Help with S2+, MTN-17DDM + Bistro

[Cythras]

I have been wanting a Convoy S2+ Triple for a while, and finally decided to build it.

I ordered the MTN-17DDM driver from MtnElectronics, with Bistro, Thermal based stepdown and pre-soldered wires and spring bypass. Emitters used are XP-L HI V2 5As.

The once assembled, the light turned on, but behaved oddly, and I think I have narrowed it down to the driver.

The light will always step down to the previous brightness, all the way to moonlight, and then turns off. The step down occurs once every 5 seconds exactly, and will always step down through the 6 main modes, even if the light was set on muggle mode.

The only time it will stay on at a constant brightness is when it is in thermal calibration mode, where it will turbo until it gets too hot to hold (40s or so). This happenes even with thermal stepdown turned off.

On top of this, the light takes a very long time to ‘forget’ it’s last used mode when memory is turned off. It takes 30+ minutes for it to start on moonlight again, even if I disconnect the power. Also, I cannot do a long press to step backwards, or enter the special modes at all. I have factory reset the driver options several times, but that did not help.

I posted on reddit asking for advice, and was advised that the behavior was partially consistent with a low voltage issue, but I cannot seem to find the cause. The tailcap spring is bypassed with 22AWG silicon wire, and I have also tried using a tailcap from a known good S2+ (also running a mtn17ddm with bistro) and by using a brass shell casing to close the circuit instead of the switch.

The batteries used to test were unprotected NCR18650Bs, and also Nitecore 18350 IMRs. The light will not run at turbo for any length of time with a protected cell, but I am aware that it is caused by the protection circuit.

I reached out to Richard, and he replied with “I would double check to make sure that the OTC capacitor and any other parts weren’t damaged during installation. Also, check the R1 and R2 capacitors. I have seen them get cracked before from being too tight then being pressed into a pill… ”
He told me that all drivers get tested before being shipped out, and ones in international orders doubly so, and is confident that the driver worked prior to shipping.

However, I do not feel like I have done anything that could have damaged the driver. It was not press fit, I used a copper pill from kiriba_ru (thanks dude) with a retaining ring; I only had to carefully file down the ‘teeth’ on the board to get it to fit. Nor is the damage from overheating, as I did not need to solder anything to the driver, or the driver to anything as I requested that to be done for me by mtn. I have checked all the solder contact points on the components, and everything looks to be good, and I see no shorts caused by solder/metal shavings. I have since disassembled everything, and cleaned out all my contacts (retaining ring, gold ring on driver, flashlight ends) with light abrasives and alcohol. Upon reassembly, nothing has changed.

At this point, I am at a loss as for what to do. I took some pictures of the driver hoping someone with more experience will be able to spot anything that I missed.

Help me please, I want my light to work

Starting fires is fun

[gchart]

Apologies for the stupid questions, but you’ve made sure the battery is fully charged? Do you have another 18650 to try? I know you said you tried a couple already, just want to eliminate the easy stuff first

Outside of the memory issue, it sounds like low voltage protection is kicking in. The temp calibration mode operates outside of the LVP checks, so that could explain your observation there.

Side note that may be unrelated: NCR18650B’s are a little weak for a FET triple build, unless you’re intentionally looking to limit your light. Those cells are good for 5 amps or so.

[vwpieces]

My guess was the NCR18650B may be dropping in voltage too fast because of the load of a triple. Most will pull over 10 Amps.
Samsung 30Q are popular and I have many in triples and quads.

[Cythras]

@gchart

Apologies for the stupid questions, but you’ve made sure the battery is fully charged? Do you have another 18650 to try? I know you said you tried a couple already, just want to eliminate the easy stuff first

Outside of the memory issue, it sounds like low voltage protection is kicking in. The temp calibration mode operates outside of the LVP checks, so that could explain your observation there.

