Hotwire Regulator Design Collaboration Project

[Alan B]

...
I'm happy to give out my source code, but on one condition: I want first shot at selling it. I'm happy for people to build their own or modify my code, just no selling. ...

Thanks for the offer. Unfortunately, the conditions are incompatible with the goals for this collaboration, so I cannot accept. I wish you the best of luck on your project, and I hope you do proceed. I look forward to seeing your product on the market, and I believe that having a couple of products in this area would be good.

 

[radellaf]

Re: Hotwire Regulators

1.2 or 2.4V. 2xAA is just my favorite size (and running price).

 

[cy]

Re: Hotwire Regulators

AWR has some well earned animosity from a load of folks. besides taking thousand of folks hard earned $$$ and not delivering.

his activities had a chilling effect on other budding hotwire regulator mods.

instead of delivering product, AWR was all time announcing some new latest greatest product.... and of course collecting more $$$ from gullible folks.

the list of regulated hotwire projects, killed by AWR's announcements was complete. almost all competition was killed off, resulting in shortage of regulated hotwire boards that exist today.

correct me if I'm wrong, but willie hunt's PWM hotwire board as used in JS's M6R addressed all issues mentioned so far. problem is willie hunt is not interested in selling individual boards

willie hunt's design might be a good place to revisit.

Jimmy & Alan, here is the link to his thread with photos. Given the animosity towards AWR, I don't really want to get into presenting specifics of his design....but you can follow the links of his first post to more specifics such as this. He has since incorporated various Inrush Current Limiters (i.e. NTC) to prevent the high watts from blowing his FET.

 

[Alan B]

Re: Hotwire Regulators

...
correct me if I'm wrong, but willie hunt's PWM hotwire board as used in JS's M6R addressed all issues mentioned so far. problem is willie hunt is not interested in selling individual boards

willie hunt's design might be a good place to revisit.

I have read the entire threads that you have quoted, and that was AWR's linear regulator.

Thus far I have not found Willie's threads. I did some searching for them but did not find them. Any links would be appreciated. Is Willie still around??

-- Alan

 

[cy]

Re: Hotwire Regulators

do a google search for willie hunt , he's a designer for surefire. his home page should pop up. willie hunt has no threads on cpf.

black diamond headlamps has used his boards.

also do a search on CPF for M6R by JS

willie hunt will do custom batches of boards to customer's spec's.
that is if you can get him to return your phone call.

I have read the entire threads that you have quoted, and that was AWR's linear regulator.

Thus far I have not found Willie's threads. I did some searching for them but did not find them. Any links would be appreciated. Is Willie still around??

-- Alan

 

[Ctechlite]

Re: Hotwire Regulators

I've got a few of Willie Hunts LVR (3D variety) if you want to have a look at them to see if you can figure out how they work. I am working on incorporating them in a torch but have stalled due to lack of readily available parts. Custom for me right now is an obstacle so they are just waiting on my work bench.

Some Willie info...

 

[That_Guy]

Thanks for the offer. Unfortunately, the conditions are incompatible with the goals for this collaboration, so I cannot accept. I wish you the best of luck on your project, and I hope you do proceed. I look forward to seeing your product on the market, and I believe that having a couple of products in this area would be good.

In what way are my conditions incompatible? It is still an open collaboration, and people can still build their own, the only difference is that instead of you getting to sell it it would be me. I'm just kind of annoyed that I've given everything away which will allow you to get a product to market faster than I can, making all my efforts amount to nothing.

What sort of time frame are you looking at? Even though I have already designed, built and tested my regulator it will take at least two months before I can go into production. This is because my design is currently all through-hole using a vero stripboard which is fine for personal use, but not really something I can sell to others. For that I need a professionally made surface mount PCB which, living in Australia, will take a month to get here for the prototype (from batchpcb), and assuming it all works another month for the production run to arrive. In reality it will probably take more like three months once construction and testing is taken into account. If you can get something up and running earlier than that then there is little point in me continuing.

 

[Alan B]

In what way are my conditions incompatible? It is still an open collaboration, and people can still build their own, the only difference is that instead of you getting to sell it it would be me...

