and because presision op-amps have extremely low input current bias (70pA if you’re using a TLV/OPA333), it will raises the voltage at the op-amp input.

Paragraph about leakage current

So if I understand correctly we would need one that has less than 70pA leakage current, TI offering only go down to 400pA, Maxim down to 10pA but no SPDT, Analog has some in SC-70 and CSP with 10pA.

Edit : BLF is cutting |70p at the end of the link.

Thanks. I have a 17mm buck boost driver with FET for full blast but I’ve just been focusing on high power boost for now. I’ve also designed a 17mm linear driver. Not sure it’s all that useful compared to buck boost, but it’s fun playing around with this stuff.

I see. Thanks for this information! I’ll have to see what packages I can fit, and choose lowest leakage current based on that.

What Rdson do you get with the gate driver ? I would have thought it would be quite low and not affect the total Rsense too much.

I haven’t done the measurements but rather some practical tests. When changing the voltage on my PSU I can notice small differences in LED output, at least when using a 0.005 Rsense resistor. So obviously something is having effect on it, probably charge pump output voltage.

I switched from 0.010 to 0.005 when I wanted to run it at full blast. It’s not a big deal I guess, but if I can fix it I will, at least for learning purposes. I have a little wiggle room for moving components so I’ll see what I can do, I might be able to shuffle around enough to make room for SOT23-6. I refuse to use 0402 though, so I’ll just have to see.

I have to admit, I don’t know much about what characteristics to look for other than what you have suggested. So may I ask, would the ADG819 be a good choice? I can definitely fit the 6-ball 1.14 mm × 2.18 mm WLCSP package: ADG819 Datasheet and Product Info | Analog Devices

Hmm, strange, the datasheet for mic5019 says 8V at 2.7V, the Rdson shouldn’t change much above that, some measurments would be interesting.
If it does varies, one option would be a zener to clamp the gate voltage at 8V, that would be smaller than a SC-70 mux.

I also thought about using a regulated 5V charge pump to power everything, so in my anduril compatible design, that would be in place of the LDO.
But in your design that would remove the need for the gate driver and this mux. There are small 2x2 QFN ones (and maybe smaller ones), I’m not too sure about the inpact of the noise on the op amp from the switching of the charge pump. Also even at 5V there might still be Rdso variance between parts.

Output difference is visible so what ever it is I was thinking that the mux should take care of everything as current sensing will only be over the sense resistor. I’ll give it a go.