P7 Mag with high efficient converter and variable output

[sector_cleared]

Hello Candlepowerforums :wave:

I am leeching here for a while now and was waiting to finish my mod to present it here. Some threads really inspired me and made me to the flashaholic I am

Enough of that. Now to my mod. Basically it is a standard Maglite mod but with some extras.

Parts:
4C Maglite
P7 CSWOI
UCL Lense
Selfmade heatsink
Selfmade synchronous step-down converter board
10kOhm potentiometer with integrated switch
3x 18650 Trustfire 2500mAh

Some facts:

Adjustable current between 100mA - 3A
Runtime between 75h (@100mA) - 2,5h(@3A)
Output should be somewhere around 700-800 Lumen @3A
Efficiency is 95% (real 95%, not like Kaidomain oder DX 95% )

This is my selfmade heatsink. Nothing special here...

I wanted to have a variable output for my Maglite but I didn't like the solution to drill a hole in the Mag-head or body and attach a potentiometer. So i decided to use one of this potentiometers with a integrated switch like the one which were used in car radios a while ago The good thing is I have only one button for switching the light on and off and for adjusting the output. The bad thing is...it's a PITA to get the potentiometer in the switch housing. :hairpull:

Lower part of the switch housing. Abused with Dremel, file and a jigsaw

With inserted potentiometer

Upper part, same story

And the whole thing put together

On the left side you see the heatsink with attached P7 and my driverboard. The potentiometer is already connected to the board. Next thing that happens on this pic is that I use a IRF1104 MOSFET to bypass the switch. It has an Rdson of 9mOhm. There is enough reserve for higher currents
(In the background you see the the ciruit diagramm of the switch)

Here is a close-up of the switch assembly. You see the MOSFET attached to the upper part of the switch. The red wire is the current path from the batteries.

Now i soldered everything together and pushed it in the handle. Instead of the push button i have a rotary one now.

In action

I guess most of the more advanced modders here ask themselves now: why is this guy not giving more information on the driver... ok ok here we go.

I started this project when the first P7 samples where availiable (yes, took me quite a while). I got my hands on 2 samples and was wondering how to drive this babys. I dropped the first one in a 3C Mag DD which worked out pretty well. But since I am a electrotechnics engineer and know how this stuff works I wasn't feeling to good about direct-driving a LED. And I wanted an adjustable output.
While the time passed by, Kaidomain and some members came up with drivers or driver mods but all of them had one thing in common: low efficiency.

The problem with the buck converters which are used normally is the voltage drop at the free-wheeling diode. At high currents like the P7 needs it, you loose 0.4-0.5V. With a CSWOI the Vf @ 3A is somewhat around 3,5V. While free-wheeling you lose 1/7th of the energy in the diode. With that and the other losses (FETs, Circuit) it's barely possible to reach a efficiency above 85%. I bed most of the drivers are around 80%. Ok, enough bashing.
There is a way to avoid most of the losses. It's called synchronous step down converter. Instead of the diode there is a MOSFET which switches when the current is free-wheeling. Due to the very low resistance of the MOSFETs you lose only a few Millivolts instead of 0.4-0.5V. In combination with a good switching MOSFET you reach a efficiency of >95%. To measure the current i am using a 20mOhm Shunt with a current sense amplifier. That gives me a very accurate current reading but only 60mV voltage drop at 3A.

Okay folks, thats it so far. I hope my electronics mumbo-jumbo was not too tiring :sleepy: Now i have to go and shock some ppl with my new toy :duck:

 

[3rdrock]

:welcome:


That is a neat build.

 

[Al Combs]

:welcome:

A very respectable first post. Would you mind sharing some of the details of your driver? By that I mean the specifics. Part numbers and schematics so we can make our own. Or are you perhaps thinking of going into production? If so, how soon?

Is it constant current output, or just a voltage regulator? With 3 amps variable output and 95% efficiency, it sounds like you've got what everyone's been looking for. That's very nice work.:thumbsup:

 

[sector_cleared]

Thanks for the welcome

With 3 amps variable output and 95% efficiency, it sounds like you've got what everyone's been looking for.

Thats the point. I was also looking for a driver like this but it was not availiable so i had to build my own one. The circuit is current controlled. I am using a current sense amplifier to measure the current. Using a voltage controlled driver for a LED is careless in my opinion. :thumbsdow

Would you mind sharing some of the details of your driver? By that I mean the specifics. Part numbers and schematics so we can make our own. Or are you perhaps thinking of going into production? If so, how soon?