What sort of time frame are you looking at? Even though I have already designed, built and tested my regulator it will take at least two months before I can go into production. This is because my design is currently all through-hole using a vero stripboard which is fine for personal use, but not really something I can sell to others. For that I need a professionally made surface mount PCB which, living in Australia, will take a month to get here for the prototype...

By the way, what is your name? You don't sign your posts. It is nice to have at least a first name.

I don't intend to sell anything. My intent, which I tried to state in post #1 (which I have updated slightly a couple of times), is to facilitate an open design and software. I expect that one or more persons will build these and make some profit in doing so, and hopefully contribute a little something to CPF from those sales (perhaps 5%?), and I will be able to buy them for my own use. You are welcome to be one of the builders of our group design. I hope JimmyM will also be interested in building them. If no one at all builds them I might consider building them myself, but I don't have the time at this point and don't expect to anytime soon. If I have to do my own, I would probably put the PCB layout files on the web and folks could use the layout.

You could sell hardware of our design and put your secret sourcecode on it. You are way ahead of us there.

One thing I'm trying to end here is the exclusivity of making regulators that has resulted in limited or no availability. I also want something that is user programmable so it can be tailored for different applications. As far as I can tell, zero products on the market or promised will do that.

The timeframe is not really up to me. I plan to publish a schematic within a week or so, and then we'll see what happens. I don't know what the situation is in Australia, but here in California I could layout a PCB and have prototypes in my hands in under a week for something with so few parts.

In terms of the design, I'm not planning to change it from what was stated. I did the algebra and came up with duty cycle = Vrms^2 / Vbat^2, as you had pointed out. No getting around the algebra. I still plan to use a feedback loop as described. I won't worry about the details of the calculation until I get farther along. I'm familiar with integer / floating point calculations and tradeoffs in microprocessors, I've been working with them since 1974 (and boy were they primitive). What I'll do there is see what the compiler generates and decide if I need to improve it or not. With the 8k program space it may not be necessary to work as hard as you did to get the algorithm in. Later, as the software is improved and more features are desired it will likely be required to drop more code into asm, or tweaked C code optimized for the environment.

As I plan to put the software into the public domain as well, I expect others will take it up and make improvements. Perhaps I won't end up writing the code at all - perhaps someone will volunteer to do that. We'll have to see how it plays out.

-- Alan

 

[Mr Happy]

If you can get something up and running earlier than that then there is little point in me continuing.

I really don't see this. This is a kind of enthusiast/hobby environment where people will buy stuff for the enjoyment of it. It can't hurt to have choices out there. If there is more than one offering available, people may well buy samples of all of them to compare.

It is not a business enterprise that is going to make anyone a millionaire. We are talking about a bit of pocket money to help defray the cost of parts and development and make it a less expensive hobby.

You should also realise that it's a big bad world out there. If you publish any of your designs, you really have no legal way to stop anyone manufacturing them and selling them as their own work. I'm not saying anyone here would do that, but once published anyone with access to Google can find the information. If you really want to keep things for your own benefit, you will have to keep them confidential. I'm afraid that's just the way the world is.

One thing I'm trying to end here is the exclusivity of making regulators that has resulted in limited or no availability. I also want something that is user programmable so it can be tailored for different applications. As far as I can tell, zero products on the market or promised will do that.

From what I read, this is true. I don't think there is anything secret about the kind of design and technology involved here. There should be options available for the people who aren't in a position to design and build their own, and if there are competing designs out there rather than none at all it can only be a good thing.

Every time someone says "X is doing it, so I won't bother," the world loses a bit of richness.

 

[That_Guy]

While there is no reason why I can't continue to develop and sell my design alongside this collaboration I really can't see the point. Choice and competition is great in the commercial world, but for hobbyists on a small budget it isn't so good. There is little incentive to spend any significant amount of money developing or improving something if there are no guarantees of being able to recoup the costs because there is already something similar or the same on the market.

