Hotwire LDO driver! 13mv drop with 1185

[NewBie]

can fit 9x 17500 in a 3D... you can fit 12 14500 in a 3D.. now that's a very interesting solution.. See.. with my LDO.. putting 1185 into a 14.4V solution is horribly inefficient.. oddly enough even so it's just as efficient overall as the 1154 which runs at 14V but is far less efficient in lumen/watt..

That would be a most excellent use of the switching driver.

What you can fit into that host is 12xR123 for 14.4V... you could also put in 12X NiMH... also for 14.4V.. at 35W and say 95% efficiency.. means you need.. only 2.55A from the cells vs 3.35.. that is not insignificant a difference.. very interesting though there.

In addition.. it would be nice for the primary based solutions like running the mag 85 from 12x 123 primaries.. which would be 12V no-load but 10V under load.

I will definitely be exploring 'version two' of the hotdriver in the form of a switcher.

-awr

Speaking of the 12 R123 cell configuration, and your 12.5V bulb at 2.05A, do you realize you'd only need to pull less than 0.6A out of the cells? There would be a tremendous efficiency increase in the batteries alone, with a switcher vs. a linear.

Hopefully I can get you interested into researching it more for your products. There are some really nice improvements you can make with a switcher in those configurations...

 

[andrewwynn]

yes i did realize.. actually i already knew all this.. just wasn't going to happen for 'round one' because of the simplicity of the LDO solution and that i already within 2 days had a schamtic and within 2 weeks had an operational breadboard model.

Most of the designs the LDO is intended for are basically only for over-voltage on fresh cells protection.. my brother just blew the second 1185 bulb within 1 week from freshly charge CBP 1650s.. i'm thinking it could have something to do with the colder weather? (nothing else changed).. I would put the LDO ckt set for 10.8'v into the 10.8V hosts for exasmple.. then for the majority of the use.. the efficiency is like 98% or more.. that's what this circuit was designed for.. being able to use a bulb that will handle nominal voltage just fine but not the overvoltage:

prime example: [email protected] which likes to blow at 12.6.. .1331.. same scenario, but even more flashable.

the 14.4 host to run the 1185 with a switcher is a very interesting option.. because the bulb can be pushed to a constant 1300 lumen vs the 1200 and dimming you get from 10.8V.. but with 3.39A at 1316 lumen with the 1185.. (11.0V).. that means.. 39W on the input side of the switcher.. divide that by 14.4V and get only 0.9A from each stack..

What's it all meab Basil? (austin powers quote)

Means that running an 1185 from a 14.4V host can get you 96% vs 76% efficiency which would convert to a runtime improvement of 47 minutes vs 37 minutes with the same exact bulb and batteries, only using the switcher you propose vs the LDO of my current design..

As you can see i'm aware of the times when the switcher kicks the snot out of the LDO, and haven't turned my back on it.. who wouldn't like a mag 85 that stays bright like that first juicy minute or two.. but also won't blow your bulbs trying to run it at 12.6V!

Oh....i think i mentioned it above.. but i was able to witness an 1160 bulb operating at 8V! (i think that calculated to like 1300 lumen!).. of course only for a fraction of a second but it was one BRIGHT fraction of a second!

-awr

 

[legtu]

How about using a microcontroller?

A couple of caps, a microcontroller, FET, regulator and resistor(s)/voltage ref should do it. The regulator takes care of the vin to the microcontroller. Microcontroller takes care of the voltage regulation via PWM, undervoltage shutdown/lockout, soft-start, etc. It can even be made to 'auto-sense' the vin and adjust accordingly to a pre-programmed bulb parameters. The circuit itself should be fairly simple but the 'complexity' goes into programming though...

Efficiency should be fairly high regardless of vin right?

 

[andrewwynn]

There are people working on µC hotwire drivers.. it's a really neat idea to be sure... i like the idea of the auto-sense for the Vin.

-awr

 

[Bimmerboy]

Hope nobody minds me jumping into this highly technical thread for a moment, but I just discovered this thread and skimmed through it some. I gotta' keep an eye on this.

Just so I don't have to try to understand everything going on here, could I ask for a quick and dirty explanation on what possibilities this holds for the 1185?

One thing I'm wondering is if an eventual driver would fit in with my 1185's present battery scheme of using a 9AA to 3D holder with NiMH.

 

[cue003]

is this mod ready to go into lights... namely my lights...

I would love to get the full potential out of the hotwire mods I already have.

Curtis

 

[andrewwynn]

the main deal is that you can push bulbs harder w/o worrying about instant flashing them.. for example.. 9x17500 cells is nominal 10.8V which is great.. 1234 lumen.. but at 12.6V.. (what fresh LiONS x3 will be).. it's a barnstorming 2052 lumen.. but the CCT is 0.9% from the melting point (though i understand that that is not 'exactly' the melting point)... it compares to 5.3% away at 10.8V....

