Why no high-power boost drivers?

30 posts / 0 new
Last post
zak.wilson
Offline
Last seen: 19 hours 47 min ago
Joined: 09/29/2014 - 14:27
Posts: 1034
Why no high-power boost drivers?

Looking around at commercial and community-designed drivers, there’s one for just about everything. Want a direct-drive FET with a sub-lumen moonlight? No problem. Constant current buck for a 3V emitter with multiple cells in series? Easy. Run an XP-L off a AAA? No problem. Run an XHP50 from a single 18650? Keep an XP-L regulated for the whole life of the battery? Crickets….

Zebralight, Armytek and the Noctigon Meteor do this, so it’s obviously not a technical limitation. I know they’re a little more expensive to produce, but that doesn’t typically stop people here. Zebralight is even running a 12V XHP35 from a single 18650 now. With quite a few posts on running series 18350s in 18650 hosts for 6V emitters, there’s clearly interest in building lights that would benefit from such a driver. Why isn’t there one?

Ceilingbounce – flashlight testing and runtime graphs for Android

pilotdog68
pilotdog68's picture
Offline
Last seen: 7 months 2 weeks ago
Joined: 05/30/2013 - 23:31
Posts: 6422
Location: Held against my will in IOWA, USA

I imagine mostly time. Something like that takes an entire R&D department a while to design, test, and redesign. It’s hard for one person to do it, especially when it’s just a hobby not a full-time job.

Maybe one day it will be real, but even a programmable buck driver took a long time to become a reality.

My Favorite Modded Lights: X6R, S8 , X2R , M6, SP03

Major Projects:  Illuminated Tailcap, TripleDown/TripleStack Driver

Diode663
Offline
Last seen: 5 years 4 weeks ago
Joined: 05/26/2015 - 22:24
Posts: 27
Location: United States

I tried and I can’t get mine small enough. It’s 1 inch square and doesn’t have a micro on it. The only real way to do it would be with a pic micro that can handle being a smps. The other problem is that even with moddest output currents the input current is significant and it requires a decent size inductor.

doubleone44
doubleone44's picture
Offline
Last seen: 2 years 5 months ago
Joined: 08/05/2015 - 11:42
Posts: 172
Location: The Netherlands

There are some higher power boost drivers available. Over at LPF, we have the X-boost, which can do up to 4.5A from a 18650. We use them for our 445nm lasers with higher forward voltages. They also have PWM input, so they could be stacked upon an MCU. The only downside: Cost. These drivers are 20 dollars+, mostly because of the very good stability laser diodes need.

zak.wilson
Offline
Last seen: 19 hours 47 min ago
Joined: 09/29/2014 - 14:27
Posts: 1034

The X-boost looks ideal. It’s not clear to me what the output voltage options are though.

Ceilingbounce – flashlight testing and runtime graphs for Android

Diode663
Offline
Last seen: 5 years 4 weeks ago
Joined: 05/26/2015 - 22:24
Posts: 27
Location: United States

The output voltage would vary depending on current

zak.wilson
Offline
Last seen: 19 hours 47 min ago
Joined: 09/29/2014 - 14:27
Posts: 1034

I tried taking a closer look at the X-boost to try, with my fairly basic understanding of electronics, to understand how it works. I noticed that in all pictures of them, one of the ICs has its markings sanded off, painted over or photoshopped out (example).

This… is not to my liking. It’s as if the producer of these drivers does not want others to learn about how they work and potentially compete with him. If there was some super unique secret sauce going on here, I could see it, but these are off the shelf parts put together in more or less the manner depicted in widely available reference diagrams.

So with that in mind, I did some looking based on the stated capabilities of the Super X Boost version of the driver and the appearance of the secret IC. It is almost certainly the Texas Instruments LM2588 5A flyback regulator, and indeed, the SXB does look quite a bit like the “typical application” reference diagram.