Side note that may be unrelated: NCR18650B’s are a little weak for a FET triple build, unless you’re intentionally looking to limit your light. Those cells are good for 5 amps or so.

No worries, I’ve worked as tier 1 tech support, so I understand

I’ve tried multiple batteries, all fully charged. I know that the 18650b are not the best battery for this, but they are all the decent 18650s I’ve got that are unprotected. The light just trips the circuits on my protected cells. However, I tested with high(er) drain 18350 IMRs as well, so I know that that is not the issue.

I probably would have purchased different batteries for a triple build if I were using Nichias, but the XP-L HI should be alright with the Panasonics.

[Cythras]

@vwpieces

My guess was the NCR18650B may be dropping in voltage too fast because of the load of a triple. Most will pull over 10 Amps.
Samsung 30Q are popular and I have many in triples and quads.

I would consider that a possibility were it not for the fact that
A) It turbos fine in temp config mode, and
B) It exhibits the same behaviour with a 18350 IMR.
C) It doesn’t just step down from turbo. If I turn the light on cold on moonlight, it steps down to off in 5 seconds. I’m sure a 18650B can handle that amount of current without sagging too much.

[gchart]

With all due respect, point A is invalid. Temp config mode operates outside of the low voltage protection. Part of the reason I jumped straight to LVP.

Can’t explain B though. I’d really try to get ahold of a 30Q or something just to be sure.

[Cythras]

@gchart

With all due respect, point A is invalid. Temp config mode operates outside of the low voltage protection. Part of the reason I jumped straight to LVP.

Can’t explain B though. I’d really try to get ahold of a 30Q or something just to be sure.

There’s a C now too, I edited it in
Fair enough for point A though.

[gchart]

Point C… hmm. Damaged voltage sense resistor? If it’s hardware related, I only know enough to be dangerous. I’m a software guy (CS major for my bach.)

[Rufusbduck]

Sometimes a poor ground can be mistaken by the mcu as a low battery condition. Current draw causes a voltage drop across the weak connection. I would disassemble the entire light except for solder joints including the switch, make sure there wasn’t any loose bits in the way, and reassemble. It might even be a bad switch, I had one with oddly high resistance that when changed sorted it out. Also make sure the retaining rings actually clamp the driver and switch pcb’s. On occasion threads have been cut not far enough and the ring bottomed out before clamping(pretty much just the A6 but doesn’t hurt to check everything you can). Good luck.

[gchart]

RBD, the fact that it turbos fine in temp cal mode would give me the (potentially incorrect) impression that the electrical path is at least passable… it’s just incorrectly sensing voltage. No?

[Cythras]

@Rufusbduck

Sometimes a poor ground can be mistaken by the mcu as a low battery condition. Current draw causes a voltage drop across the weak connection. I would disassemble the entire light except for solder joints including the switch, make sure there wasn’t any loose bits in the way, and reassemble. It might even be a bad switch, I had one with oddly high resistance that when changed sorted it out. Also make sure the retaining rings actually clamp the driver and switch pcb’s. On occasion threads have been cut not far enough and the ring bottomed out before clamping(pretty much just the A6 but doesn’t hurt to check everything you can). Good luck.

I have completely unassembled my light and cleaned all the contacts with light abrasives and alcohol, and both used a known good switch, as well as bypassing the switch entirely with a piece of pure brass. I have tightened all the rings as well, I am fairly certain that the issue is in the driver now.

[Rufusbduck]

@gchart

RBD, the fact that it turbos fine in temp cal mode would give me the (potentially incorrect) impression that the electrical path is at least passable… it’s just incorrectly sensing voltage. No?

You’re probably right but the mcu could detect a drop in voltage that the eye might not notice and since temp calibration doesn’t step down from a voltage drop it wouldn’t change modes then. I just make a habit of checking everything to be sure something wasn’t missed. I don’t generally make things up so if I suggest a possibility it’s something I’ve either seen myself or seen in a post. Eliminate all the easy stuff and then move on to the more difficult or time consuming.