In my case if I were to continue the only costs would be transitioning from a through-hole design to surface-mount. It might not sound too costly but the reason I've stuck with through-hole this whole time is that surface-mount is expensive in Australia. With through-hole I can order all the parts locally and use vero stripboard instead of a PCB. With surface-mount I need a proper PCB. Using batchPCB, which as far as I know is the cheapest for single prototypes, I'm looking at around $20 and a one month wait. There are faster options, but they are even more expensive. The most costly thing is the parts. Most need to be ordered from overseas, and Digikey charge a minimum of $30 to ship internationally, so it can get expensive quickly, especially if you realise you need another part after you've placed your order which almost always ends up happening. So to sell my design I'll probably be looking at spending a minimum of $100 on top of what I've already spent. Since my costs are much higher I would also have to sell my design at a much higher price. Since it isn't really any different to what's going to be designed here I doubt many people would be interested so it is unlikely I would be able to recoup my costs.

While talk of profit always sounds evil next to talk of open source projects, it really is necessary to drive innovation, even amongst hobbyists who aren't in it for the money. The money made from projects goes beyond just recouping the costs spent developing it. The money I was planning on making was going to be reinvested back into the design to make it better. One of the first things I would have done would be to purchase a True RMS multimeter to allow me to properly measure the output voltage which would allow me to improve the accuracy of the regulator.

Open collaborations like this are good because they allow people to contribute and make improvements without having to bear the costs of doing such things themselves, and there's nothing stopping me from contributing. In fact I would be happy to because it doesn't cost me a cent. However there are limits to this. With open source software the only investment needed from the developers is time. But this is hardware, and hardware costs money. Sooner or later someone is going to have to pony up the cash and that person is going to be less willing to spend much money if they can't get any of it back. They may be happy to spend a little bit, but to make more dramatic improvements in a design it often takes a lot of money and the only place that money is going to come from is profit. The fact is profit drives innovation, without it things will stagnate.

Despite seemingly being motivated by profit it doesn't mean that I want to keep everything a secret. I like the idea of an open source regulator. My plan all along had been to do something similar to what you hope to achieve. I was going to announce my regulator, with the source code, schematic and PCB layout all freely available, along with a detailed description of how it all works and instructions on how to build your own and modify the code, hardware and PCB layout to your own requirements. The only difference was that I wanted exclusive rights to sell it so that I would be guaranteed of making a profit to both recoup my costs and fund further development. This restriction was only going to be for as long as I sold it, so that when I stopped other people could take over and CPF wouldn't be left without a regulator.

I was really hoping that since our goals were so similar that I would be able to convince you to settle for what I had already designed and let me make some money selling it, but it looks like that's not going to happen.

Since the only point of restricting the sale of my design was to prevent any competition, it doesn't make any difference to me if someone else sells a regulator based on my design or your's. I am therefore willing to give out my source code with no restrictions. However there probably isn't much point since you want to do it in C. The only reason I used assembler is because I don't know how to program microcontrollers in C. You could probably write a program in C in less than 15 minutes which would be easier than trying to understand what I've done.

Despite the impression I may have given you there really isn't anything at all special about my source code. It really is quite simple. It's only 226 bytes long, doesn't use any RAM and isn't at all optimized. What's special is the equation. That's the beauty of a microcontroller based PWM regulator – its elegance. All it is is that one little equation. Very simple yet very powerful. My main piece of advice would be to drop your idea of using a feedback loop and just calculate the duty cycle directly. As far as I can see there is absolutely no benefit to doing it your way. Apart from slowing down the response it ruins the elegance. As far as I know my method is the same used by everyone else: Willie's LVR, winny's PIR and the Lightbrain. I'm still a novice when it comes to electronics so if there is an advantage to your method I'd be interested to hear it but at the moment I just can't see any point.

To design the regulator in C all you need to do is configure the timer for PWM on one channel, and then configure the ADC to be triggered by the timer on the other channel. You then use the ADC interrupt to calculate the duty cycle. You use an if statement to check if Vin < Vout, in which case the duty cycle is 100%, and if it isn't you then use the formula to calculate the duty cycle. Simple as that. Are you still interested in my code?

Regards,
Tim

 

[Akarusa]

There is little incentive to spend any significant amount of money developing or improving something if there are no guarantees of being able to recoup the costs because there is already something similar or the same on the market.

Tim,
While I can understand wanting a reasonable probability of profit before investing, nobody in a free market gets a guarantee that they'll recoup their costs. Profit and risk go hand-in-hand, and there is no practical way to close this market for regulators in order to break that rule.