Ceilingbounce – flashlight testing and runtime graphs for Android

Fritz t. Cat
Fritz t. Cat's picture
Offline
Last seen: 3 years 1 week ago
Joined: 05/07/2013 - 00:33
Posts: 2535
Location: Si Valley

A boost driver has constant input current and intermittent output. So one can’t drive any LED to it maximum with a boost driver without inefficiency and possibly burning it out. One might use a filter or a boost buck combination, but those may not fit in a flashlight. So a one cell light with an XHP-50 might work, but it isn’t clear that one can get more light out of it than with say a Nichia 219C and regulation.

Flashlight designers should look at lighthouses and pottery.
这些谁设计的手电筒应该看灯塔,以及在陶器。

zak.wilson
Offline
Last seen: 19 hours 47 min ago
Joined: 09/29/2014 - 14:27
Posts: 1034

Tomorrow, Zebralight is scheduled to release specs for the SC600 Mk. III, which uses a single 18650 to power an XHP35. Those only come in 12V. I’m eager to see the claimed output, which I do believe is usually found to be accurate from Zebralight in independent testsing. Zebralight doesn’t usually make bad engineering decisions or build impractical hot-rod lights, so I imagine they’re getting enough out of the new emitter to make up for the inefficiency of more than tripling the voltage. I’d like to be able to replicate that in DIY lights. Also, there are XHP35s listed with 90+ CRI in tints where high-CRI was previously unavailable. It is not clear to me yet if all of these parts are actually being produced.

I love the 219C for its low forward voltage, nice tint and good color rendering. I’ve put them in three different lights so far and plan on building several more.

Ceilingbounce – flashlight testing and runtime graphs for Android

Ronin42
Ronin42's picture
Offline
Last seen: 2 years 1 month ago
Joined: 12/17/2014 - 22:17
Posts: 1760
Location: Alameda, CA

Of course it can be done
Fenix PD40 1×26650 MT-G2 Light
JM-35
etc etc.

Part of the issue is Why?

If you want to drive a 6v or 12v emitter (well) use multiple cells.

sure it is cool to do it on one cell but it would be cool to drive an F1 car to work (neither make as much sense as the right tool for the job).

True running an MT-G2 off a 32650 3.6v cell probably makes more sense then 18650 but folks just seem to want what is not available (because) it is not available.

So yes it can be done but most will agree it has not been done to the emitter capability, we seem to always be limited by the single battery being used.

The MT-G2 and XHP70 can pull 6 amps, 8 amps, heck 10 amps consistently. For an hour that would need 60WH battery assuming a perfect 100% driver.

http://flashlightwiki.com/Battery_capacity

Humm, I know lets add a cell, or 2, bingo! Bobs your uncle.

I guess the answer to your question is the Battery.

(“It’s good that most people can’t remember their previous lives. Otherwise
things would be a lot more complicated than they already are.”
Ajaan Lee Dhammadharo)

Ronin42
Ronin42's picture
Offline
Last seen: 2 years 1 month ago
Joined: 12/17/2014 - 22:17
Posts: 1760
Location: Alameda, CA

But on a different note, start getting your pitch forks ready to start a revolt so that the next BLF designed light is the … wait for it…

V3 ×7 aka 26650/26700 6v emitter light.

(“It’s good that most people can’t remember their previous lives. Otherwise
things would be a lot more complicated than they already are.”
Ajaan Lee Dhammadharo)

Ronin42
Ronin42's picture
Offline
Last seen: 2 years 1 month ago
Joined: 12/17/2014 - 22:17
Posts: 1760
Location: Alameda, CA

And

Boost Driver Challenge – Technical Discussion Thread – Experts PLEASE step inside.
http://budgetlightforum.com/node/21596

(“It’s good that most people can’t remember their previous lives. Otherwise
things would be a lot more complicated than they already are.”
Ajaan Lee Dhammadharo)

zak.wilson
Offline
Last seen: 19 hours 47 min ago
Joined: 09/29/2014 - 14:27
Posts: 1034

There are a few reasons to want boost drivers, though the appearance of the 219C has mitigated some of them some for me. Draining a VTC5 in 10 minutes trying to get 5000 lumens out of an XHP70 in a pocketable host is not among them; I don’t own any asbestos gloves.

