phosphor22
Enlightened
LOoking forward to hearing about Phase 2 testing... and the lowest mode on this sounds just right to me.
Build Log: The 1.5V Project
- Thread starter calipsoii
- Start date
Help Support Candle Power Flashlight Forum
calipsoii
Flashlight Enthusiast
- Joined
- Apr 21, 2010
- Messages
- 1,412
Putting It All Together
Despite being done the machine work months ago, I only just now (grudgingly) assembled the light. The reason is pretty simple: once the driver is sealed up inside the host it's very difficult to change the firmware. I've spent many nights flashing and reflashing code changes at my computer desk trying to get things just perfect but there's no substitute for real-world use so it's time to say "good enough" and take this thing out on the road.
-----------
We start with the MCPCB. Most of you already know what it is, but the Metal Core Printed Circuit Board is what you mount the LED on. Sinkpad's are extremely expensive, solid copper MCPCB's that directly connect the LED thermal pad to the flashlight body. They're measurably the best MCPCB's on the market in terms of wicking away excess heat from the emitter into the body. They're also really thick and heavy - my little 10-pack weighs more than most of my lights.
These are pretty much impossible to solder with an iron so I applied a dab of solder paste to each pad, mounted the LED, and reflowed it on my skillet. For this prototype unit I've chosen a N219B HCRI LED.
After physically examining and poking the LED to make sure it was completely soldered, I tested it for functionality. If you set your multimeter to 'continuity test' it will ramp up the voltage until a small current flows - the perfect way to test an LED without blowing it away.
Next up: getting the driver mounted inside the driver pill. Here I've soldered some power wires onto my little boost converter (two to the battery, two to the LED).
The fit is perfect! The board is suspended neatly inside the driver pill, resting on a tiny ledge which contacts the ground ring on the driver and serves as both a ground and heatsinking path. The white Delrin collar is the one you saw earlier - it acts as an insulating spacer which will a) keep the battery contact PCB the correct distance from the converter and b) give a bit of room for the wires to fit.
Here I'm trimming the driver wires to the correct length. This is seriously hopeful wishing on my part: my goal is to be able to flip the contact board completely out of the way, allowing access to the microcontroller for reprogramming without completely disassembling the light. It might work, it might not, but I'm going to try. If it does, it'll mean flashing a new firmware is a 10-minute ordeal with no soldering and only a pair of tweezers needed for disassembly.
Hopefully all that extra wire will fit in the driver cavity! The solder points get some Kapton tape to prevent shorts.
These springs were marketed as 'carobronze' which is apparently a Phosphorus Bronze alloy. They have a higher tension than my old ones and seem to be very low resistance.
With all the soldering done it's time to try fitting everything into the driver pill. Fingers crossed - this is where we find out if my estimated dimensions match up to real-life.
Looks good.
A touch of Arctic Silver thermal paste and then we're mounting the LED up on top of the pill.
The Kapton tape is quite important up here because the metallicized reflector will conduct electricity and could cause a short.
A quick test-fire to make sure everything is working before we proceed.
These rubber switch boots are intended for use with short e-switches. To modify them for a McClicky the central pillar needs trimmed off.
Most McClicky's screw down into the body. Screwing them up into a tailcap (like in a Surefire 6P) is fine so long as the flashlight body contacts the silver ring on the bottom of the switch. In our case neither scenario is happening, so we need to add an additional component: a threaded retaining ring that grounds the switch against the body of the tailcap.
And with that, we're all done! As an added bonus, with our switch boot on one end and an o-ring on the other, this tailcap is almost certainly waterproof.
A tiny Delrin centering ring keeps the reflector centered on the LED. If you look down on a reflector and see the yellow phosphor of the LED you're on the path to a nicely focused beam.
Looking like the rocket from a Saturn V, the reflector sits atop the driver/emitter stack.
The reflector is slightly larger (18mm) than the glass window (17mm). I was having issues getting the o-ring and glass centered during assembly, so it was back out to the lathe for ANOTHER Delrin centering ring (you can see it sitting at the bottom of the head). The part count keeps risin' but if you want great results you have to work for them!
Now THAT'S a centered o-ring. The slight blue tinge is the anti-reflective coating of the ultra-clear lens (UCL).
The whole assembly is threaded down into the head...
... and we're finished!
Ah heck, I can't keep you in suspense. Shall we sneak a peek of the Spyglass v2 compared to the previous version?
In addition to the driver being completely redesigned from the ground up, we've changed a few external things as well:
This was my first foray into the wild world of waterjet cutting. A local shop in town offered me a small run of prototype pieces, so I brought in my drawings and they went to work. This was the result: 10 pieces of 0.060" stainless steel.