I really can't see what you expect to gain from getting some kind of exclusivity agreement. Even if someone were to agree to your terms, once your code is publicized, any 3rd party who wasn't part of the agreement could do whatever they want with it. Short of copyright or patent protection (and their attendant costs), that's just the reality of things.

And if, as you contend, your simpler design is more elegant and efficient than the one discussed here, wouldn't that alone create a separate market niche for your product and partially insulate it from competition?

 

[Alan B]

Here is a pencil draft of the design so far. Sorry I have not had time to make a pretty schematic. In essence, it is not much different from the LightBrain linked above. I included a capacitor on the ADC input but that may not be required or even desirable, but it can be left unloaded. Many details regarding i/o and programming pins are different for this CPU. About 12 parts including PCB and programming connector.

http://akbeng.com/flash/hwreg/attic/20080118 loside.pdf

Jimmy proposed a regulator, I did not investigate to see if it requires the support parts shown. The FET needs to be selected, a logic level gate N channel in the 30V, 30A, 5 milliOhm region.

Form factor would seem to be circular to fit on a KIU socket. The FET might heatsink on the socket adapter, perhaps. Some access to the programming jack should be provided for development and future software or parameter changes. Of course a prototype does not have to fit, but with this few parts I might go for a fitting prototype...

-- Alan

 

[JimmyM]

That's nice and simple. I see the voltage divider you have there would be sensing battery voltage. Be careful of that capacitor though. When the mosfet turns on, it's going to dump when the battery voltage sags. The resistors will have to be pretty large value. 1M on the high side, and 100K on the low side. I like the simplicity. I'll do some looking around for logic level mosfets.

OK. IRLR6635 Int Rectifier, 30V, 32A, 2.4 mOhm @ 4.5V. DirectFET MX, 6.4mm x 5.1mm footprint. Odd package though. It would require oven reflow soldering.
or
IRLR7843 Int Rectifier, 30V, 100A, 3.9 mOhm @ 4.5V, DPak package. 6.7mm x 10.4mm footprint. Slightly higher gate charge would slow down switching speed, but not very much.

 

[That_Guy]

While under normal circumstances I would love this project it's just disheartening for me because I had already gone so far down the path of designing something so similar myself, and now I'm going to lose my baby. Despite seemingly being motivated by profit it was never meant to be a commercial venture to make me rich and I never wanted to compete with anyone. I was just wanting to fill the void left by winny, the same as this project, and hopefully make a bit of money to fund further development and buy better equipment. If I had announced my idea earlier I doubt this thread would exist, it's just that I wanted to wait until I had finished before announcing annoying. Of coarse it's my fault for waiting so long in the first place. If I had built it once I had finished designing it 8 months ago instead of sitting on it there wouldn't be any need for this project. Sure others would eventually have made commercial units and gone into competition with me, such as AW with his commercial Mag regulator, but it wouldn't matter because I would have already made enough money off the initial period of exclusivity that I could either drop my prices or stop production and let someone else take over with my design.

The fact that my design is simpler and more elegant won't make the slightest bit of difference because consumers won't be able notice and won't care. It's really just an intellectual nicety. But as I've said I don't want to compete.

I'm happy to contribute as much as I can to this project. At least that way all my efforts go completely to waste and I won't feel as bad.

Does anyone know of any good guides on the internet for programming AVRs in C? I shouldn't have any problems making the program myself once I know how to manipulate the registers.

Regarding hardware the first thing that should be decided is what are the requirements with regards to size and power handling? A small low power regulator could be made to fit inside a C Mag, or a large high power regulator could be built for powering monster bulbs like the 36V 400W 64663 used in the Ken 5 or the 28V 600W Q4559X used in "The Sleeper" and the Larryk14. Something to fit a D sized Mag would probably be the most popular, but since AW is already coming out with a regulator to fit a D sized Mag with the added convenience of a professionally designed drop-in including switch and socket there isn't much point. Of course there can be a number of different versions, but something should be settled on initially.

You also need to decide if it's going to be surface-mount or through-hole. The Tinys are the best choice for through hole because at 28 pins the smallest Megas are too large. However for surface mount this isn't a problem. The 8 pin Tinys are only available as SOICs while the Megas are available as TQFPs so despite the greater number of pins they aren't really any bigger. Both are also available as the tiny 4 x 4mm QFNs but they are much more difficult to solder. I would recommend using the AtMega48 since it isn't much bigger or more expensive, but has a hardware multiplier, more timers including a 16 bit for higher resolution PWM, an AREF pin for reducing the noise on the ADC and more IO pins allowing the option of things like LEDs to indicate battery voltage.