Zebralight has been using them for years to ensure its lights never fall out of regulation. There’s some thermal throttling if the light isn’t adequately cooled and a low-voltage stepdown, but otherwise the runtime graph of a Zebralight is flat. Some Armytek models do this as well. I like this feature and would like to be able to build DIY lights suitable for pocket carry that have it. Using a buck driver and two 18350s in series would do it, but that cuts capacity in half or worse. To my way of thinking, that defeats the purpose. The 219C’s low forward voltage makes this a little less important.

Another reason, as I mentioned is the XHP35. Sure, it can put out more light than any of the 3V emitters, but what’s really compelling is the high-CRI range. This is a link to order the XHP35 HI with a color temperature of 7000K and a minimum CRI of 90. It might not actually exist. It seems like it shouldn’t exist, but it’s in Cree’s datasheet and you can order a reel of them from Mouser. I want to see what that looks like (daylight with the sun blocked by clouds, I imagine). I do not want to have to build a 4S light to use those. 4S is getting out of my comfort zone for not being able to monitor the voltage of each cell individually. I also suspect the XHP35 HI will be a compelling emitter for throwers. No, it’s probably not going to beat a dedomed XP-G2 if you’re just trying to post big numbers, but if you want a versatile beam that’s useful for seeing stuff at a distance, this could be the next big thing.

Ceilingbounce – flashlight testing and runtime graphs for Android

cncyana
Offline
Last seen: 3 months 3 days ago
Joined: 03/24/2013 - 02:21
Posts: 517
Location: los angeles

Ronin42 wrote:
But on a different note, start getting your pitch forks ready to start a revolt so that the next BLF designed light is the … wait for it…

V3 ×7 aka 26650/26700 6v emitter light.

A 26650-powered X7 big brother version of the X6 would be great but which emitter?

Life’s Questions/Answers here: http://www.answersingenesis.org/home/area/qa.asp “So then, each of us will give an account of himself to God.” Romans 14:12

Hikelite
Offline
Last seen: 8 months 2 weeks ago
Joined: 07/13/2011 - 16:18
Posts: 3888
Location: Honeyland

zak.wilson wrote:
but what's really compelling is the high-CRI range. "This":http://www.mouser.com/ProductDetail/Cree-Inc/XHP35A-H0-0000-0D0UA40DT/?q... is a link to order the XHP35 HI with a color temperature of 7000K and a minimum CRI of 90.

 

Seems a bit redundant as in reliality a great majority will not care about High CRI 7000K light really. That slight warmness of neutral white is what is appealing (5500K to 5000K), not the fact that 7000K have normal 65CRI or 70CRI (depending on the LED) to run form them. I understand it is a curiosity, that could be satisfied for the moment by buying the LED and testing out with a DC power supply (an immediate test) 

 

zak.wilson wrote:
 I also suspect the XHP35 HI will be a compelling emitter for throwers. No, it's probably not going to beat a dedomed XP-G2 if you're just trying to post big numbers, but if you want a versatile beam that's useful for seeing stuff at a distance, this could be the next big thing.

Since the die is even bigger than XM-L2, it is not that of a huge deal in throw (next big thing), reality is we still need smaller die LEDs that are more efficient for throwers, not larger and larger that are more efficient (XHP35).

Halo...
Halo...'s picture
Offline
Last seen: 5 years 3 months ago
Joined: 12/15/2011 - 02:39
Posts: 3304
Location: Halo island

zak.wilson wrote:
Using a buck driver and two 18350s in series would do it, but that cuts capacity in half or worse. To my way of thinking, that defeats the purpose.

2* 18350 does not cut capacity in half. The mA is not actual capacity when comparing 1* cell to 2* cells. You need watts, volts * amps to compare. You do get less watts from 2*18350s than from 1*18650 since only 18650s get the latest li-ion technology but it’s not automatically just half. And after straight battery watts you do need to consider driver losses. The more you have to boost (or buck) the higher the losses are. Over the whole discharge curve 2*18350s will give closer to the voltage you need than 1*18650.
zak.wilson
Offline
Last seen: 19 hours 47 min ago
Joined: 09/29/2014 - 14:27
Posts: 1034

There aren’t any other high-CRI emitters in daylight tints (5000-5500K) either. The coolest 219B R9050 is 4500K. There’s a 90+ CRI XHP35 listed for every color temperature the emitter comes in, and the HI version of the same. I care about both tint and CRI, and while I mostly prefer ~5000K, options and variety are fun.

