The Whispering Gallery - Modified Category - 7th Annual BLF/OL contest entry

29 posts / 0 new
Last post
The Whispering ...
Offline
Last seen: 13 hours 54 min ago
Joined: 06/25/2018 - 21:36
Posts: 40
Location: Florida
The Whispering Gallery - Modified Category - 7th Annual BLF/OL contest entry

I love AAA flashlights.

This is a Thrunite Ti3 V2.
Ti3 V2 parts

I like a lot of things about it except that it has next mode memory. Initially I thought that the next mode memory only lasted a minute or so, which wouldn’t be so bad. But sometimes it would last longer than that, and sometimes it would come on in low after just a few seconds. I wondered if the length of time that the next mode is remembered was dependent on the on time, as if the capacitor takes some time to charge up and then have a longer memory. (Full disclosure, I have no background in electronics).

I did a test where I turned the light on in low for varying lengths of time, and then waited varying lengths of time before turning it back on. In between tests, I waited at least 1 hour. This took a few weeks to complete.
Thrunite Ti3 V2 NMM_b

I still don’t know why the behavior of this light is unpredictable and would love some insight from the experts here.

Here is the driver. I pushed it out of the pill with a cotton swab after desoldering the LED MCPCB. The red arrow points to a piece I broke off of a component in the process (an inductor?). I don’t think it is a problem.

Thrunite Ti3 V2 driver

Then, I used a pencil to draw lines on capacitors to see if that would shorten the length of the NMM. The red arrow points to the one that worked. With the line, the NMM seemed to last pretty consistently ~10 s. Odd to observe predictable behavior after starting with something so unpredictable.

Thrunite Ti3 V2 driver with pencil mod

Testing with alligator clip wires.
image2 (5)

The next step will be to solder a resistor on that capacitor, testing to find the optimal resistance. I now have 1800 size 1608 metric resistors ranging from 10 Ohm to 1 megaohm. The first and hopefully not last giveaway in this thread will be my extra resistors.

SMD resistors

Edited by: The Whispering ... on 01/17/2020 - 05:44
The Whispering ...
Offline
Last seen: 13 hours 54 min ago
Joined: 06/25/2018 - 21:36
Posts: 40
Location: Florida

Reserved.

pinkpanda3310
pinkpanda3310's picture
Offline
Last seen: 10 hours 9 min ago
Joined: 12/28/2013 - 08:45
Posts: 2625
Location: Perth

Welcome to the comp whispering Smile

Next mode memory would have to be one of the worst ideas in the history of flashlights Sick So good choice of mod Thumbs Up

Agro
Agro's picture
Offline
Last seen: 5 hours 3 min ago
Joined: 05/14/2017 - 11:16
Posts: 6169
Location: Ślōnsk

Welcome to the party The Whispering. Smile
I hate NMM so much…I agree with pinkpanda, this is a nice choice.

CRX
CRX's picture
Offline
Last seen: 33 min 34 sec ago
Joined: 04/02/2013 - 15:27
Posts: 4194
Location: Scotland

The old pencil on capacitor trick, nice Wink
Good deduction, perseverance testing and showing results of the NMM times too Thumbs Up
Looks like a big bag of resistors from FT, always come in handy.

Scallywag
Scallywag's picture
Offline
Last seen: 51 min 28 sec ago
Joined: 01/11/2018 - 22:23
Posts: 1185
Location: Ohio, United States

Nice. NMM is a deal breaker.

Well… Confession time. My DQG Slim AA has NMM sometimes. It’s pretty unpredictable on mine. But I put up with it because it’s so otherwise nearly perfect.

EDC Rotation: ZL SC62(w) | Jaxman E2L XP-G2 5A | Purple S2+ XPL-HI U6-3A | D4 w/ Luxeon V | RRT-01 | Purple FW3A, 4000K SST20
EagTac D25C Ti | DQG Slim AA Ti | Jaxman E3 | UF-T1 by CRX | Nitecore EX11.2
L6 XHP70.2 P2 4000K FET+7135 | Jaxman M8 | MF02 | Jaxman Z1 CULNM1.TG | Blue S2+ w/ ML Special
Unfinished: Supfire M6 3xXHP50.2, Sofirn C8F, Sofirn SP70
Others: Nitecore EC23 | Nebo Twyst | Streamlight ProTac 1AA | TerraLux LightStar 100

The Whispering ...
Offline
Last seen: 13 hours 54 min ago
Joined: 06/25/2018 - 21:36
Posts: 40
Location: Florida

Thanks all for the kind words. I should clarify that I’m well aware that I didn’t invent the pencil trick, but I’ve never seen it done on a Ti3 V2 so I thought it was worth posting about.