Regarding the choice of MOSFET I've found switching losses to be far more important than a low Ron. Even when switching at only 294Hz with a MOSFET driver and a mediocre MOSFET with a Ron of 5.3mOhms I've found switching losses to dominate at higher input voltages.

The resistors used for the voltage divider can't be very high, the ADC requires an input impedance of less than 10k for good accuracy.

 

[Mr Happy]

While under normal circumstances I would love this project it's just disheartening for me because I had already gone so far down the path of designing something so similar myself, and now I'm going to lose my baby.

I think it's a shame you are disheartened so readily. With no disrespect to Alan, whatever is discussed in this thread so far is vaporware. It may come to nothing. There have been any number of threads like this in the history of discussion forums that have gone nowhere.

I'm sure everyone would welcome your input if you choose to join in and add more of your contributions here, but there is some realism to face up to as well. From what you have said, Australia is not such a kind place to innovate in as the USA. I find that resonates a little since I have met Australians who have emigrated in order to find a more stimulating high technology environment than their home country. If it's as costly as you say to work from your home base, you do face some obstacles that set you back from the start. Is there really no domestic Australian equivalent to Digikey?

If you have built a working prototype already, then I think you have achieved two of the most important things, namely (1) your own personal education and edification, and (2) the satisfaction of designing and building something that works. I think your dismay here arises from building a future vision in your mind and then knocking it down. Remember that right now you have not lost anything except a future of your own imagination. Keep imagining and don't be disheartened! I'm sure you have a long career ahead of you.

 

[That_Guy]

Is there really no domestic Australian equivalent to Digikey?

There's Farnell which has a huge range, but they still don't have the more exotic parts, and despite being in Australia shipping is still pricey at $10. They can also be very expensive, making Digikey cheaper in some cases even with the excessive overseas shipping charges. Despite this I'm actually quite lucky because there's a brick and mortar chain called Jaycar which caters to the hobbyist and has a surprisingly good range. It's great being able to walk in and buy individual components and even though they are very expensive it doesn't matter when most parts are no more than a couple of dollars anyway. The only problem is that they cater a little bit too much to hobbyists - they hardly sell any surface-mount stuff.

I think it's a shame you are disheartened so readily.

It's probably because this is the first time I've ever attempted anything like this. After I had realised that I could actually sell something I had designed I got really enthusiastic about the whole idea and went getting my hopes up.

I think your dismay here arises from building a future vision in your mind and then knocking it down.

Yep, that's pretty much it. Next time I design something or have an idea I need to go out and do it, rather than waiting around for months and then whinging when someone else decides to do something similar.


I'm happy to contribute, it will make me feel useful and better about this whole thing. Is this project going to be done in C or assembly? I would really like to learn how to program micro controllers in C. I can already write basic C programs for DOS, it's just C for AVRs specifically that I need help with. I've been trying to find some guides on the internet about it but I haven't found any simple and clear guides yet. It's so confusing with all the different compilers and linkers and makefiles and things. Assembler seems so much easier, it is nice and simple and straight forward.

 

[Alan B]

I like to use a mix of C and assembler, but if we have a standard hardware platform there can be different codes for it, and some could be C and some assembler.

AVRFreaks website is the place to get the info on the gccavr compiler. I have not used it recently, but there is a package to download and install. It is confusing to get it working at first, once going it is really a snap. There are a few special C idioms that are specific to I/O that need to be learned. And one has to be very careful. For example, doing a floating point anything brings in a lot of library code, so on a small chip it can overflow it. Once in it can be used without much further code growth. 8K parts have quite a bit of space, but can be filled up. A 2k part can be overflowed with a single formula, so no floating point there.

On the processor, my tendency for the first go-round is to stay with an 8 pin part. If we want to make it simple enough that someone can dead-bug it, the megas are probably too fine pitch in the smaller packages. They probably don't have a temperature sensor either, though I didn't verify that.

