If you had your wish...

[dat2zip]

Here's the deal. Most of the converter boards I have designed to date have been targeted for CPF or custom for a specific CPF'ers light, but, driven by my best guess at what would be accepted by CPF. When there was no converter boards available anything was acceptable since in the beginning there was a vacuum and void. Today that is no longer the case.

Going forward I would like your input on what converter would you desire the most. I'll list some that I have on the plate that are either concept (idea only) or in process. Your ideas/thoughts may become the next product from the shoppe.

GDuP. Take the GD buck/boost and add a uP on it. The UI I would propose is a simple on state timing with three levels (low, med, high). The current thought would be to put the microprocessor on the backside of the GD. This additional height along with a dedicated blank anode contact board would still be mountable in the current Aleph Light engines allowing the three level UI to be upgraded into existing Alephs or E series lights.
Restrictions: Single Lithium cell only. (Primary and rechargeable)
Current state: Just starting preliminary dual board prototype.


Hammerhead: Next generation Shark. The microprocessor would be included on the board. Dimensions would grow and it would probably be a square board and not round. A large plane on the back along with a couple of holes would allow bolting a heatsink and allow mounting the board/heatsink to a larger heat sink assembly. The microprocessor would accomodate the power switch for input, dedicated rotary switch and a dedicated analog pot for diverse application usage.
Current state: Concept only.

LED clock: A redesign of my LED clock I made after I graduated from High School. Integrate an atomic clock receiver for accurate time all the time and simplify assembly and costs.
Current State: Concept only

BBM: New X2 driver for Don's Lunasol project. The base design is the original Badboy driver put on the X2 driver footprint. Will drive 6 3mm LEDs off a single primary Lithium battery.
Current state: Completed prototype and testing. Moving into production.

So, that's it. I can't divulge too much information. I would like to hear or discuss any other ideas.

I'd like to also have you think about your idea. Making a single converter board for your custom upgrade of the light you are holding in your hand right now will more than likely not be feasible. Making a converter board that many people beyond just the one is more like feasible. It takes quantity to make it into a production part.

Let the discussions begin.

Wayne

 

[paulr]

1. A board for running a powerful led on a single high-current NiMH AA cell, i.e. something that can put a solid 3.5 watts through an SSC or Cree on 1.1 volts input. That might take bigger inductors than on the existing boards but it's ok.

2. A boost converter for running the 3-led Rebel Star on two 123's or 3 or 4 AA's or high current AAA's. That Star has three Rebels in series so it needs around 9.6 volts input.

3. A multi-led boost converter, to run 3 to 5 independently regulated 5mm leds on a single AA or AAA cell. Existing multi-led lights tend to just put the leds in parallel through a single resistor, so small differences in vf make some of the leds get more current than others.

4. The microprocessor boards should have open firmware and should have PCB contact points for loading configuration and maybe reloading code.

5. This idea is completely insane but how about a user-programmable, replacement controller board for the Photon Freedom and/or the new 4-led version of the Freedom.

 

[Kiessling]

The GDuP will most likely be a universal approach and could be used in many lights. Also, you'd be able to make a DBuP et al without too much hassle from this concept, too, I guess.

My concern is, and has always been, the interface. Which is another can of worms, but certainly a can worth opening and discussing before going into production.

Then ... the Hammerhead sounds promising, too, as multi-LED big lights can become a very bright reaity now, and such a driver with a nice interface (think trim pod for brightness) will be a killer.

bernie


P.S.: I am arguing from a KISS POV. Not a multi-klick wizbang POV.

 

[paulr]

How about including charge controllers on some of those boards, to allow smart recharging of an nimh or lithium ion pack inside the flashlight, from an external 5v or 12v source.

 

[Gryloc]

Dat2zip,

Do you plan on making a boost circuit for multi-LED lights with a higher Vf output? The shark is limited at 22V, which can make a 7 LED Rebel more difficult to work out. Sure, if the current is lower on the Rebel, the total Vf of all 7 will be in the 22V range, but what if you want to drive 7 or more Crees, Seouls, or Rebels at more current? The Vf output would bot be enough.

The only way around this is using multiple driver boards on multiple banks of LEDs (meaning more $ for us cheap modders, as well as added complexity and more space inside the light consumed). Plus, this would work perfectly with the new 7 Rebel Flower board you just released. Maybe this higher output Vf can be incorporated into this Hammerhead circuit you mentioned. The only other alternative is George's MaxFlex. The MaxFlex has max Vf out of 24V (just enough, but what about another 2V more?), and it has a UI interface. Currently, it is not available yet.