Hopefully going to do a few other things to this light besides fixing the NMM…we will see how it goes.

The Whispering ...
Offline
Last seen: 13 hours 54 min ago
Joined: 06/25/2018 - 21:36
Posts: 40
Location: Florida

I have now tested several resistors soldered over the capacitor that I suspect is responsible for the NMM. By the end, I got better at the soldering. These things are really small.

Various resistors on Ti3 V2 capacitor
Resistor on capacitor

After trying a couple of resistors and testing using the test wires, I got NMM length of ~6 seconds with a 100 kohm resistor. Great! So I reassembled the light. But the NMM was lasting longer when assembled, ~17 seconds. I’m still not sure if this is a real difference, or how to explain it. At some point, I realized that I could assemble the driver/pill/MCPCB and that the wires were long enough to still access the driver, so I didn’t have to use the test wires like I showed in the first post.

Soldering holder_cropped

So I tried more resistors, trying to tune the NMM to a desirable length of ~5 seconds. Unfortunately, it seemed like I was getting diminishing returns, and the light when assembled never went below NMM of ~11 seconds. The NMMs with a resistor on the capacitor were generally very repeatable, and I tested each resistor several times. Here is a table of the results.

NMM length vs. resistor value

At this point, I’ve started working with my brother, who is an expert in electronics and such. He made some assumptions about the circuit, and then did some circuit simulations, but couldn’t replicate the results that I was getting. One of the oddest parts is the capacitor discharge time, which can be easily calculated. Even if the only parallel resistance is one of the resistors that I tested, and the capacitance is the highest possible for 0603 package (which I believe is 47 uF), the capacitor is generally discharged long before the NMMs that I observed. This is true if the capacitor is charged to 1.5V battery voltage, or by the boost converter to ~3V.

Maybe some of the assumptions about this circuit aren’t true though. So I bought my brother a new Thrunite Ti3 V2 from Amazon to look at (since he lives on the other side of the country). He found that the NMM only lasts ~10 seconds though, unlike my example. However, it goes into strobe seemingly randomly. The mode sequence is supposed to be L-M-H-L-M-H-S, but his light often goes into strobe straight from low or medium.

That’s all for now. We are going to try to learn more and figure out what can be done.

MRsDNF
MRsDNF's picture
Offline
Last seen: 5 hours 4 min ago
Joined: 12/22/2011 - 21:18
Posts: 13339
Location: A light beam away from the missus in the land of Aus.

Interesting mod The Whispering. You are certainly doing a lot of experimenting to get rid of an old Chinese favourite. Beer

 

djozz quotes, "it came with chinese lettering that is chinese to me".

                      "My man mousehole needs one too"

old4570 said "I'm not an expert , so don't suffer from any such technical restrictions".

Old-Lumens. Highly admired and cherished member of Budget Light Forum. 11.5.2011 - 20.12.16. RIP.

 

CNCman
CNCman's picture
Offline
Last seen: 1 day 13 hours ago
Joined: 04/07/2018 - 10:10
Posts: 591
Location: Mississippi Gulf Coast

Welcome “ The Whispering “ to the fun. Beer
You certainly are experienced at modding lights, nice explanations of your operations. Thumbs Up

CNC & Manual Machinist. Think outside the box too long , cannot find your way back in.

Never give up, Never surrender. Make someone Smile today.

The Whispering ...
Offline
Last seen: 13 hours 54 min ago
Joined: 06/25/2018 - 21:36
Posts: 40
Location: Florida

Thanks to you both. I wouldn’t say I’m all that experienced…especially compared to some of the amazing projects done by BLF members. I’ve enjoyed reading through the other threads in this contest, including both of yours. Just trying to learn and have fun with this Big Smile

NeutralFan
NeutralFan's picture
Offline
Last seen: 21 min 23 sec ago
Joined: 03/20/2014 - 19:22
Posts: 1031
Location: Wisconsin, USA

I hope you can figure this out The Whispering. So does it seem the pencil trick had the best results so far? My Astrolux A01 has similar behavior, but much longer off time. I give it a pass though due to it’s Nichia 219B.

I’d rather use my flashlight around the house than turn on the lights.