The XHP35 HI isn’t likely to set throw records. That wasn’t my meaning in saying it would be good for throwers. I think it’s a good bet it can match the throw of the XP-L HI but with more output and a larger hotspot, making the same thrower more useful. It also comes in more tints. Personally, I’d go with something fairly warm to better penetrate any fog or haze.

I didn’t realize there was an 18350 that hits nearly 1000 mAh under light loads, exceeding its advertised capacity. My back of the envelope calculation was based on 850 mAh for the 18350 and 3400 for the 18650, giving 6290 mWh for 2×18350 and 12580 for the 18650. That’s half. The Xtar 18350 does a little better than that. Would you consider “a bit more than half” a fair description?

Ceilingbounce – flashlight testing and runtime graphs for Android

Fritz t. Cat
Fritz t. Cat's picture
Offline
Last seen: 3 years 1 week ago
Joined: 05/07/2013 - 00:33
Posts: 2535
Location: Si Valley

There is an easy way to drive an XHP35 in a large but not huge light. That is the LED Lenser P14 with 4 × 14500. http://www.tmart.com/Lenser-P14-LED-Flashlight_p236578.html The beauty of this is that since the driver consists of nothing but resistors and a mechanical switch, one only needs to file down the pillar that the LED sits on so a filed down star will fit and maybe swap some resistors. The switch and resistors don’t care what voltage they are handling. This host offers Fresnel-like optics as well as high build quality, and absolutely no ripple in any mode. The down side is that the cells are not closely packed in the battery tube and 14500s don’t have the energy density of 18650s.

Flashlight designers should look at lighthouses and pottery.
这些谁设计的手电筒应该看灯塔,以及在陶器。

zak.wilson
Offline
Last seen: 19 hours 47 min ago
Joined: 09/29/2014 - 14:27
Posts: 1034

That’s actually a cool idea. I’m a little hesitant about running 4 Li-ions in series though. I guess protected cells would be a good solution to prevent reverse charging.

I still want good boost drivers. I may try my hand at building one on a breadboard.

Ceilingbounce – flashlight testing and runtime graphs for Android

Halo...
Halo...'s picture
Offline
Last seen: 5 years 3 months ago
Joined: 12/15/2011 - 02:39
Posts: 3304
Location: Halo island

You also need to pull over twice the amps from the 18650 so you need to look at capacity there. HKJ lists Wh for a few different currents he tests at. And then is driver losses.
I didn’t do any calculations before posting, half the capacity just sounded off to me.

We have high CRI nichias in 5000k. 90+ CRI XHP35 may be listed in the datasheet in every color temp but are they actually available for sale? I’ve seen datasheets claim something should be available but it never appeared. So for the best stuff I believe it when I see it.

Boost drivers just take more work. And are more difficult for high currents in a small space. You can’t just throw a reference design together and have it work well. If you read through the LPF boost driver threads you’ll see a lot of testing and revisions went into them. Unfortunately it seems not many want to do it for free. Some of the LPF boost driver have been shared freely. Just the parts to build one cost a decent bit. Shows us how cheap our 7135 or fet drivers are.

Halo...
Halo...'s picture
Offline
Last seen: 5 years 3 months ago
Joined: 12/15/2011 - 02:39
Posts: 3304
Location: Halo island

zak.wilson wrote:
I still want good boost drivers. I may try my hand at building one on a breadboard.

If you have the motivation to make us some nice open boost drivers I wouldn’t want to discourage you. I’d much rather be wrong about it being more work than most want to do for free.
Hikelite
Offline
Last seen: 8 months 2 weeks ago
Joined: 07/13/2011 - 16:18
Posts: 3888
Location: Honeyland

Buck is the best in practice in a flashlight due to lower current input and lower current output when lots of watts power is needed as output, resistance does not become enemy number one in this case, neither the cells need massive current capability, in theory people get all super extremely scared of "series issues", yet the whole flashlight industry runs on 4S 18650 or 3S 18650 flashlight and I have not heard or read any issues whatsoever.