Aside from that, a Badboy (or equivalent) buck converter that can handle more current out safely would be great! Possibly 1500mA, or an extreme max of <2000mA. This would be awesome for those who wish to overdrive their high-efficiency emitters for that extra light. The Cree can handle the extra current (ooh, the Q5), and the Rebel can almost take it. However, with this new Badboy (call it "StrongBoy" he he), you will be ready for the "New K2" emitters using the Rebel die technology in a beefier package. There are production samples floating around, and I bet you have seen them. Think of the future...

I know of one good application for this driver. Using a 2C Mag with the "New K2" 0240 part (240lm at 1A), or a Rebel 0100, on a C hotlips heatsink. This is all powered by two 18650's or 2 of AW's li-ion C cell. Talk about a real throw-master, all with long lasting light due to a nice, efficient current-regulated buck driver. I am sure the Badboy, or the SOB can be "beefed up" to handle the extra current, right?

Well, thanks for listening. I am happy to hear that the driver selection is growing. It is awesome knowing that you can get a good current-regulated driver for almost any application.


-Tony

 

[paulr]

While we're talking about m*g conversions, maybe a hotwire controller (along the lines of the LVM) could help some projects out.

 

[dat2zip]

Dat2zip,

Do you plan on making a boost circuit for multi-LED lights with a higher Vf output? The shark is limited at 22V, which can make a 7 LED Rebel more difficult to work out. Sure, if the current is lower on the Rebel, the total Vf of all 7 will be in the 22V range, but what if you want to drive 7 or more Crees, Seouls, or Rebels at more current? The Vf output would bot be enough.


-Tony

Tony,

Funny you should ask. The next batch of Sharks coming out soon will have a few component values tweaked to raise the output limit to ~27V or so. This should accommodate 7 LEDs in series like a 7 up flower arrangement.


Good discussion. Keep the information coming.

I doubt I will work on a single cell driver beyond a real simple Zetex style driver in a real tiny package. I can see possibly offering this, but, there are now many offshore drivers that are single cell (1.5V) and offer a uP based UI.

Wayne

 

[havand]

How about including charge controllers on some of those boards, to allow smart recharging of an nimh or lithium ion pack inside the flashlight, from an external 5v or 12v source.

How about making independent boards that could do this too? I've been killing myself lately looking for a premade board that can boost/buck regulated.

 

[paulr]

Wayne, I wish you would do some good single AA converters. I think the lack of such converters is one reason the top modder lights all use 123's instead. Of course I can't tell the modders what to build. But I really wish there was such as thing as a 1AA McLux PD-S, and I don't think there's a practical way right now for Don to build that even if he wanted to.

One other idea: how about making the microprocessor board able to communicate with a computer through an optical interface, maybe using the flashlight led as a photodetector (I think all leds are photosensitive). Then you'd have a USB dongle with its own led and detector to wirelessly exchange data with the flashlight. If the flashlight led is IR sensitive and (being so powerful) it emits significant IR of its own when powered, then it might even be able to talk to a computer IRDA port directly (kluuuudge!!!).

 

[Gryloc]

Dat2zip,

I am all tingly inside knowing that the "New Shark" will have higher Vf capabilities of 27V! :twothumbs You can run 7 emitters at a decent current, or even squeeze in an 8th at slightly lower currents. Yipee! I am glad that the Shark is already pretty beefy so you can use higher current levels over 1A. Could you release to us how much power that the "New Shark" can deal with? Sure it can handle a Vin of <20V and a max Iin of 4A, but what current can be supplied to the emitters when the Vf is 27V? This is considering the efficiency of the board and assuming that the driver is attached to a heatsink. I would love to see this new circuit power 7 rebels or the "New K2" at 1500mA or more safely. I understand that you would have to have a higher Vin (4-5 li-ion or 12NiMH).


Also, I was doing some thinking about concepts. I think that it is neat that you flashlight can have communication abilities with PC in some way (whether a small wired plug, an optical sensor, or some wireless transmission like wifi). But anything too exotic will translate into higher costs and extremely long development costs, especially if optical and wireless interfaces are a complicated matter to learn. Maybe it is more viable to use a small 2-4 pin data port using some cheap and small nylon connector. Maybe just providing the option to include an inch-long wired nylon connector with the converter, then the end user can solder each wire to the appropriate wire to the solder pads on the circuit board. This is definitely doable, but the only obstacle is making the UI of the converter compatible with an ordinary PC using Serial or even USB connection. Maybe this is something you can incorporate in the more distant future. This will give more of the average joe the ability to add features to the microprocessor for a more personal flashlight.


Anyway, I just wanted to throw out the idea of an alternative way to indicate flashlight modes when using multi-mode circuits. One way is to guess what mode you are in while pressing the buttons, but working in the menu modes could be more challenging. Most make their flashlights flash the main LED emitter to signify a change in mode (which can look cool, but can be a nuisance, too). Some could use external LED indicators, but this would make it more difficult to waterproof your light.