[gchart]

For sure! You’re a veteran here and I’m sure have learned a lot of lessons the hard way. When you guys speak up, I try to learn as much as possible. Sometimes that’s just sitting back and absorbing what I read, sometimes it’s venturing out there to explore a bit. Not challenging anyone, just trying to learn.

[Rufusbduck]

I don’t get it right every time. Most of us are just trying to help and sharing what we’ve learned. Question everything, new ideas are how we get new toys. These DD drivers only have a few years under the belt and are still changing as developers learn more and come up with new firmware. I can’t really keep up with it all anyway but some things are common to all lights.

[Cythras]

@Rufusbduck

@gchart

RBD, the fact that it turbos fine in temp cal mode would give me the (potentially incorrect) impression that the electrical path is at least passable… it’s just incorrectly sensing voltage. No?

You’re probably right but the mcu could detect a drop in voltage that the eye might not notice and since temp calibration doesn’t step down from a voltage drop it wouldn’t change modes then. I just make a habit of checking everything to be sure something wasn’t missed. I don’t generally make things up so if I suggest a possibility it’s something I’ve either seen myself or seen in a post. Eliminate all the easy stuff and then move on to the more difficult or time consuming.

Would measuring the voltage at the tailcap potentially verify whether or not it’s actual low voltage due to high resistance/whatever or a faulty sensor?

[kiriba-ru]

I would say that regulated power supply is must have tool for all light builders. If you dont want to spent $50 or more, you can use any constant voltage supply (~12v ps) with cheap chinese dc-dc converter with current limiting.
In this case, using constant current supply could help you to make right answer within a minute - if there is a problem with driver, or some other issues.

[Cythras]

@kiriba-ru

I would say that regulated power supply is must have tool for all light builders. If you dont want to spent $50 or more, you can use any constant voltage supply (~12v ps) with cheap chinese dc-dc converter with current limiting.
In this case, using constant current supply could help you to make right answer within a minute - if there is a problem with driver, or some other issues.

Oh hey, it’s you. Thanks for making this light possible
But yeah, I need more/better tools and supplies; but a RPS is but a dream at the moment. First in line is a digital multimeter. (I’m not sure I even have an analogue one :person_facepalming: )

[Rufusbduck]

If you measure tail current in temp cal mode it should indicate whether your getting the full battery output. Might be a pain to do without the switch though. The driver pics are well focused but too small to see much, any chance for close ups of R1/R2? C1 and OTC looked ok.

[Cythras]

@Rufusbduck

If you measure tail current in temp cal mode it should indicate whether your getting the full battery output. Might be a pain to do without the switch though. The driver pics are well focused but too small to see much, any chance for close ups of R1/R2? C1 and OTC looked ok.

The light is back together again, and I swear my soldering gets worse every time I redo something; so I don’t think I’ll be pulling it apart again unless there’s a very good reason to. For the record, if what you are looking for is good solder joints, cracking, shorts or any other physical anomaly, I went over every component with a magnifying glass and it all looks good.

[djozz]

I have a same story here: Mtn 17FET+7135 driver with bistro, built into an Astrolux S41 quad shorty with 219C leds. The exact continuous step-down on all levels behaviour that you describe, and I’m sure that all connections were solid (I have learned that over the years). In the end I swapped the driver with a Banggood bistro driver (with added ‘pyramid fix’ ) and now it is good.

I have no idea why the driver does that, I have another of these Mtn drivers that I’m not tempted to use now…

My opinion: the development of the bistro driver here on BLF and by Richard appears an interesting ongoing project, this driver especially showed several unexpected side-effects (that I do not understand, but I sometimes try to follow what they discuss in the Attiny development thread) , and there may still be new insights and fixes (hardware and software) before the driver is stable in all sorts of circumstances. But it nevertheless is my favourite driver!