Fritz t. Cat
Fritz t. Cat's picture
Offline
Last seen: 3 years 1 week ago
Joined: 05/07/2013 - 00:33
Posts: 2535
Location: Si Valley

Halo… wrote:
zak.wilson wrote:
Using a buck driver and two 18350s in series would do it, but that cuts capacity in half or worse. To my way of thinking, that defeats the purpose.

2* 18350 does not cut capacity in half. The mA is not actual capacity when comparing 1* cell to 2* cells. You need watts, volts * amps to compare. You do get less watts from 2*18350s than from 1*18650 since only 18650s get the latest li-ion technology but it’s not automatically just half. And after straight battery watts you do need to consider driver losses. The more you have to boost (or buck) the higher the losses are. Over the whole discharge curve 2*18350s will give closer to the voltage you need than 1*18650.

It appears to me to cut it more than in half. 18350s, like 14500s don’t have half the capacity of 18650s as one might hope. I see numbers like 800 mAh, not 3.4 / 2 = 1,700.

Flashlight designers should look at lighthouses and pottery.
这些谁设计的手电筒应该看灯塔,以及在陶器。

zak.wilson
Offline
Last seen: 19 hours 47 min ago
Joined: 09/29/2014 - 14:27
Posts: 1034
Fritz t. Cat wrote:
It appears to me to cut it more than in half. 18350s, like 14500s don’t have half the capacity of 18650s as one might hope. I see numbers like 800 mAh, not 3.4 / 2 = 1,700.

We were talking about running two in series, so it’s necessary to compare Watt-hours, not Amp-hours. They would be at different loads too, and then there’s the question of driver efficiency. We can compute the capacity, but have to guess a bit about the load and driver efficiency. My estimate is that two 18350s in series to run a 6V emitter probably have a little more than half the runtime of one 18650 and a boost driver, assuming the highest capacity batteries for both and on the order of 2A (12W) to the emitter.

Ceilingbounce – flashlight testing and runtime graphs for Android

texas shooter
Offline
Last seen: 14 hours 2 min ago
Joined: 08/26/2012 - 02:14
Posts: 1821
Location: Texas

I prefer single cell lights for size and ease of use, I’m also a lumen junkie. Let me explain my view on ease of use, I can take any cell of the right size and put it in a light. No grouping cells by manufacturer or age or its use history like in multi cell lights. A group of drivers for the newer higher voltage Cree leds would be very welcomed in my book. Single 18650/26650 with XHP35, XPH50, and MT-G2 are the lights I’m wanting. Fenix PD40 gets a lot of use from me. The SC600 Mk III 18650 XHP35 is already ordered. I have almost no electrical engineering skills so I buy what’s out there. If our forum engineers or hobbyist designed a driver that meets these needs I’ll be in hog heaven. These are the levels of modification I can “safely” do. Being a P60 diehard fan I’d also love to see XHP-35, XPH-50 and MT-G2 single cell dropins.

pilotdog68
pilotdog68's picture
Offline
Last seen: 7 months 2 weeks ago
Joined: 05/30/2013 - 23:31
Posts: 6422
Location: Held against my will in IOWA, USA

I guess I don’t see the point of boosting 4.2v all the way up for an XHP35… Just use an XP-L

My Favorite Modded Lights: X6R, S8 , X2R , M6, SP03

Major Projects:  Illuminated Tailcap, TripleDown/TripleStack Driver

Fritz t. Cat
Fritz t. Cat's picture
Offline
Last seen: 3 years 1 week ago
Joined: 05/07/2013 - 00:33
Posts: 2535
Location: Si Valley
zak.wilson wrote:
Fritz t. Cat wrote:
It appears to me to cut it more than in half. 18350s, like 14500s don’t have half the capacity of 18650s as one might hope. I see numbers like 800 mAh, not 3.4 / 2 = 1,700.