I had the idea when playing with my cell phone. All you need is a little bit of drive circuitry for a low power vibrating motor! This would mean two solder pads for this function and a small driving transistor or mosfet. That wouldn't take up much board space. You always hold the flashlight in your hand when changing modes, so why not? Vibrator motors are very small and probably very cheap. If the motors can be bought fully enclosed, then the end user could easily hot-glue them to the light body on the interior. When you press the button, you will feel a confirming vibrate pulse. These pulses can be short, long, or both for a more complex interface. Therefore, your UI can include more features.

That is just an exotic idea. It could be done and I can see how practical it could be. It is easier to feel the vibrating pulses, or beats (as to count them to see which mode you are in) then trying to feel light pulses. The vibrating pulses can be shorter and closer together, and it might make it faster to navigate through modes. The only other way to show modes in a more detailed manner is to use an LCD screen (going overboard on a flashlight). With this simple, low current, 2-wire output, people could also hook up small buzzers or beepers as an sound indicator. The sound would be muffled a bit in more solid lights, but you may still be able to hear it. Some may hear the vibrating pulses. A vibrator motor in a flashlight would be cool, as long as people do not get the wrong idea...


It would be cool to see a single cell converter circuit, but then again, I see the issues of single cell form factors. For a single emitter to run at a decent current, such as 900mA (and a good emitter with a Vf of 3.3V), the current input for a 1.2V NiMH cell would be insane (over 3A with a converter efficiency of 80%). I hear that efficiencies of such converters are always pretty low, but maybe not. Due to the chemistry of the battery, some batteries could fall flat on its face. It seems to work in Fenix lights, but maybe small li-ions are best for these lights (although pricey and not as easy to work with). I guess that if you want more light than what a Fenix (or equivalent) can dish out, then a larger form factored light is needed. I agree that it could be a useful driver. Not everybody has the access (or money -me) for li-ion. I rely on NiMH AA's for nearly all my lights. It is a very common battery.

Well, enough with my ideas and all my ranting. Good luck with developing new circuits. Just let your imagination run wild! I will be thinking about all this for a while and this may not be the last you will hear from me. :devil:

-Tony

 

[dat2zip]

paulr,

It's really tough and very difficult to make a good efficient single AA driver. Losses in trace resistances, inductance wire losses etc are small, but, percentage wise a large percentage of the total battery voltage and thus make for a very inefficient driver.

An interesting point is to look at the latest high efficiency LEDs like the Rebel and couple that with an efficient driver pushing the LED at say 100mA. For a single cell configuration like a keychain light would still be very bright and make the driver more reasonable to make with reasonable efficiency.

havand,

That's definitely possible. I'd have to say there is some difficulty in the configuration depending on battery technology and that could lead to a myriad of configurations. I could be wrong, but, I think the vast majority would still use primary cells over rechargeable.

Since I can only work on one project at at time I think the best bang for the buck in terms of my time and investment will still be GDuP for now and a simple 3 level brightness. After that depending on what comes out of this discussion or be my second choice in terms of the priority list above and that would be the hammerhead.

What is so special about hammerhead? Well, I'm glad you asked. What will be unique with Hammerhead is that like Shark will support boost driving multiple LEDs and have variable brightness. It will maintain a true white balance over the full brightness control range from dim to full brightness. The other nice feature will be the mechanical mount arrangement will ensure optimum heat transfer from the converter to the heat sink it will need to be mounted to as compared to the Shark which is not optimimal configured to get the heat away from the PCB.

As the uP tasks become more difficult the effort rises almost exponentially to develop/test and troubleshoot the code to make it error free.

I would say the three level driver for the Shark is the simplest of code with no feedback control. The Shark only generates three fixed levels of approximate brightness. The GDuP will servo the actual LED current so each boards behavior will have the same LED current. This feedback loop is more difficult, but, I believe I have most of the code in bits and pieces from other projects so it should go pretty fast.

As for battery voltage monitoring, charging, detection would all be new code and new code development takes the most time.

Anyone that is interested and has time to develop the uP code, is fluent in Atmel AVRs I would be more than happy to talk to you. Doing the hardware or providing prototype boards doesn't take that long. Getting the code to run correctly takes a lot more time and effort.

Wayne

 

[LED Zeppelin]

Wayne, they all seem like great products.

Can you provide some details on the LED clock? Sounds interesting.

 

[dat2zip]

Wayne, they all seem like great products.

Can you provide some details on the LED clock? Sounds interesting.

You can read a little more about it here.

Some of my older projects that I posted some time ago here.

