17mm/22mm MTN-MAXlp - Low-Profile 1A-4A Programmable Buck Driver + 22mm MTN-MAX

The timing, even with the correct resistor, is a bit tricky. You can't just tap the switch as fast as possible like on most drivers. There is sort of a rhythm to it. It usually takes me a few tries before I get it.

Yeah I keep trying but… I don’t get it, I know it’s there but I can’t catch it. :–5

Did you ever get it going?

Hopefully if I get some time later this week I will post up some new variants that I've been working on for a while, that are finally ready to go:

  • Updated 17mm MTN-MAXlp HP
    • Increased efficiency
    • In-system-programmable
  • Updated 22mm MTN-MAX
    • Increased efficiency and stability; fixed a few small bugs
  • New 26mm & 27.5mm MTN-MAX

No I've done all I know to do but can't get into programming. The timing just isn't quick enough or something. I have been meaning to ask you if I should stack a 36k on there and try it.

I have been burning thru some primarys with it though. It's a 3.5 amp so in high it steps down in like 3 sec (with fresh batteries) but it's still more than 500lm and lasts a long time so I'm happy with that. That's not the way I intended to run it all the time but I just wanted to know how it would handle them. It gets really warm for the first 20 min or so but after that it's fine.

I would be worried a bit about pushing those primaries that hard. Rule of thumb is 1.5A maximum continuous, so you're pushing them a bit harder than that probably. Be careful!

When you switch modes, do you feel like you have to tap the switch really fast, or do you have to wait a moment before it will switch?

I have to wait a moment to get it to switch.

Yeah I was worried about it too and have been paying close attention to the batteries. They're not really getting hot. It's the driver heating. I was thinking that when it steps down (almost immediately) it wouldn't be pulling the max from the batteries?

If you have to wait, then stack another resistor on top of the existing and try it.

I now have 2 19k resistors stacked. 9550 ohm? The mode changes are quicker and I am able to enter configuration mode and select the mode group. Seems to work like it should now. Thanks Richard. I do believe I’ll be getting a couple more of these soon.

Glad you got it working!

I usually haven't had that much of a hard time with guppydrv ones, but it does take a bit of timing to get it.

OK, finally got time to get the new 4L 17mm and 2L 22mm uploaded and shared. Should I keep everything in this thread and just update the OP, or should I start something newer and tidier?

17mm 4L v1.1 w/ ISP Programming (of course, more details to come)

22mm 2L v1.12

You will notice that besides the ISP on the 4L, there is a new part on the board. I also need to update some of my component recommendations, especially those regarding inductors.

In the end these drivers are still pulse width modulate for lower modes right? As in the LED sees on/off pulses right?

That is correct. You can't see it or hear it, but it is PWM.

Thanks. I asked because of this:

https://budgetlightforum.com/t/-/41762#comment-1004493

An idea to use a Ti IC (that I think you recommended) for analog buck control. Better discuss there I guess to not split conversation.

Is there any way to make the high-power version run above 6A?
I would prefer 7A for a big light with XP-L which I have.

I haven't tried it...it may or may not work. I think that 6A is already pushing it pretty hard, but 7A may be possible. Remember, that 7A will be the average current and that the peak current seen by the LED will be a bit higher than that, so keep that in mind (this is all assuming that the driver maintains stability at 7A, which it may not).

This 10% ripple thing is something I'm pretty curious about. I don't know about your circuit but the application diagram for the MAX16820 shows no output cap, ie no cap in parallel with the load. I get that the current sense/switch-control is histeretic, and that means just like a constant off time setup, it needs to sense real time fluctuations in the inductor current (difference is constant off time only triggers on the up pulse not up and down). But an output cap shouldn't have a big effect on the inductor ripple current, and that sense resistor, while on the other side of the load in that diagram, should still effectively see the inductor ripple I think, not the load ripple if there's a cap. (The cap probably has some effect on inductor ripple as the usual simple inductor ripple current equations assume a constant output voltage, but the fact we're driving diodes still keeps a reasonably close approximation to that anyway.) It's certainly possible to get big inductor ripple current (which I suspect the histeretic circuit needs) and a low output ripple current at the same time anyway, in worst case might require a bit smaller inductor value or a bit lower switching frequency.

Not trying to redesign the driver here though. What I'm wondering though is what happens if one were to wire a 10 or 20uF cap across the led's without modifying anything else? Does it break your buck? Might be worth a try for someone who really just has to squeeze the last drop out of their diode, but they still better have spare diodes to kill I guess.

If you read the entire datasheet you'll see that they do indeed talk about an output capacitor to reduce output ripple, and I have implemented one into the design. That said, you severely limit your moonlight capability if you go much bigger than 1uF, and if you go way too big you can compromise stability. I can measure a reduction in output ripple up to 10uF, but more than that doesn't seem to help (at least not with the non-ideal capacitor size and PCB location).

But remember that 10% ripple? Yeah, that's if everything was running perfect, with an ideal PCB layout and ideal components; in the real world as you start to push things sometimes that ideal 10% grows far beyond 10%, so I always say 10% with the capacitor to give some safety margin. This PCB layout is far from ideal, but here we have to work within the constraints of a 17mm PCB with pre-defined ground and battery input points, limited heatsinking---a far cry from the demo circuit.

Ok, thanks. Yeah, I'm definitely learning these bucks are complicated, and indeed when pushed hard one approximation breaks down and the rest all become nonsense too and the simple hopes of any hand calculations may be moot.

I guess this stability issue at high capacitance may be related to the histereitc feedback, or maybe I just hope. The numbers I've crunched for the texas buck look optimistic that we'll be able to get low ripple, but I'm also well aware that there are tons of demons hiding in the corners of these bucks, so we'll just have to see. Really while that IC is probably nicer, there are things I still don't like about it too, oh and predictions now are based on components we could fit on a 49mm board, not a 17mm or whatever board.

Oh that moonlight thing is a really good point too. We're thinking on the LM3409 of ramping down to low current and then adding PWM to go super low from there. It might be easier because the PWM won't have to take it as far, but still better think it through, or maybe just find out. Thanks again for the detailed reply.