[ GXB172 - 50W Single Cell 17mm Boost Driver! ]

I started out trying to do tail cap measurements with three or four different DMM’s and every measurement would be radically different for each meter. The higher the amps the worse they measured. I quickly gave up on using DMM’s.

I do get repeatable results with my UT210E, but I don’t have a high precision, and I assume very expensive, calibrated meter to verify it’s accuracy.

The typical things I measure are flashlights at the tail cap and they can vary due to the battery, driver, led, etc… You get the usual peak when you first turn it on and it typically drops a little as the seconds go by so I don’t need super high precision.

Plus a lot of folks on BLF use the exact same meter. So if it’s got a wierd curve to its measurements or an inaccuracy, then hopefully everyone else measures about the same.

Repeatability is important to me. If a light measures 15.1A@30s and I switch to a better driver spring and get 17.6A@30s then I know it’s probably a legit 2.5A increase. Even if the meters calibration is 4% too high for instance, it’s still showing a legit 2.5A increase.

Us flashlight hobbyists are not working in a laboratory, so we don’t need super high precision. We just need something reasonably accurate for our purposes.

For it’s reletively low cost and seemingly “accurate enough” measurements, I think the UT210E is a good deal. Just my opinion.

Suggestion: use a low value precision current sense resistor with 4 wire measurement since most multimeter are very good at voltage/milivolt range.
You can buy this slightly expensive but good FPR2A-0R005F1 no-inductance resistor, 5 milliohm. At 20A, this will only have 100mV drop. Not the smallest in flashlight where every little bit is important, but easy to measure with 0.1mV resolution with most cheap multimeter. Then you can just put this across tail of flashlight and do quick measurement.

Are you suggesting we do this once in order to determine how accurate our clamp meter is or are you suggesting we do this every single time we want to do a tail cap measurement?

The uni-t clamp meter look like ±(2%+3) specification so it should be ok. But at low current measurement or if you want to make lower cost and very accurate measurement (but with some burden voltage), it seem like using a good current sense resistor and multimeter is easiest way, so it can be a backup or different method.

Yep.

Using a current shunt is how I personally roll.

I’ve now also built one, using slightly different components though.


I am having some issues though, I think I might have fried the OpAmp. I’ll have to try replacing it. The mess on the bottom right is some fooling around with the current sense resistor, and I got a bit sloppy.

Wow very exciting work! More people building the GXB172, but this looks like your design! Do you use the same MP3431 ic? I see different component on your board, like maybe bigger resistor and different regulator and attiny?

Nice Design Mike C. Looks like your using 0603 components. Is this PCB still 17mm.
From the pics I count A few less components, but I might have missed something the pics don’t show.
I have built several of these drivers now with problems, my fault of coarse. The design is good it’s the small components that makes bad connections or shorts so hard to spot. What kind of problem are you seeing?

The most recent one, I didn’t have much difference in modes. That’s the one that was reporting 19 amps on the 87V setup for the XHP35.2. The modes were there, the output didn’t change much from mode to mode when switched. After a few hours of trying to figuring out what was going on, I tested R11 which was reporting a very low resistance. Unsoldered the resistor and it read fine off the PCB. Checked the pads and saw a very low resistance. I chased it back to the mp3431 and saw that pin 5 and 6 (FB and AGND) looked like the were touching. Applied some Amtech NC-559 flux and hit it with a fine tip soldering iron. The solder jumped right apart and flowed to each pin, amazing flux.
Now the modes work fine.

I did get the UT210E clamp meter yesterday. This is after I fixed the short at pin 5 and 6. The UT reports 16 amps and 2600 lumens at start up running the XHP35.2 and the 87V reports close to the same. I actually ran the 87V homemade thick leads threw the UT to get the reading. Before I found the short it reported 19 amps and 2800 lumens at start up. I can’t get it to run a 19 amps again, so I’m guessing the short had something to do with that.

Yes, my board is my own design and it is 17mm. However, the boost circuit is entirely based on loneoceans schematics. I use the same boost IC, op-amp and component values. He hasn’t made the schematics public but was kind enough to send me them on request although he has not finalized them.

I made my own because I want to use my 1634 firmware, I don’t want to go back to the 841. And I really hate working with 0402 so I attempted to make it with 0603 instead. The main differences with the layout is that my 1206 caps are on the battery side and the MCU is on inductor side. I have a few less components because I don’t use an external temp sensor, E-switch pull ups, reset pin pull up and OTC cap. I did forget to add an input cap to the op-amp though, so I might be getting noise. I can add one from that pin to GND ring though, the op-amp is orientated that way on my board.

The issue I’m having as that the output as varying up and down although I’m increasing PWM duty cycle. I have to hook this driver up to an oscilloscope and really check everything before I start replacing components. It will force me to learn a few things about this boost circuit design. I’ll also have to check everything with a magnifying glass and maybe use some flux here and there.