The Whispering ...
Offline
Last seen: 13 hours 54 min ago
Joined: 06/25/2018 - 21:36
Posts: 40
Location: Florida

Ok, the next thing I did was solder some wires on the capacitor so that I could measure the voltage across the capacitor when the light is on. I found that it varied from about 2.4-2.7V depending on the mode. The capacitor is charged by the boost converter, so low mode gives the lowest voltage and high mode give the highest voltage.

Testing voltage on capacitor_cropped

After some discussions with my brother, I made some rough assumptions that the minimum voltage to power the MCU is ~0.6V, and that it uses on the order of uA. At this point we don’t know the capacitance of this capacitor either, so I made the table below to show the expected length of NMM for various capacitors and parallel resistance (the resistor which I soldered onto the capacitor). 50 uF is about the max capacitance for a 1608 capacitor, and 1E6 kohm represents the unmodified flashlight where only a few uA are drawn by the MCU. I defined NMM length as the time for the capacitor to discharge from 2.4V to 0.6V. The equation is Vc=Vo*e^(-t/RC), where Vc is the capacitor voltage at time, t, Vo is the initial voltage, R is resistance, and C is the capacitance.

NMM table_3

The problem is that this doesn’t match any of the behavior that I observed, namely that I was getting NMM lengths of >10s even with the 1 kohm resistor. So I concluded that I’ve been working with the wrong capacitor, and started trying the others.

This one turned out to be the right one. I drew a pencil line first, and then cleaned it off and soldered on some wires to check the voltage, which was also about 2.4-2.7V. Then soldered on a 1 Mohm resistor (the highest that I have), and the NMM was ~3s.

1 Mohm resistor capacitor two_cropped

Based on the assumptions used to generate the NMM table, the capacitance of this capacitor is calculated to be ~2 uF.

I was happy with these results, so I decided to reassemble the light.

Agro
Agro's picture
Offline
Last seen: 5 hours 3 min ago
Joined: 05/14/2017 - 11:16
Posts: 6169
Location: Ślōnsk

Good to see it working. Your persistence is impressive. Smile

The Whispering ...
Offline
Last seen: 13 hours 54 min ago
Joined: 06/25/2018 - 21:36
Posts: 40
Location: Florida

Reassembly:

I first had to do something about the inductor that I damaged, so I put on a little bit of this MG Chemicals 832 HD potting compound with a toothpick.

Potting compound on inductor_before and after.jpeg

Then I used my reflow station to install a Nichia 219b 3500k R9080 LED, and reassembled the driver, pill, and MCPCB.

Reflow station and N219b 3500k

Added some commemorative laser marking for the 7th Old Lumens contest, and took a quick beam shot next to my Ti3 titanium with 219b 4500k R9080 (which has the new shorter and fatter V2 body and came with only 10s NMM and works flawlessly). Both LEDs look great in real life.

Reassembled Ti3 V2 plus beamshot comparison

Here is a close-up of the laser marking. Mine didn’t come out quite as sharp as that done by Thrunite.

BLF and Ti3 laser mark

Edit 2/28/20:
To put some data to it and check the consistency of the NMM length, I did 10 tests each of two different off times. Unlike the original light, which had seemingly random NMM times tested up to 10 min (shown in the first post), after the resistor mod I found the behavior to be repeatable.
NMM test after 1M resistor

I consider this light to be complete. I’m going to use it for a little while to make sure that there are no issues and then post a giveaway for it. My brother has also reverse engineered the circuit now and I will post that information soon.

The Whispering ...
Offline
Last seen: 13 hours 54 min ago
Joined: 06/25/2018 - 21:36
Posts: 40
Location: Florida

It’s been an honor to participate in this contest. Congratulations to all the other participants on their excellent builds and modifications.

pinkpanda3310
pinkpanda3310's picture
Offline
Last seen: 10 hours 9 min ago
Joined: 12/28/2013 - 08:45
Posts: 2625
Location: Perth

Congrats on finishing Party The laser treatment is a nice touch Thumbs Up

MtnDon
MtnDon's picture
Offline
Last seen: 28 min 51 sec ago
Joined: 08/27/2015 - 18:25
Posts: 3152
Location: Canuk in NM
Thumbs Up Beer
CNCman
CNCman's picture
Offline
Last seen: 1 day 13 hours ago
Joined: 04/07/2018 - 10:10
Posts: 591
Location: Mississippi Gulf Coast

Good presentation Thumbs Up
That really helps us to understand the detail operations going on inside a driver. Thumbs Up
Great Finish with the laser Wink

CNC & Manual Machinist. Think outside the box too long , cannot find your way back in.

Never give up, Never surrender. Make someone Smile today.

