Fatman driving 3 Luxeons

[georges80]

Did a run with the prototype Fatman driver today with 4 AA powerex 2000mahr Nimh cells.

I got 57 minutes of runtime at 750mA pushed into 3 series connected Q2H stars - yeah they got hot. See the picture below. Note, Fatman was just warm/hot to the touch throughout the entire 57 minute run - no problems holding/keeping your finger on the PCB.

Note that in a real life the wires would be considerably shorter and/or thicker and the runtime would be a few minutes longer due to less loss in the wiring.

The camera fired the flash (spot metering) so the Luxeons weren't totally blinding to the camera /ubbthreads/images/graemlins/grin.gif

Of course with the dimming pot - lowering the current would increase the runtime proportionally.

A neat light would be a triple Lux3 with the Fraen triple lens in a stubby light with 4 series AA nimhs in a side by side configuration or 6 AA nimhs in those 3AA-D cell holders. With 6 series nimhs Fatman will happily push 1A or more into 3 series Lux3's.

I'm hoping to receive the production blank PCBs next week. All the components have been ordered and should turn up next week as well.

Fatman minimum input voltage is 2.7V so this is NOT a single 123 or two AA driver. It is a boost converter.

I will working to put up more technical details and documentation on my website this weekend.

george.

 

[Kiessling]

wow!
maybe you should talk to Elektrolumens, ha has some flashights that should need such a marvel!
bernhard

 

[Icarus]

Very very nice! /ubbthreads/images/graemlins/thumbsup.gif

 

[andrewwynn]

Neat... can i run from 3 AAs? what's the efficiency? What voltage possibilities for output? Is it current or voltage constant?

 

[georges80]

Yes, you can run 3 AA's (Vmin input is 2.7V). 3 'flat' batteries is 3 x 0.9 = 2.7V, so a good match.

Vout max is 16V (output capacitor voltage rating). Vin max is 12V (limit of switcher chip).

Fatman is fully current regulated and has a trimmer on board that allows adjust from about 30mA to over 1000mA (see below).

Since Fatman is a boost converter for any specific output current the input current will rise as the output voltage requirement rises. The switcher on Fatman has a minimum 2.5A current limit, with a typical of 3.6A. It has a max of 80 mOhms on resistance. These two parameters work together to affect the efficiency and the available current ouput. Basically lower current output capability (and worse efficiency) once the input/output voltage differential becomes large enough that the input current rises up and approaches the switch current limit.

Anyhow, after all that is said efficiency of Fatman is still VERY good. Typically it is 90% or higher.

On the above test here are some numbers (using 4 fresh nimhs). Using 2 meters:

Vin = 4.93V (measured at the converter)
Iin = 1.60A (measured coming from the batteries)
Vout = 9.76V (measured at the converter)
Iout = 750mA (measured across the 0.10 ohm 1% sense resistor)

Efficiency 92.8%

When you asked what the efficiency is - I will ask the question back - at what input voltage, what output voltage and what output current?

george.

 

[andrewwynn]

Is it possible to drive a single lux3? or is the Vf too low?

Say if i wanted to drive Lux3s... from 3 cells.. that's the problem i was having a while back... very difficult to find a happy camper solution... i can go to 4 cells and use a buck or go to 2 (or more) luxes and use a boost... i'm liking fatman... it wouldn't hurt me to put two lux3s in series to make a design that would run from 3 cells even if i have to under-drive the luxes to keep the I(batt) under control.

so.. Vbatt=4.5.. Vout =3.5 (probably not feasable).. so Vout=7.0... Iout maybe 350mA... 2.5Wish at the emitter(s).

I suppose a single 5W emitter might work with it's higher Vf... i'm just trying to make a 3cell lux light.

-awr

 

[georges80]

[ QUOTE ]
andrewwynn said:
Is it possible to drive a single lux3? or is the Vf too low?

Say if i wanted to drive Lux3s... from 3 cells.. that's the problem i was having a while back... very difficult to find a happy camper solution... i can go to 4 cells and use a buck or go to 2 (or more) luxes and use a boost... i'm liking fatman... it wouldn't hurt me to put two lux3s in series to make a design that would run from 3 cells even if i have to under-drive the luxes to keep the I(batt) under control.

so.. Vbatt=4.5.. Vout =3.5 (probably not feasable).. so Vout=7.0... Iout maybe 350mA... 2.5Wish at the emitter(s).

I suppose a single 5W emitter might work with it's higher Vf... i'm just trying to make a 3cell lux light.

-awr

[/ QUOTE ]

A single Luxeon from 3 cells is problematic - especially alkalines since as you've found you really need a buck/boost topology or an inverting design due to the input voltage bracketing the output voltage as the battery runs through its life cycle. And those topologies are generally much less efficient than a straight buck or boost implementation.

Also, as you vary the current to a LED the Vf changes. A Lux3 can be 3.6V at 750mA and require only 2.8V at 30mA, further compounding the buck/boost dilemna.

A single 5W would work just fine - the issue there is beam quality due to the large die. So you need to find a well matched reflector or optic.