The edges were a lot rougher than I expected. This is probably because most clips are stamped with a die, not sliced with a waterjet.
I wouldn't want to do any great number of them, but for a single prototype I was willing to go at the edges with a file.
There is a nice smooth edge hiding under there, just need some elbow grease to bring it out.
The surface finish could use some love as well. I started with some wet 600 grit and moved to 0000 steel wool. I brushed the steel-wool lengthwise to give a uniform appearance.
A little elbow grease later and we're left with a much nicer looking clip.
Adding the bends to this clip was MUCH easier than bending the old wire clip of the v1 light.
So? What does it look like?
I spent a good portion of the day comparing this clip to those on other lights I own.
If I had to rate the holding power and tension I'd probably say:
I'll finish this post with a couple glamour-shotsdevil. Next update will cover the new driver and functionality so check back later will ya?
>> Return to root post
>> Continue to next post in the series
Despite being done the machine work months ago, I only just now (grudgingly) assembled the light. The reason is pretty simple: once the driver is sealed up inside the host it's very difficult to change the firmware. I've spent many nights flashing and reflashing code changes at my computer desk trying to get things just perfect but there's no substitute for real-world use so it's time to say "good enough" and take this thing out on the road.
-----------
We start with the MCPCB. Most of you already know what it is, but the Metal Core Printed Circuit Board is what you mount the LED on. Sinkpad's are extremely expensive, solid copper MCPCB's that directly connect the LED thermal pad to the flashlight body. They're measurably the best MCPCB's on the market in terms of wicking away excess heat from the emitter into the body. They're also really thick and heavy - my little 10-pack weighs more than most of my lights.
These are pretty much impossible to solder with an iron so I applied a dab of solder paste to each pad, mounted the LED, and reflowed it on my skillet. For this prototype unit I've chosen a N219B HCRI LED.
After physically examining and poking the LED to make sure it was completely soldered, I tested it for functionality. If you set your multimeter to 'continuity test' it will ramp up the voltage until a small current flows - the perfect way to test an LED without blowing it away.
Next up: getting the driver mounted inside the driver pill. Here I've soldered some power wires onto my little boost converter (two to the battery, two to the LED).
The fit is perfect! The board is suspended neatly inside the driver pill, resting on a tiny ledge which contacts the ground ring on the driver and serves as both a ground and heatsinking path. The white Delrin collar is the one you saw earlier - it acts as an insulating spacer which will a) keep the battery contact PCB the correct distance from the converter and b) give a bit of room for the wires to fit.
Here I'm trimming the driver wires to the correct length. This is seriously hopeful wishing on my part: my goal is to be able to flip the contact board completely out of the way, allowing access to the microcontroller for reprogramming without completely disassembling the light. It might work, it might not, but I'm going to try. If it does, it'll mean flashing a new firmware is a 10-minute ordeal with no soldering and only a pair of tweezers needed for disassembly.
Hopefully all that extra wire will fit in the driver cavity! The solder points get some Kapton tape to prevent shorts.
These springs were marketed as 'carobronze' which is apparently a Phosphorus Bronze alloy. They have a higher tension than my old ones and seem to be very low resistance.
With all the soldering done it's time to try fitting everything into the driver pill. Fingers crossed - this is where we find out if my estimated dimensions match up to real-life.
Looks good.
A touch of Arctic Silver thermal paste and then we're mounting the LED up on top of the pill.
The Kapton tape is quite important up here because the metallicized reflector will conduct electricity and could cause a short.
A quick test-fire to make sure everything is working before we proceed.
These rubber switch boots are intended for use with short e-switches. To modify them for a McClicky the central pillar needs trimmed off.
Most McClicky's screw down into the body. Screwing them up into a tailcap (like in a Surefire 6P) is fine so long as the flashlight body contacts the silver ring on the bottom of the switch. In our case neither scenario is happening, so we need to add an additional component: a threaded retaining ring that grounds the switch against the body of the tailcap.
And with that, we're all done! As an added bonus, with our switch boot on one end and an o-ring on the other, this tailcap is almost certainly waterproof.
A tiny Delrin centering ring keeps the reflector centered on the LED. If you look down on a reflector and see the yellow phosphor of the LED you're on the path to a nicely focused beam.
Looking like the rocket from a Saturn V, the reflector sits atop the driver/emitter stack.
The reflector is slightly larger (18mm) than the glass window (17mm). I was having issues getting the o-ring and glass centered during assembly, so it was back out to the lathe for ANOTHER Delrin centering ring (you can see it sitting at the bottom of the head). The part count keeps risin' but if you want great results you have to work for them!