The thing with making my own board is that I can’t really say if any issues I have is due to layout, because layout is important. I’ve tried to follow MP3431 layout guidelines, and tried to stick to loneoceans layout, but by using 0603 and larger MCU I have had to make some compromises.

Good luck with the problem solving Mike.

Can I ask you EE guys a couple of off topic MOSFET questions real quick? You all seem to know your stuff and I’m just barely learning.

1. Are the majority of MOSFETs we use in flashlights N channel?

2. All N channel normally resist flow between the source and drain, correct?

3. Then your P channel always allows flow, correct? I think this is the big difference in them.

4. Providing a positive signal to the gate of an N channel allows flow and on a P channel it stops flow, correct?

5. Lastly, are the gate signals always a positive voltage or can some MOSFETs be triggered with a ground signal?

Thanks in advance for any help.

1. Yes
2. No, you can sub-divide the n- and p-channel MOSFETs into “enhancement-mode” (normally-off) and “depletion-mode” (normally-on) categories (these 4 types have their own schematic symbols). Enhancement-n-channel resists flow (hence the name normally-off), and is the most-often used type of MOSFET. And usually when people talk about MOSFETs, they usually mean a normally-off MOSFET, if they don’t specifically say it’s a normally-on type (since they’re not that often used)
3. That’s what depletion-mode MOSFETs do. A p-channel-enh. MOSFET (which you are reffering to) normally stops current flow the same as a n-channel-enh. Just the “activation method” to let current flow is the other way around. Instead of positive voltage compared to Source pin (positive V_GS) (that’s what a n channel needs), for a p-channel you need a negative voltage compared to Source pin (negative V_GS) to let curent flow.
4. See answer no. 3
5. I’d like to clarify some things here: MOSFETs generally don’t know what a positive or negative voltage is. They have no reference (ground) to be able to tell positive and negative apart. This is just us, that defined a specific voltage potential as “ground”. That way it’s easier for us, and the calculations. What MOSFETs however do know, is “more positive/less negative”, and “less positive/more negative” (the words separated with the Slash are equivalent). It compares the the Gate potential to the Source potential, and acts accordingly.
So to answer the question: MOSFETs don’t care whether the Gate signal is positive or negative, it depends on the combination at the Gate and Source. So yes, all MOSFETs can be triggered with a ground signal.

To easily explain depletion Mosfets: Think that they are like enhancement Mosfets, but a bias is already at the Gate pin, that turns them already on. So you need to go in the other direction (in terms of voltage) to overcome the bias to turn them off again.

And just to avoid some misunderstanding, in short:
(V_GS is voltage between Gate and Source. “Positive” means Gate more positive than Source, “negative” means Gate more negative than Source”)

n-channel-enhancement Mosfet:
normally-off, positive V_GS to turn on

p-channel-enhancement Mosfet:
normally-off, negative V_GS to turn on

n-channel-depletion Mosfet:
normally-on, negative V_GS to turn off

p-channel-depletion Mosfet:
normally-on, positive V_GS to turn off

Okay, I was hoping it would be a bit simpler, but I guess I need to keep studying. :smiley:
Thanks

This might help a little. https://www.quora.com/When-is-it-good-to-use-N-MOSFET-and-when-do-you-use-P-MOSFET

I wonder….
I see more and more enquires about high voltage drivers. Hyperboost is the only solution out there.
I see that the IC here is rated to 90V output.
What would it take to turn GXB172 to a high voltage driver?

It is limited due to the IC it uses. It has a max voltage of ~16V IIRC.

So an entirely different IC would be needed and that basically means a whole new driver.

Ouch, I see that I mixed up different products. Yes, 16V limit. That kills the idea. Thank you for setting me straight!

Wait…

Were you planning to use a COB?

As of now I don’t have any such plans. Now that you’re saying it - a flooder with 3000 lm CRI97 COB would be interesting. :wink: Though definitely out of my skills range.
I think I’ve seen at least 4 inquires about it in the past quarter - and that’s it.

Glad to see others have gotten this working, too! :slight_smile:

Thus far, I’ve built a total of 4 of these—two of the original Rev A and two more of the Rev A2. Sadly, I managed to fry one of the original Rev A boards, so I only have 3 operational right now. I am rather happy with them, but still would like to hack on the firmware a bit.

When I get a moment, I’ll try to get some more pictures of the new builds, along with some measurements. However, I wanted to check in here and ask a question, as my fifth build is currently having some issues: after I power it up, it blinks the LED (at lowish power) 4 times before turning off. My guess is that communication to the temperature sensor isn’t working? But wondering if anyone else ran into the problem / how to solve it.

I’ve got some of the 20mm boards (finally!), and will be doing some builds of those with some new S11 hosts from MTN.

(And yeah, I have had trouble getting my hands on the 47uF 25V recommended caps; 10V and 16V are fine for 6V and 9V operation, but 12V is too little margin. Managed to get a few from Mouser but they are since out of stock.)