Actually I am thinking of going into production with that baby. But first I want to change some parts of the driver. I found a better MOSFET with lower resistance and the inductor ist too big for mass-production (Its specified for 9 Amps. I just used it cuz I had them already). Also the current sense amplifier is quite expensive. With minor changes a normal op-amp would do the trick (Same thing, it was laying around so i used it). Also some other things are to do like optimizing the parts to minimum heigth (as you see I had to cut down my heatsink ) and I want to check how far I can go with input- and output-voltage. At the moment the circuit can drive one LED of a maximum voltage of 16V. 3-4 P7s of a maximum voltage of 20V would be neat. But I dont think that fits in the MagC form factor...but maybe MagD. Lets see

But due to personal circumstances I don't know if I will have the time to do all that stuff. What I gonna do is I'll start doing the modifications listed up above. If I dont find the time to finish it I'll will post the circuit diagramm, the schematics and the part list so everybody can build it on their own. :twothumbs

 

[TexLite]

Great job and first post!

-Michael

 

[Firecop]

Spectacular - thanks for sharing!

Keep us posted on your progress...I'm looking forward to your results.

 

[Visible EM Wave]

...I am a electrotechnics engineer...

Now you can be proud to announce yourself as a flashlight engineer!

Thanks for sharing this and look forward to your high efficiency P7 V2.0

:thumbsup:

 

[PhotonAddict]

Nice work! Thanks for sharing and welcome to CPF!

 

[bluecrow76]

:welcome:

And I'm looking forward to that production driver run!!!

 

[Al Combs]

I am using a current sense amplifier to measure the current. Using a voltage controlled driver for a LED is careless in my opinion.

Using an amplifier to monitor current, nice. Most of the constant current regulators I've seen are really voltage regulators that monitor current across a sense resistor. The best I've been able to find by way of voltage regulating IC's to build my own go down to about 0.6 volts. So if the LED needs 3.6 volts (for easy math), you're automatically down to (edit 85.7%) efficiency before you even get to the regulator. Perhaps even more than the inefficiency is the question of how to get rid of all that heat.

Actually I am thinking of going into production with that baby.

Actually I was rather hoping you would. The diagram you had your parts layed out on showed an inductor, a couple of IC's and several discrete components. More than would look pretty on a breadboard arrangement. I'm afraid etching my own boards is a little more than I want to take on. Do keep us posted and again, great job.

 

[SafetyBob]

Superb craftsmanship!!

I cannot tell you how much many of us would love to make multiple flashlights with this setup. Please just give us the board with the stuff on it and we can modify the switch or whatever else needs to be done. You don't need to make some crazy new switch or anything.....others will probably do that later, but we need the electronics!!!

Everyone is right, this is EXACTLY what we have been waiting for.

Please keep us up dated with what you decide to do.

Bob E.

 

[Aircraft800]

sector_cleared,

Put me on the list for two of your buck drivers!

Most people come here to learn how to mod a flashlight, you already are teaching others! Awesome work! I am using the low efficient diode type buck drivers, and am always looking for someone to make something that fits into a 2 "C" M@g, my favorite duty light.

 

[mitch79]

You can put me on the list for a couple as well.

Question, is it possible to use PWM dimming instead of the pot?

 

[sector_cleared]

Thanks a lot for the feedback

I will hurry to get things done as soon as possible. But for the peace of conscience I have to modify some things to make it perfect. I don't want to cast a damning light on the label: Made in Germany

@mitch79: Sorry, in the actaul configuration it's not possible to use a pwm to dim

 

[astrotec]

really nice build sector:thumbsup:
put me in for up to 4(four) of the bucks when they are ready... depends on cost... thanks again...
a.t.

 

[2xTrinity]

Using an amplifier to monitor current, nice. Most of the constant current regulators I've seen are really voltage regulators that monitor current across a sense resistor. The best I've been able to find by way of voltage regulating IC's to build my own go down to about 0.6 volts. So if the LED needs 3.6 volts (for easy math), you're automatically down to (edit 85.7%) efficiency before you even get to the regulator. Perhaps even more than the inefficiency is the question of how to get rid of all that heat.