Now THAT'S a centered o-ring. The slight blue tinge is the anti-reflective coating of the ultra-clear lens (UCL).
The whole assembly is threaded down into the head...
... and we're finished!
Ah heck, I can't keep you in suspense. Shall we sneak a peek of the Spyglass v2 compared to the previous version?
In addition to the driver being completely redesigned from the ground up, we've changed a few external things as well:
- Driver pill has been changed to aluminum. Not only is it a great conductor of heat, it offers a nice contrasting color band around the head.
- Body on this light has been machined from bearing bronze, lending the red warmth of copper with the nice black tarnish of brass.
- Tailcap is now separate instead of integrated, greatly improving waterproofing.
This was my first foray into the wild world of waterjet cutting. A local shop in town offered me a small run of prototype pieces, so I brought in my drawings and they went to work. This was the result: 10 pieces of 0.060" stainless steel.
The edges were a lot rougher than I expected. This is probably because most clips are stamped with a die, not sliced with a waterjet.
I wouldn't want to do any great number of them, but for a single prototype I was willing to go at the edges with a file.
There is a nice smooth edge hiding under there, just need some elbow grease to bring it out.
The surface finish could use some love as well. I started with some wet 600 grit and moved to 0000 steel wool. I brushed the steel-wool lengthwise to give a uniform appearance.
A little elbow grease later and we're left with a much nicer looking clip.
Adding the bends to this clip was MUCH easier than bending the old wire clip of the v1 light.
So? What does it look like?
I spent a good portion of the day comparing this clip to those on other lights I own.
If I had to rate the holding power and tension I'd probably say:
- Stronger grip than Surefire long clips (A2, E2) but less springy and more rigid.
- Weaker grip than McGizmo death-grip clip (I'm fine with this; McG's clips have shredded numerous pairs of pants, they're crazy tight). More flex.
- Stronger grip and more rigid than DarkSucks Eiger clip.
- Closest comparison is probably the Moodoo HDS clip, though the Moodoo clip is probably stainless spring-steel as it seems less likely to deform than this one.
I'll finish this post with a couple glamour-shots
devil. Next update will cover the new driver and functionality so check back later will ya?
>> Return to root post
>> Continue to next post in the series
Last edited:
Another most excellent update! 
gunga
Flashaholic
How about a deep carry clip? Please say you are making a few more for loyal Canadians who would love a sample... Like me.
- Joined
- Jan 26, 2015
- Messages
- 2,123
Gorgeous work!
archimedes
Flashaholic
When can I send you money ?
phosphor22
Enlightened
Wow - the end result is stunning - the bearing bronze is really fine looking - along with the two silvery elements with it
I second Archimedes, I'm ready to Paypal now. I "need" this light in the interest of fostering good international relations with our Canadian neighbors. And the temptation is too great, for me to miss the opportunity, of owning a real Canadian light, and a prototype at that!
Well done Sir!!!
Well done Sir!!!
Last edited:
calipsoii
Flashlight Enthusiast
- Joined
- Apr 21, 2010
- Messages
- 1,412
Thanks for the comments all! It's great to hear you guys like what you see so far. It's a solid little light and I'm packing it around testing it at every opportunity I get. I still have to try a runtime test or two with the new driver and then I'll post some technical observations up. Then we can discuss building a few more for anyone who's interested. 
phosphor22
Enlightened
Thanks for the comments all! It's great to hear you guys like what you see so far. It's a solid little light and I'm packing it around testing it at every opportunity I get. I still have to try a runtime test or two with the new driver and then I'll post some technical observations up. Then we can discuss building a few more for anyone who's interested.
oo:
- Joined
- Jan 26, 2015
- Messages
- 2,123
That sounds like a great plan [emoji2]
You want to test it? Send it to me and I'll make sure Candle Power Forums most distinguished tester "run4jc" puts it through it's paces. He definitely has the equipment and more importantly the drive to do whatever it takes to evaluate your prototype. And if you so desire he can test the limits of your light as well.
calipsoii
Flashlight Enthusiast
- Joined
- Apr 21, 2010
- Messages
- 1,412
15VP v2 Measurements
Hi all! With the driver installed in a host body it now has proper heatsinking and a reflector which means I can do some real-world testing!
For this prototype driver I swapped out the 2.2uH inductor coil for a 4.7uH model. This adds a bit of stability at the cost of lower average current on high. This increased stability means I was able to experiment with microamp drive levels again (something that was very unstable with the 2.2uH inductor).