MRsDNF
MRsDNF's picture
Offline
Last seen: 5 hours 4 min ago
Joined: 12/22/2011 - 21:18
Posts: 13339
Location: A light beam away from the missus in the land of Aus.

Nice work The Whispering. Nothing like a custom light to suit your own needs. Beer

 

djozz quotes, "it came with chinese lettering that is chinese to me".

                      "My man mousehole needs one too"

old4570 said "I'm not an expert , so don't suffer from any such technical restrictions".

Old-Lumens. Highly admired and cherished member of Budget Light Forum. 11.5.2011 - 20.12.16. RIP.

 

MascaratumB
MascaratumB's picture
Offline
Last seen: 28 min 18 sec ago
Joined: 10/29/2016 - 12:12
Posts: 5217
Location: Portugal

Great work and persistence Wink
Surely a keeper after all this Beer

[REVIEWS] ACEBEAM: H20 / TK16 // AMUTORCH: S3 / S3 vs 219c / AM30 / AX1 / VG10 // BLITZWOLF: BW-ET1 // BRINYTE: T28 / PT18pro / PT28 // DQG: AA Slim Ti // FIREFLIES: ROT66 GEN II // HC-LIGHTS: SS AAA // KLARUS: XT1C // LIVARNOLUX: 314791 // LUMINTOP: Tool AA V2.0 + Tool 25 // NITEFOX: UT20 / ES10K / K3 // ODEPRO: KL52 / B108 // OLIGHT: M2R Warrior // ON THE ROAD: M1 / i3 / M3 Pro // ROVYVON: A2 + A5R / E300S / A8 // SKILHUNT: M150 / M200 // SOFIRN: SF14 + SP10A / SP32A / SP10B // WUBEN: TO10R / E05 / T70 / E10 / TO50R / E19 // XTAR: PB2 Charger // Tricks: 1 / 2 / 3/ 4 / 5 / 6 / 7 / 8 // TIR: 1 / 2 // Others: Biscotti 3 + 1*7135 / Triple TIR w/ XP-G2 /// My Collection /// My Review's Blog (PT) /// OL Contest 2019 /// GIVEAWAY: 1 / 2

NeutralFan
NeutralFan's picture
Offline
Last seen: 21 min 23 sec ago
Joined: 03/20/2014 - 19:22
Posts: 1031
Location: Wisconsin, USA

Wait, what, you have a laser that does that? Impressive The Whispering!

I’d rather use my flashlight around the house than turn on the lights.

The Whispering ...
Offline
Last seen: 13 hours 54 min ago
Joined: 06/25/2018 - 21:36
Posts: 40
Location: Florida

Thanks all. I’m really grateful to those that organized this contest and to BLF in general. I learned a lot doing this project, and that is by far the most valuable takeaway. Stay tuned as I’ll be giving away this light as well as a few kits from the leftover resistors to raise awareness for next year’s contest and encourage more projects and new participants (will take some time to sort through the resistors).

P.S. I don’t own a laser engraver, just had the opportunity to use one.

The Whispering ...
Offline
Last seen: 13 hours 54 min ago
Joined: 06/25/2018 - 21:36
Posts: 40
Location: Florida

Here is my bro’s report. This is all his work and I didn’t contribute anything. Note that his example of the light had consistent NMM of 10 s, unlike mine.

1. Reverse engineer the design and capture schematics.

Here is a description of the main circuit components/functions:

  • B1: AAA battery
  • D2: The LED
  • U2: Microcontroller (Probably ATTiny4 or PIC10F200)
  • U1, C1, L1, D1, C2, C3: Boost converter to drive forward voltage of LED
  • FB node of U1: feedback pin – constant voltage reference that provides control loop input for regulating current through D2. I_LED =
  • V_FB/R_Sense, where R_Sense is the parallel combination of all enabled sense resistors R1, R2, R3.
  • R3: LOW mode resistor (always on)
  • Q1A/R1, and Q1B/R2: FET which enables the corresponding sense resistor; enabled by MCU to change brightness
  • R4/R5: Pull-down resistor on gate of FET to prevent undefined state
  • C4: The decoupling capacitor for the MCU. This is what allows the MCU to stay alive during the NMM time. In general a decoupling capacitor provides a low-inductance voltage supply to satisfy the load’s (IC) transient current spike requirements.
  • D3: prevents discharge of decoupling capacitor through LED. This is what allows the MCU to stay alive during the NMM time.
  • Methods
    R1 and R2 values were read off the resistor and confirmed to be within range using multimeter. Since the values were so low, a 4-wire resistance measurement out of circuit would have been required, but was not done since the value was printed on the resistor.
    R3 was measured in-circuit
    C4 was measured in-circuit