Current control on Fatman is easy - just tweak the trimpot to set the maximum current level. Minimum current output is set at a nominal 30mA (on board fixed resistor in series with the trimpot). Maximum output current is limited by the switch current limit, efficiency curve and schottky diode current and runtime requirement.

Simply put, Fatman Vin (min) is 2.7V and it is a boost converter and Vin < Vout.

george.

 

[mrsinbad]

How can I get me a Fatman?

 

[georges80]

[ QUOTE ]
mrsinbad said:
How can I get me a Fatman?

[/ QUOTE ]

Lots & lots of pizza /ubbthreads/images/graemlins/grin.gif

Check my Fatman thread in "CPF's Custom & Mod B/S/T"

cheers,
george.

 

[georges80]

Production PCB's arrived this morning. I've assembled the first 8.

Front picture:

Back picture:

I will be building more and of course testing them.
There's a thread on custom/modified b/s/t on prices etc.
They'll be ready later next week.

cheers,
george.

 

[georges80]

I've added some specs & technical info (and will add more). There is a start to a table with resistance measurements (of the trimmer pot) versus drive current for folk that want to adjust Fatman to their own current output requirement.

fatman tech info

george.

 

[4sevens]

I got'em today! I just stuck one into my new 5w nuwai

700ma current to led
850ma from battery (2 123's)
efficency 81% (tell me if this is right (850-700)/700 = .21% loss)

I took out one of the cells and got 2250ma from the battery.
Same brightness. yay!

color still sucky but i have a wx1s on the way from klaus!

quick question.. when measuring resistance between vin- and pot
I get 6.6k for the preset 700ma
and 7.25k for the preset 1000ma

according to your specs on your website shouldn't it
be 9.37k and 13.50k respectively?


PICTURES! oh yeah.
web page

 

[georges80]

[ QUOTE ]
4sevens said:
I got'em today! I just stuck one into my new 5w nuwai

700ma current to led
850ma from battery (2 123's)
efficency 81% (tell me if this is right (850-700)/700 = .21% loss)

I took out one of the cells and got 2250ma from the battery.
Same brightness. yay!

color still sucky but i have a wx1s on the way from klaus!

quick question.. when measuring resistance between vin- and pot
I get 6.6k for the preset 700ma
and 7.25k for the preset 1000ma

according to your specs on your website shouldn't it
be 9.37k and 13.50k respectively?


PICTURES! oh yeah.
web page

[/ QUOTE ]

You're measureing efficiency incorrectly. If I undertand your calculation, you're just dividing the input & output current.

You need to divide the input/output POWER. Fatman should be in the 90% efficiency range.

Efficiency is (Vout * Iout) / (Vin * Iin),
i.e Power_Out/Power_In

Also, it is hard to measure the input current or output current without swaying the efficiency measurement somewhat - due to the voltage loss across the meter.

To do it properly you really need to have two meters, one measure input voltage (at the board) and the other measuring input current to the board. Output current will be constant, so you can just measure the output voltage & multiply it with the know output current. Also, do the measurements quickly then remove power - so the LED remains at a relatively constant temperature and the changing Vf (lowers as the LED gets hotter) doesn't affect the measurements of Power_out/Power_in.

The resistance measurements on my web site are accurate. I used a Fluke 87III to do them. I HAVE NOTED that cheap meters can give different readings since they apply a different current and since you are measuring the resistance in circuit with active components - the higher current can bias the active components and cause the effective resistance to be different.

Reverse the leads on your meter when you measure the resistance and tell me if the reading is different - it isn't with my Fluke...

Nice mod, congratulations on being the first to publicly show Fatman operating in a mod. Thanks!

george.

 

[4sevens]

Yeah, it was my stinky meter. I got another one and the
resistance is accurate.

I haven't gotten around to unscrew the head and measure efficiency.

Specific question. What exactly happens when Vin drops below 2.7v?
I drove the 700ma for an entire hour using 2 2000mah nimh's.
They were pulling 2.2A. About half way through it drops to 1.5A.
Strange. I can't tell the difference in the brightness by eyeballing
since I didn't have a light meter.

david

 

[georges80]

[ QUOTE ]
4sevens said:
Yeah, it was my stinky meter. I got another one and the
resistance is accurate.

I haven't gotten around to unscrew the head and measure efficiency.

Specific question. What exactly happens when Vin drops below 2.7v?
I drove the 700ma for an entire hour using 2 2000mah nimh's.
They were pulling 2.2A. About half way through it drops to 1.5A.
Strange. I can't tell the difference in the brightness by eyeballing
since I didn't have a light meter.

david

[/ QUOTE ]

As the Vin drops below 2.5V and lower, the feedback circuit that is monitoring the current will start to go out of spec and the current output will drop.

Since the human eye has a logarithmic response to intensity it is not surprising that even a 30% drop in intensity (current) is not too obvious.

Good to hear the better meter matches my measurements.

george.

 

[NewBie]

I don't like utilizing meters inline for current measurement, as they affect the source resistance (power supply).