Good points about the FET specs. If we choose a generic enough package the FET can be easily substituted. I would expect that any logic level gate FET would be potentially suitable. Look in Digikey and suggest something. We're wide open on this.

Another issue is off-current. If we have to make the divider too low in impedance it will draw some current. Also the 5V regulator we use should have very low idle current, or we may have to turn it off. I'd like to keep the parts count low. We should choose an off-current maximum spec. Something like 2500 mAH divided by six months??

In terms of a space requirement, my own needs at this time are for D Mag, which would be into a KIU socket base. Perhaps fitting in a KIU C socket base would be more universal, however. I have neither at this time, so someone could help here by putting the available size of those two environments. Are there other packages that we should consider?

On the AW regulator, I don't know for certain what it will do, or when it will be released. It looks very nice physically (the non regulated version). I think the differentiating factor that makes me interested in this collaboration's version is that it will be more flexible than one from any vendor with closed hardware/software.

If you look at LED regulators, there seems to be plenty of market space for many units. Even a single vendor offers almost a dozen variations. This market is possibly smaller, but don't forget there are many cavers, etc who really prefer incans and not necessarily the super power models, but more practical long running stuff. At least one designer will only answer the phone for 10,000 units per order!

I am interested in a fairly basic hardware design, but perhaps a phase II would be much more sophisticated. How about a zigbee RF interface, so you could control your flashlight and set its parameters as well as read out battery status, etc from a distance. Save that for later.

I am going to Quartsite Quartzfest tomorrow and I need to get the RV packed. Any CPF'ers going should look me up. We'll be with the Ham Radio group at mile marker 99. I may have internet, or not.

See you folks in a week. Play nice!!

-- Alan, WB6ZQZ

 

[JimmyM]

That Guy,
There's nothing wrong with having more people at the party. AW has a fancy soft starter with 3 levels in a custom switch body. I have my JM-SST. It's just the board and sells for 1/3 the price. They're both soft starters and both are selling.
I'm happy to help this project in what ever way I can, but it won't stop my development on a PWM regulator of my own. I can't do PICs or AVRs, so I'm going with what I know. A PWM controller and a P-channel MOSFET. If your design is better and you want to release it on your own. Please do. It will have it fans and detractors as everything does. But it will be your design and your baby.
I kind of felt the way you do when AW brought out his soft starter before I announced mine was ready to go. My mine still sells well for a couple of reasons, it's cheaper and super durable.
Keep at it!

For my part, I'm still in the hunt for the "perfect" PWM controller.

 

[petrev]

Hi All,

No help in design terms here I'm afraid - just a bodger.

Had a PIR (or 2) - Loved it but it broke . . . Winny Off-Line.

AWR-HD broke the fragile little trim pot and never got it back from AWR.

I tend to buy multiple different things to try them so . . .

Make mine nice and robust with provision for an external volume pot - love the JM-SST for that and looking forward to Jimmy's regulated version soon.

Have to keep my eye on this one - looking forward to the results . . .

Cheers
Pete



ps. Kiu C-Adapter space is really quite small 17.9mm Dia. by 6.9mm deep

D-Adapter 30.4mm Dia by 7.0mm deep

there are quite a few places where a bit of Dremeling can be done to make extra room for odd components . . .

 

[JimmyM]

Hi All,

No help in design terms here I'm afraid - just a bodger.

Had a PIR (or 2) - Loved it but it broke . . . Winny Off-Line.

AWR-HD broke the fragile little trim pot and never got it back from AWR.

I tend to buy multiple different things to try them so . . .

Make mine nice and robust with provision for an external volume pot - love the JM-SST for that and looking forward to Jimmy's regulated version soon.

Have to keep my eye on this one - looking forward to the results . . .

Cheers
Pete



ps. Kiu C-Adapter space is really quite small 17.9mm Dia. by 6.9mm deep

D-Adapter 30.4mm Dia by 7.0mm deep

there are quite a few places where a bit of Dremeling can be done to make extra room for odd components . . .

I have a design that kind of works off the JM-SST design basics but does regulation. Now I'm in the process of trying to find chips that do more than one thing at a time. I found a chip that is an oscillator, comparator, error amplifier, and a high-side NFET driver. I think it's been discontinued though. AAARRRRRRGGGGGGHHHHH. I'll be breadboarding this weekend.