NOTE: while sub-milliamp drive currents are possible with this driver, THEY ARE NOT 100% ROCK SOLID.
The issue boils down to the microcontroller being unable to accurately read these tiny drive levels 100% of the time. Adding a bigger sense resistor would fix it (at the cost of overall efficiency). So would re-designing the circuit from scratch (something I'm unwilling to do right now). So I'm leaving the ability to set microamp drives in the firmware but recommending anyone seeking stability uses 1mA+.
In daily use I'd say I notice something wonky with the lowest-low about 20% of the time. I have a small firmware change planned that should reduce that to 10%. But again: experimental.
Ok! With that disclaimer out of the way, let's dig into some performance measurements!
Eneloop (old cell, freshly-charged)
So while we're nowhere near Zebralight output or efficiency, we're still getting decent output for a decent timespan. Good enough for my daily use.
I'd mentioned microamp drive levels:
They're pretty good! Heck, even 1mA drive levels are still very low output. Here's a comparison shot against some other lights all set to their lowest outputs:
I busted out the light-meter and makeshift light-tube to take some rough output measurements:
Some comparison readings:
Lacking a proper integrating sphere to measure lumens, I'm going to tentatively estimate output somewhere around 90lm on High? Would need run4jc or someone to measure them directly so perhaps I can see about that a bit later here.
Lastly, I shot a few comparison shots to help judge the beam & output against a couple more well-known lights:
Right: McGizmo Haiku 119 CR123
Right: Spyglass v1 Oslon
Right: Tain Flute N219
Right: Quark XP-E NW
Right: Muyshondt Aeon MkIII N219
So there we go! A short summary of my initial driver tests. I have a few changes to make to the firmware to correct things I don't like in real-world use and then it's onwards and upwards. Thanks for reading along!
>> Return to root post
>> Continue to next post in the series
Hi all! With the driver installed in a host body it now has proper heatsinking and a reflector which means I can do some real-world testing!
For this prototype driver I swapped out the 2.2uH inductor coil for a 4.7uH model. This adds a bit of stability at the cost of lower average current on high. This increased stability means I was able to experiment with microamp drive levels again (something that was very unstable with the 2.2uH inductor).
NOTE: while sub-milliamp drive currents are possible with this driver, THEY ARE NOT 100% ROCK SOLID. The issue boils down to the microcontroller being unable to accurately read these tiny drive levels 100% of the time. Adding a bigger sense resistor would fix it (at the cost of overall efficiency). So would re-designing the circuit from scratch (something I'm unwilling to do right now). So I'm leaving the ability to set microamp drives in the firmware but recommending anyone seeking stability uses 1mA+.
In daily use I'd say I notice something wonky with the lowest-low about 20% of the time. I have a small firmware change planned that should reduce that to 10%. But again: experimental.
Ok! With that disclaimer out of the way, let's dig into some performance measurements!
Eneloop (old cell, freshly-charged)
- 350mA: 90 minutes until it drops out of regulation
- 35mA: 950 minutes / 16 hours
- 1mA: 24k minutes / 400 hours / 16 days (calculated)
So while we're nowhere near Zebralight output or efficiency, we're still getting decent output for a decent timespan. Good enough for my daily use.
I'd mentioned microamp drive levels:
They're pretty good! Heck, even 1mA drive levels are still very low output. Here's a comparison shot against some other lights all set to their lowest outputs:
I busted out the light-meter and makeshift light-tube to take some rough output measurements:
Some comparison readings:
| Light | Reading |
| 4sevens 1xAA Quark NW XP-E | 2977 |
| Tain Flute N219 | 2215 |
| Spyglass v2 N219 | 3421 |
| Muyshondt Aeon MkIII N219 | 3902 |
Lacking a proper integrating sphere to measure lumens, I'm going to tentatively estimate output somewhere around 90lm on High? Would need run4jc or someone to measure them directly so perhaps I can see about that a bit later here.
Lastly, I shot a few comparison shots to help judge the beam & output against a couple more well-known lights:
Right: McGizmo Haiku 119 CR123
Right: Spyglass v1 Oslon
Right: Tain Flute N219
Right: Quark XP-E NW
Right: Muyshondt Aeon MkIII N219
So there we go! A short summary of my initial driver tests. I have a few changes to make to the firmware to correct things I don't like in real-world use and then it's onwards and upwards. Thanks for reading along!