I've blown 2 1185 bulbs in the past week.. one from 1/2 hr aged and one from 3hr aged CBP 1650s.. charged as always with the batteryspace smart charger which tops them off to 12.4 or 12.5 just as expected.

the 9AA->3D is exactly the host i blew my two 1185s using.. Oh i blew another one in my 2 1/2D host

I wouldn't expect 'quantities' of this driver before the first of the year.. but dec. is a possibility.

-awr

 

[NewBie]

How about using a microcontroller?

A couple of caps, a microcontroller, FET, regulator and resistor(s)/voltage ref should do it. The regulator takes care of the vin to the microcontroller. Microcontroller takes care of the voltage regulation via PWM, undervoltage shutdown/lockout, soft-start, etc. It can even be made to 'auto-sense' the vin and adjust accordingly to a pre-programmed bulb parameters. The circuit itself should be fairly simple but the 'complexity' goes into programming though...

Efficiency should be fairly high regardless of vin right?

The major problem with the uC with MOSFET switch is that you are still whacking the bulb with the full voltage, even if you go fast. A uC with MOSFET switch doesn't regulate the output voltage at all. It may achieve an average current, but that is done by turning the MOSFET off and on. Thats what PWM is, Pulse Width Modulation. You adjust the on time to off time ratio to achieve a different average current.

Additionally, you are still drawing the power out of the batteries at full current, during the time you are whacking your poor filament with the PWM pulse. So you don't pick up as big of gains for efficiency inside the cell.

The other problem with PWM, is that you can't go too high with the battery voltage.

Yes, it may be better than nothing, but highly doubtful you could push the bulb as far as you could with a linear or a switcher.

You can get the uC and MOSFET PWM to not stress the filament as much by running the PWM at a higher frequency, but in doing so, you burn more power switching the MOSFET off and on, and also burn power in the MOSFET during the transistion time.

But it is possible to implement a switcher with a uC, by adding an inductor and capacitor. Henry of HDS did this a few years back with the ARC4.

 

[innerlight]

I'm in for one of these drivers when they become available

Rick

 

[glire]

This thread wants to make me learn stuff I don't know at all... yet (hopefully)

Newbie,
Can you point me to a 'modern' boost switch regulator example (shema, parts, actual experiences, etc).
Thank you.

 

[andrewwynn]

Very exciting news!

the 'shrinking hotdriver'.

the prototype! a few bugs went away going from the breadboard to the prototype board! now it works exactly as designed.. not a tiny bit of voltage swing depdending on load anymore.. you can set it with no load.. and the voltage will be the same with 4A.

Here's the first operational 'in the light' hotdriver.. I put it in a 1D host from fivemega, using (4) 14500 cells for 14.4V.. and running an 1166 bulb at 12.48V.. 909 bulb lumens.. output is absolutely constant 'til the last 30-40 seconds where you can just start to detect it's dimming before it shuts off.

the rest of the slideshow is here:
shrinking hotdriver

 

[StoneDog]

Standing by w/ PP!

 

[wquiles]

Andrew,

Like StoneDog, I got Paypal ready - I was going to order parts and try to build one, but I rather pay a little more and get the module from you

Let us know when these will be ready

Will

 

[savumaki]

I've been following this for sometime and I get the same feeling I got when I was waiting for my firstborn to come along.

I'm not sure I can afford him but here he comes :goodjob:

Karl

 

[andrewwynn]

I have parts lined up.. the plan is as soon as nano production is done to crank out about 20-25 of these complete kits.. just drop in a maglight with a 5/64th hex put in your bulb and your battery pack and turn it on.

Not exactly sure what the total price will be but it will be reasonable for what it is... considering the ludicrous cost of high-power lights 'in the real world'.. like the M6 battery holder for $90.. holy cow!.. get a much nicer one for $35 from Fivemega!

-awr

 

[greg_in_canada]

I hope you wore a faceguard when you powered up the breadboard prototype to protect you from the naked bulb (possibly exploding).

Greg

 

[andrewwynn]

I use a piece of welder's glass.. you can see it in some pictures.. sadly i broke it so i need a new one.. so today and lately when testing my 'faceguard' was my forearm... does a decent job of knocking down the directed lumens too.

We got what we are now calling the "Mag100R" fully operational last night.. 3000 estimated torch lumen.. basically the same as the USL but doesn't dim... it is a sight to see it here (beamshots comparing to a mag85).

 

[vacuum3d]

Andrew, Great stuff. I've been following the thread as well. Where does the line start?
Can this driver be easily adjusted to handle different bulbs and battery voltage input?

Thx,
ernest

 

[andrewwynn]

voltage is adjusted by an accessable trimpot.. anywhere from like 1.5V to 30V is possible.. Vbulb has to be reasonably close to Vbat... but the FET can easily handle 10W or more.. it is possible to run an 1185 at 11V from a 14.4V host.. just not very efficient.

-awr

 

[cue003]

deleted....sent PM instead