We were talking about running two in series, so it’s necessary to compare Watt-hours, not Amp-hours. They would be at different loads too, and then there’s the question of driver efficiency. We can compute the capacity, but have to guess a bit about the load and driver efficiency. My estimate is that two 18350s in series to run a 6V emitter probably have a little more than half the runtime of one 18650 and a boost driver, assuming the highest capacity batteries for both and on the order of 2A (12W) to the emitter.


Yes, for the watt hours to be the same the two would have to have half the charge capacity each.

Flashlight designers should look at lighthouses and pottery.
这些谁设计的手电筒应该看灯塔,以及在陶器。

Ronin42
Ronin42's picture
Offline
Last seen: 2 years 1 month ago
Joined: 12/17/2014 - 22:17
Posts: 1760
Location: Alameda, CA

cncyana wrote:
Ronin42 wrote:
But on a different note, start getting your pitch forks ready to start a revolt so that the next BLF designed light is the … wait for it…

V3 ×7 aka 26650/26700 6v emitter light.

A 26650-powered X7 big brother version of the X6 would be great but which emitter?

which emitter indeed. Let the mob decide?

(“It’s good that most people can’t remember their previous lives. Otherwise
things would be a lot more complicated than they already are.”
Ajaan Lee Dhammadharo)

zak.wilson
Offline
Last seen: 19 hours 47 min ago
Joined: 09/29/2014 - 14:27
Posts: 1034

So specs are out on the SC600-III. It falls a bit short of what I’d hoped for – 1300 lumens on high for the cool white version, and only 1126 for the neutral. Neutral takes a bigger hit here than it did in the XM-L2 model.

On the other hand, runtime is improved at every output level. That doesn’t mean much at the highest level, as Zebralights are thermally regulated and would run longer if they get throttled for producing too much heat, but at the lower levels, it’s significant. 24 minutes more at 360lm. 3 hours more at 70lm. ZL’s spreadsheet claims a slight CRI bump for both cool and neutral tints and the cool is now a more daylight-like 5700K as opposed to the old 6300K.

Is this all worth the effort? Well… maybe. Certainly, if you’re buying a $95 single-cell flashlight, I think Zebralight does more to justify its price than many, perhaps any similarly-priced products. If you’re building a light and you have the choice between a $5 driver and $5 emitter or a $20 driver and $10 emitter, the relative impact of those costs is more significant.

Worth it or not, once I have a couple other things completed, I think I’ll try my hand at building a boost driver. I’ve never really done any electrical engineering, so I probably won’t come up with the best driver ever, but I bet I can come up with something that works.

Ceilingbounce – flashlight testing and runtime graphs for Android

DavidEF
DavidEF's picture
Offline
Last seen: 1 week 3 days ago
Joined: 06/05/2014 - 06:00
Posts: 7699
Location: Salisbury, North Carolina, USA

zak.wilson wrote:
So specs are out on the SC600-III. It falls a bit short of what I’d hoped for – 1300 lumens on high for the cool white version, and only 1126 for the neutral. Neutral takes a bigger hit here than it did in the XM-L2 model.

On the other hand, runtime is improved at every output level. That doesn’t mean much at the highest level, as Zebralights are thermally regulated and would run longer if they get throttled for producing too much heat, but at the lower levels, it’s significant. 24 minutes more at 360lm. 3 hours more at 70lm. ZL’s spreadsheet claims a slight CRI bump for both cool and neutral tints and the cool is now a more daylight-like 5700K as opposed to the old 6300K.

Is this all worth the effort? Well… maybe. Certainly, if you’re buying a $95 single-cell flashlight, I think Zebralight does more to justify its price than many, perhaps any similarly-priced products. If you’re building a light and you have the choice between a $5 driver and $5 emitter or a $20 driver and $10 emitter, the relative impact of those costs is more significant.

Worth it or not, once I have a couple other things completed, I think I’ll try my hand at building a boost driver. I’ve never really done any electrical engineering, so I probably won’t come up with the best driver ever, but I bet I can come up with something that works.


Go for it! I’m looking forward to seeing what you can come up with!

The Cycle of Goodness: “No one prospers without rendering benefit to others”
- The YKK Philosophy