>> Return to root post
>> Continue to next post in the series
Woah, awesome machine work! I wish I had tools like that, so many things I could do. It's cool to see that you've invested the time and effort to make what will be the ultimate flashlight for you. I have to ask though, why the 1 X AA format? No offence intended, but current battery tech tells me that light will never put out more than 400 lumens at most, and even then it may run what, 10 minutes? I guess what I'm asking is what is it about the 1 X AA format that captivates your imagination, almost to the point of obsession? I'm not saying what your doing is a waste, It's absolutely AWESOME. It's also really cool to see a custom driver purpose built to pull power from a AA. The current measurement scheme is genious, BTW. ROCK ON!
calipsoii
Flashlight Enthusiast
- Joined
- Apr 21, 2010
- Messages
- 1,412
@NiTyson
AA is my favorite format, it's pretty much as simple as that! In real-world use I've never found that my pocket light needs to be >100 lumens. If I'm walking the dog I'll reach for a Malkoff or something more powerful. There are already many great offerings in the high-power arena so I don't intend to try and recreate that wheel. I'm finding the AA market continues to be disappointing though, so that's an area I'm willing to put effort into. Plus this project has been one great big ongoing learning experience in many different fields, which is good for a guy who's not getting any younger like me.
Cheers!
AA is my favorite format, it's pretty much as simple as that! In real-world use I've never found that my pocket light needs to be >100 lumens. If I'm walking the dog I'll reach for a Malkoff or something more powerful. There are already many great offerings in the high-power arena so I don't intend to try and recreate that wheel. I'm finding the AA market continues to be disappointing though, so that's an area I'm willing to put effort into. Plus this project has been one great big ongoing learning experience in many different fields, which is good for a guy who's not getting any younger like me.
Cheers!
- Joined
- Jan 26, 2015
- Messages
- 2,123
Some great updates! Beam pattern looks very nice and I love the AA format. My malkoff mdc aa 219 is one of my favorite lights.
archimedes
Flashaholic
I have used a lot of different torches ... a lot ... really.
My virtually perfect AA torch would be my Spyglass Gen1 ... with a higher high ... a lower low ... and in titanium :devil:
My virtually perfect AA torch would be my Spyglass Gen1 ... with a higher high ... a lower low ... and in titanium :devil:
@NiTyson
AA is my favorite format, it's pretty much as simple as that! In real-world use I've never found that my pocket light needs to be >100 lumens. If I'm walking the dog I'll reach for a Malkoff or something more powerful. There are already many great offerings in the high-power arena so I don't intend to try and recreate that wheel. I'm finding the AA market continues to be disappointing though, so that's an area I'm willing to put effort into. Plus this project has been one great big ongoing learning experience in many different fields, which is good for a guy who's not getting any younger like me.
Cheers!
I sure wasn't criticizing, mainly curious. I just completed a LED conversion on an old Maglite, but it's so big, I don't think I will use it too much! But I have it if I need it. At least if anyone tries to attack me, I can knock their brains out with it.:devil: That's one of the reasons the police like them so much. I can definitely tell why a AA light would be nice though. My Convoy s2+ is a little big to carry in a pocket all day, and it doesn't help that it lacks a clip to attach it on the outside of your pocket. A AA light is small enough that you can forget you have it, while a 18650 light is a little harder to miss. Btw, did you ditch the secondary LED? I haven't seen it in recent post. Anyway, keep up the awesomeness!
calipsoii
Flashlight Enthusiast
- Joined
- Apr 21, 2010
- Messages
- 1,412
A wild Spyglass sighting!
Been carrying it every day since built and getting a good feel for what works and what doesn't. Have 2 of 6 firmware changes finished and then it'll be reflashed and tested. So far results are really positive.
Still working away here so stay tuned!
Been carrying it every day since built and getting a good feel for what works and what doesn't. Have 2 of 6 firmware changes finished and then it'll be reflashed and tested. So far results are really positive.
- High is high enough (in my opinion); v1 Spyglass was definitely too low at times
- Low is too low (in my opinion); I might run it at 1-2mA instead of 0.5mA because I simply don't need that low
- 3 output levels seems perfect - 4 was too many to tap through
- I'm running it with mode memory OFF because I like cycling up through L->M->H and not guessing what mode I was on last
- The bronze is heavy as hell. I like the color though (look at that tarnish!) so maybe a bronze driver pill and titanium body would look nice?
- The groove in the clip runs too high up the center, it needs to be half as long. Holding power is ok but I'm going to look for stainless spring steel for final clip
- I'd like a hairpin at the top of the clip so that it grabs my pocket material, right now the top slides down the pocket
Still working away here so stay tuned!
Latest posts
-
-
-
-
-
-
-
-
Test/review of Xtar PB2C Charger & power bank
- Latest: Photolunatic
-
Xtar PB2 circuitry upgrades/changes?
- Latest: Photolunatic