Many make the big mistake of utilizing the voltage and current measurements from the power meter. Unfortunately, the wires running from the power supply to the converter have resistance in them. This results in the converter getting less voltage than what the power supply is indicating. As you move into higher and higher power switchers, the resistance in the power supply wires can really add up fast. Try this sometime. While the converter is running, measure the voltage drop, say on the plus wire from the power supply, from end to end with another meter. You'll be surprised at how much voltage loss there is, especially with something like a 4A input boost. Take this and double it, as the drop exists in the negative wire also.

I prefer a low value non-inductive resistor (0.025 ohms), and measure the voltage drop across that. The voltage should be measured right at the board input and the board output. Additionally, long wires (or tiny guage, such as 20ga) from the power supply affect measurements (think source resistance).

With the resistors, you can utilize one meter. This gets rid of relative accuracy issues between the meters. Since most hand held meters are typically 2-3% accurate (when still in calibration), you can have one meter 3% high and another 3% low, which results in a 6% error. Additionally, utilizing four meters could result in a 12% error. Then, if the meters have not recently been calibrated, they could be off even further. By using one meter, you end up cancelling out the error, since the same error (say 3% high), when used for both the input and output, ends up cancelling out.

A good set of high accuracy sense resistors is a good investment, imho. A Kelvin connection is a great bonus.

A great way to match the low value resistors (without expensive equipment), is to buy several, then wire them in series, and pass a few amps through them. Measure the voltage drop across each. If it is the same, they are matched. If they aren't swap them around, until you find a pair that are the closest together.

Wirewound resistors are a no-no, as with switchers, there are often pulses, and the inductance in the wirewound resistors can result in bad measurements.

 

[4sevens]

Another Q

What happens when Vin approaches Vout - say less than 1v?
Say Vin = Vout (given the appropriate I)?

david

 

[georges80]

[ QUOTE ]
4sevens said:
Another Q

What happens when Vin approaches Vout - say less than 1v?
Say Vin = Vout (given the appropriate I)?

david

[/ QUOTE ]

Fatman will transition to direct drive if the Vin > Vout. The current will then go through the inductor and series Schottky diode directly to the LED. Of course there will be no current regulation and the current will increase steeply due to the steep Vf curve of a Luxeon.

george.

 

[4sevens]

George,

check this out I just got a whole batch of these Lithium Ion cells
that are identical in size to AA's. They are rated 3.7v and 700mah
and 4.1v right off the charger.

My GROUPBUY HERE

Now, I hooked up ONE SINGLE cell to my 2D mag o-sink with a
Lux V WX1S (Vf at 6.2v) with fatman set at 1A with current at the
battery between 2.1A and 2.2A. It ran for exactly 20 minutes before
it flicked once and came out of regulation. That is really awesome
- either the cell is holding more capacity or your driver is quite
efficient!

Another setup, I put TWO cells in this time. However, due to the Vin
be higher than the Vf, the current at the battery was about 1.55A
which I understand it has transitioned into DD. Then I used a pair of these that are not fully charged (about %60) and the current was
about 1.2A. This time I left it on longer and watched the current. The neat thing is I watched the current drop slowly from 1.2A to about .950A and then it started climbing! Showing that it
transitioned into the boosting circuit mode.

I'm having too much fun with the fatman. /ubbthreads/images/graemlins/smile.gif
david

 

[georges80]

[ QUOTE ]
4sevens said:
George,

check this out I just got a whole batch of these Lithium Ion cells
that are identical in size to AA's. They are rated 3.7v and 700mah
and 4.1v right off the charger. (url=http://www.candlepowerforums.com/ubbthreads/showflat.php?Cat=&Number=626177&page=0&view=collapsed&sb=5&o=&fpart=1]My Groupbuy HERE[/url])

Now, I hooked up ONE SINGLE cell to my 2D mag o-sink with a
Lux V WX1S (Vf at 6.2v) with fatman set at 1A with current at the
battery between 2.1A and 2.2A. It ran for exactly 20 minutes before
it flicked once and came out of regulation. That is really awesome
- either the cell is holding more capacity or your driver is quite
efficient!

Another setup, I put TWO cells in this time. However, due to the Vin
be higher than the Vf, the current at the battery was about 1.55A
which I understand it has transitioned into DD. Then I used a pair of these that are not fully charged (about %60) and the current was
about 1.2A. This time I left it on longer and watched the current. The neat thing is I watched the current drop slowly from 1.2A to about .950A and then it started climbing! Showing that it
transitioned into the boosting circuit mode.

I'm having too much fun with the fatman. /ubbthreads/images/graemlins/smile.gif
david

[/ QUOTE ]

David,
certainly sounds like you're getting good mileage out of Fatman. Yes, it is a VERY efficient driver - mostly due to the low resistance of the internal switch element of the switcher and externally a good low resistance/high saturation limit inductor. For driving 1A there's not much out there that can touch Fatman for the size.

Yep, Fatman transitions pretty nicely from direct drive to boost. Just be careful in direct drive that you don't push too much current throught the schottky (rated at 1.5A)

george.