Ti3 V2 driver schematic

2. Take key measurements

The circuit was instrumented for measurements. I soldered 18AWG stranded silicone wire to the board for low resistance connections, since we are dealing with very low sense resistors, and a very low current in LOW mode. The LED anode wire had to be lengthened to fit the current clamp probe. For the battery power input, I scratched off the soldermask on the back side and soldered the wires there because I didn’t want to get solder on the battery contacts (and I wanted to preserve the ENIG finish) in case I wanted to reassemble this flashlight. These wires were connected to a 1.5V DC bench top supply.

Test set up

Results:
Table of measured parameters

Discussion:

  • High mode measured current implies a combined parallel sense resistance of 0.13 ohms, or an R2 value of 0.15 ohms. It is very plausible that the extra 0.15-0.075 = 0.075 ohms comes from the on-state resistance of the pass transistor, Q1B.
  • Low mode current is very difficult to measure. 0.68mA is very close to expected.

3. Determine voltage threshold for moving to next state

  • Initially this was attempted by probing the voltage decay on the microcontroller power rail using an oscilloscope. However, the 10Mohm input impedance of the oscilloscope seemed to cause a fast decay of the power rail. After about 6s the capacitor has fully discharged.

Voltage decay_1

  • Next a bench top multimeter with a 10Gohm input impedance was used. Since this could not capture the waveform, I set a 10s timer (NMM time determined in previous experiments and stated in the user manual) and read the voltage at this time. The voltage was about 300mV. Image is for reference only, not the actual voltage that was seen at 10s.

Bench top multimeter

  • Now, taking into account the 1uF capacitor, a 300mV threshold, and a 10s NMM time, we can approximate the load presented by the microcontroller. It turns out to be about 4.5Mohm.

MCU load calculation

  • Next, we can use a resistor to model the microcontroller load in simulation. Our results confirm that ~300mV is reached at 10s.

Model MCU load

  • Now we can add a second resistor and adjust until we achieve the 300mV threshold at the desired NMM time. In these tests we achieve a 3s NMM time by adding a 1.8Mohm resistive load.

Model MCU load plus parallel resistor

4. Limitations to this study

  • The sense resistors (1.2 ohms and 0.075 ohms) are very low values. The resistance of the transistors (Rds_on and LED wiring will start to play a role here, explaining why the measured currents (especially in HIGH mode) were different than what you would calculate using ohms law (theoretical current calculations in table above).
  • The threshold voltage measured to be ~300mV was found empirically using a stopwatch with a human in the loop. Not incredibly scientific, but a good approximation. A better way would be to place a high-impedance op-amp across the capacitor and read the output of that with the oscilloscope.
  • D3 forward voltage is 0.24V. This was not taken into account in simulations.
  • 3.0V was used for V_BOOST, but really it varied from 2.5 to 3V, minus the forward voltage drop of D3.

5. Works Cited

Finally, here is a photo of the driver with components labelled.

Ti3 driver with labelled components

The Whispering ...
Offline
Last seen: 13 hours 54 min ago
Joined: 06/25/2018 - 21:36
Posts: 40
Location: Florida
The_OL_Contest
The_OL_Contest's picture
Offline
Last seen: 3 weeks 5 days ago
Joined: 11/04/2019 - 15:45
Posts: 42
Location: Around the World

Well done on completing your build Thumbs Up

z - The Whispering

The Whispering ...
Offline
Last seen: 13 hours 54 min ago
Joined: 06/25/2018 - 21:36
Posts: 40
Location: Florida

Love it!

jon_slider
jon_slider's picture
Offline
Last seen: 42 min 35 sec ago
Joined: 09/08/2015 - 12:20
Posts: 3230
Location: The Land of Enchantment

The Whispering Gallery wrote:

added text to your photo
(photo edited to match info in next post)

thank for the info

is the pencil graphite enough to eliminate the NMM on the Thrunite Ti3 AAA lights, without getting into soldering tiny smds?

The Whispering ...
Offline
Last seen: 13 hours 54 min ago
Joined: 06/25/2018 - 21:36
Posts: 40
Location: Florida

Hi Jon, I would bet that you would see a reduction in the NMM time with just pencil. However, the capacitor I first tested with the pencil trick turned out not to be the one that controls NMM. The correct one is in the photo below, with red arrow (resistor is soldered on in the photo).

1 Mohm resistor capacitor two_cropped