Wayne

 

[MarNav1]

GduP I think would be great. Would you have 3 stages at the tap of a switch and no two stage tailcap? Would they be able to run a Seoul emitter? I only ask because I have several Aleph 1 and 2's and a couple A19's as well and I like the Seoul beam pattern better than a Cree. I'm not familiar with the Rebel.

 

[mahoney]

A "donut" shaped board to fit in the old ARC LS heads, so the drive components would be around the emitter, like the stock boards. Something sized to fit in the Rev. 2s would also work in the Rev. 1s. It could be as simple as a voltage boost board, possibly with a couple/3 levels, or a buck/boost style to also allow Li-ion use. Given the limited Z-axis space in an LS, it's hard fitting driver boards in. It could also be usefull in other mods where flashlight length is limited.

 

[dat2zip]

A "donut" shaped board to fit in the old ARC LS heads, so the drive components would be around the emitter, like the stock boards. Something sized to fit in the Rev. 2s would also work in the Rev. 1s. It could be as simple as a voltage boost board, possibly with a couple/3 levels, or a buck/boost style to also allow Li-ion use. Given the limited Z-axis space in an LS, it's hard fitting driver boards in. It could also be usefull in other mods where flashlight length is limited.

I believe the donut hole puts a severe restriction on the implementation. The original Arc board was fortunate to use a transistor based driver where component layout is not critical and components can be spaced around the perimeter of the LED. Almost all the newer more efficient drivers are based on CMOS IC fabrication technology and have a faster switching method and are very sensitive to component layout. Almost all CMOS based designs are near impossible to layout correctly on a donut style hole board as the inductor, input cap and output cap all need to be adjacent to the main IC and that's impossible or near impossible given the perimeter width of the donut board. The Z axis restriction all puts a serious cramp on design implementation. I know the Badboy design can be implemented in the donut shape, but, I don't think there is room to add a microprocessor along with it.

I have a couple of questions.

What would you consider the max drive level for this design? What LED works in this? Won't some LEDs require new optics?

Wayne

 

[havand]

Sorry, not to go off topic, but do you have any idea when you might be able to bring out those hammerhead converters?

 

[jch79]

GDuP. Take the GD buck/boost and add a uP on it. The UI I would propose is a simple on state timing with three levels (low, med, high). The current thought would be to put the microprocessor on the backside of the GD. This additional height along with a dedicated blank anode contact board would still be mountable in the current Aleph Light engines allowing the three level UI to be upgraded into existing Alephs or E series lights.

:thumbsup:

Could the user set/tweak the brightness each level, ala the SPY005?

I find that the average "low" level is not low enough for my tastes, which is one reason I LOVE the SPY005!!

john

 

[mahoney]

In the ARC LS, assuming you are not using a custom battery pack, the max voltage available from primary cells is 3 and anything above 600 ma to the LED would not stay in regulation very long, especially with Badboy style circuitry. If you are using li-ions, you could go higher with the current to the LED, but thermal management becomes an issue. IMO 750 ma with a Cree or Seoul is more than plenty bright, 500 ma is quite bright, and I have modded some rev 0.9 LSs with Seouls and they are still "wow" bright with the stock drive level of about 350 ma.

Of course everybody will have a different opinion about this, but I don't see a need for more than somewhere between 500-750 ma drive current maximum.

I've put Seouls, Edisons, and a Cree into ARC LSs. The Seouls and Edisons play well with the McR18S and the McFlood, both of which are a drop-in optic replacement in the rev 2 LSs and a fairly easy cut down to fit with a thin lense in the Rev 0.9/1 LSs. For the Cree I used a McR19 F cut way down to fit in a rev 2 LS.

(Here's my plea to bring back the McFlood. You would not even have to plate them, I find they work better with the nickel polished off)

Perhaps stick to the Lux/Seoul LED shape for simplicity, although who knows if the latest and greatest LEDs of the future will continue to come in this shape. Crees & Rebels are a bit of a pain to deal with if they are not on a star already. Unless of course you can make a circuit that could fit on the star around the LED. Then the mod would consist of trimming the star to the correct diameter, wiring the board to the contact pads of the star, and gluing the driver board to the top of the star around the LED. If you could leave a center hole that would fit a Seoul, a Cree, a Rebel, it would cover almost anything someone wanted to do...

 

[dat2zip]

Sorry, not to go off topic, but do you have any idea when you might be able to bring out those hammerhead converters?

Hammerhead is in the queue, but, not active. I can only work on one major project at a time along with all the other tasks that I have ongoing.

Unless the priority list changes the GDuP is the active project and could take several months to see completion. That would put the next project start in the November/December time frame with a possible completion early next year some time.

Wayne