Yeah, another irritating side effect of that is that most of the regulators have the sense resistor on the low side (between LED negative and ground), making it impossible to connect the LED (-) terminal directly to a common ground. In builds with a lot of LEDs on a common heatsink, or where the LEDs are separate from the drivers (as in a robot project I was working on recently) this is a nuisance as it requires a separate ground return from each LED.

Adjustable current between 100mA - 3A
Runtime between 75h (@100mA) - 2,5h(@3A)
Output should be somewhere around 700-800 Lumen @3A
Efficiency is 95% (real 95%, not like Kaidomain oder DX 95% )p

The PT4105 (eg the "Kennan" over at kaidomain) is actually a 200mV voltage regulator, measuring the current across as sense resistor. There is only once scenario where it's possible to almost reach 95% efficiency. That is, running the Kennans with 3 LEDs in series off of 4 LiIons in, with output current set to exactly 700mA.

Kaidomain spec'd their boards so they deliver 1A, despite the fact that effeciency plummets above 700mA, so it's necessary to either swap the sense resistor for a larger value one, or apply a constant bias voltage to one of the pins on the IC to accomplish this, as well.

Again: you must have ALL those conditions EXACTLY satisfied to get 95% efficient conversion. current LOWER than 700mA will also reduce efficiency by driving up the duty cycle of the switcher, increasing "freewheeling" losses you described...


Back to the main topic:

nice work on this build. I've been considering doing a project like this where I run a P7 or MC-E with a custom, fully dimmabe driver.

One quick question:

Is the potentiometer an audio taper, so as to make the light output increase exponentially? If not, did you use some other method to creating a non-linear response for your dimmer (ie, an amplifier with non-linear response)?

 

[sector_cleared]

Yeah, another irritating side effect of that is that most of the regulators have the sense resistor on the low side (between LED negative and ground), making it impossible to connect the LED (-) terminal directly to a common ground.

I am using a high-side current sense amplifier and the shunt sits between the inductor and the LED. Would have been perfect for your project.

The PT4105 (eg the "Kennan" over at kaidomain) is actually a 200mV voltage regulator, measuring the current across as sense resistor. There is only once scenario where it's possible to almost reach 95% efficiency. That is, running the Kennans with 3 LEDs in series off of 4 LiIons in, with output current set to exactly 700mA.

Getting high a efficency in a multi-LED setup with currents below 1Amp is easy since the losses at the losses due to resistance are not that high and the voltage-drop at the free-wheeling-diode is small compared the voltage of 3+ LEDs.
But when you want to drive only one LED with a high current every voltage drop has to be as small as possible. The resistance in the current path of my driver is:
RMOSFET = 10mOhm
RInductor = 14.8mOhm
RShunt = 20mOhm (I could have gone lower but to keep a good noise immunity I wanted a voltage drop of 60mV @ 3Amps... and I had the 20mOhm shunts laying around on my workbench )
-> Rtotal = 44.8mOhm

ULosses = Rtotal * I = 44.8mOhm * 3Amps = 134.4mV

Lets pretend the P7 got a VF = 3.5V @ 3Amps:

Efficiency = ULED / (ULED + ULosses) = 3.5V / (3.5V+134.4mV) = 0,963 -> 96%

Of course you lose a bit here and there like i didnt count the resistance of the circuity path and I didnt count the ESR of the capacitors (but I used Sanyo OSCON capacitors which are supreme quality and got a ultra low ESR).

Is the potentiometer an audio taper, so as to make the light output increase exponentially?

Yes, I used a exponentially increasing potentiometer with integrated switch. When you turn it all the way counter-clockwise you have a notch which switches a NOC (normally open contact).

 

[Aircraft800]

Here is a nOOb question since I know nothing about how this works,

can I use an additional PWM dimmer in series after this buck converter like the D2DIM by TaskLED? I like the idea of stock looking with everything hidden inside.

Sorta like this setup:
http://www.candlepowerforums.com/vb/showpost.php?p=2630275&postcount=42

 

[snarfer]

Just curious, did you measure efficiency of 95%, or is that your projection based on the math? What was actual input and output voltage and current?

Also, from my own experiments I have found that inductor design made a huge difference in efficiency. Did you account for such factors as core loss, AC resistance in the windings and so forth in your calculations?

 

[jeffosborne]

Good job, Sector_Cleared! Welcome to the forum. We would be delighted if you could provide a BEAM SHOT photo or two from your wonderful light. Thanks for your contribution to the collective knowledge of light builders.

Jeff O.