Announcing - Fatman lives!

[georges80]

Thanks for the compliments guys. I'm very happy with how easily Fatman fired up - been an easy journey so far (time to worry I guess...).

Did some more quick tests this morning. First with 8V into Fatman driving 1A into 3 Q2H stars. Well over 90% efficient.

Then I dialed back the powersupply to 6V and fed about 820mA (turned the dimmer down a bit) - my supply is only capable of 1.5A output. Anyhow still well over 90% efficient into 3 Q2H stars.

The pics - first with the dimmer dialed down all the way, 30mA into the stars (series connected).

And the next with 820mA into the Stars - BRIGHT!

For the efficiency tests I measure the voltage right at the input pins of Fatman. For output voltage I measure it at the output pins. For input current I take the powersupply current display - checked many times before and it is within the display resolution. For output current I measure the voltage across the 0.1ohm 1% sense resistor & multiply by 10.

I need a bigger powersupply to do more extreme testing!

george.

 

[MoonRise]

Yeah, an efficient high power regulated driver board!

I'm interested as well.

Any heatsinking issues or ideas for the driver board? 90% efficient of 5W still leaves 0.5W of heat to dump out, mostly going through that one IC in the middle. I'm thinking along the lines of extended run times, not short burst operation. Even 0.5W can add up some heat after a while, especially buried in a dead air space inside a flashlight.

Curious about your guesstimate of price as well.

 

[Ocelot]

So, what coil are you using, and where did you buy it? I've had a hard time finding small coils with low ESR and high current capability.

Thanks,
Scott

 

[georges80]

Moonrise - 1/2W isn't too much to dump out as heat. Actually on the tests the efficiency into a 5W is more like 95% /ubbthreads/images/graemlins/grin.gif

If you look at the start of the thread & the picture you can see many connections from the switcher chip's GND/Power pins to wide copper sheet. The back of the board is also all copper and multiple vias are connected through to it. The board is 0.8" in diameter so there's quite a bit of thermal surface area. It would of course be 'smart' to mount the edge of the board (bottom edge is ground) to the body of the light for some heat transfer when driving multiple emitters from lower voltage sources (efficiency will be lower).

Hmm - a NTC thermistor instead of the dimmer pot or along with it... An exercise left for the reader /ubbthreads/images/graemlins/icon15.gif

The switcher chip is thermally protected as well.

Edit: Price guesstimate - $20 < Fatman < $25

Ocelot - for inductors (free samples!) check out coilcraft's website

george.

 

[BikeLEDs]

Now THIS would suit my application very well. Keep us posted when this becomes available !!

Dave

 

[Ocelot]

>Remember, the driver is fully dimmable and is configured to put out around 30mA at the fully dim setting.

Since this is just a boost converter (not buck/boost), this 30mA minimum setting is dependent on the input voltage... As the input voltage approaches the Vf of the LED, the min current will be increasing.

Scott

 

[georges80]

[ QUOTE ]
Ocelot said:
>Remember, the driver is fully dimmable and is configured to put out around 30mA at the fully dim setting.

Since this is just a boost converter (not buck/boost), this 30mA minimum setting is dependent on the input voltage... As the input voltage approaches the Vf of the LED, the min current will be increasing.

Scott

[/ QUOTE ]

Of course. Though even at 30mA the Vf of a Luxeon is still relatively high. But yes, for it to maintain current regulation, your input voltage must be no great than the (total) Vf at 30mA. It won't damage the converter, but the dimming won't work.

george.

 

[eebowler]

Vin min is 2.7V.What are the output voltages when the input is 3 and 4.5V? I am wondering if it can be used to drive a single LuxIII. Is that possible?

Any runtime tests up and comming?

 

[georges80]

[ QUOTE ]
eebowler said:
Vin min is 2.7V.What are the output voltages when the input is 3 and 4.5V? I am wondering if it can be used to drive a single LuxIII. Is that possible?

Any runtime tests up and comming?

[/ QUOTE ]

Fatman is a boost converter, so the input voltage must be below the output voltage. A Lux 3 Vf is only around 3.5V (typ) - so 4.5V would be too much.

Runtime tests - so many combinations of Vin (and battery chemistry) and Luxeon combinations to test that it's beyond my time right now. Also, the current output enters into the equation - since Fatman is adjustable (better to say 'configurable') via the onboard trimpot - or dimmable via an external pot - that increases the # of combinations even more...

The switcher chip on Fatman in considerably better than the LT1618 that we're all familiar with. It's main limitation is that Fatman has a 2.7V min input so isn't suited to the single Lux1 or Lux3 applications.

Given the interest I've seen in the thread I started in Modified B/S/T I will do a production run of PCBs. I'll most likely have the first finished drivers available in around 2wks. When folk get them in their hands I'm sure we'll get better runtime figures for various battery/Luxeon combinations.

george.

 

[georges80]

Decided to connect FOUR series Q2H stars to really stress out the converter and more closely match the total Vf of 2 series 5watters. Did some measurements at various drive currents and various input voltages.

The efficiency is excellent at all levels over 100mA. Even at low currents it's pretty decent.


LED current Led voltage Input Voltage Input current Efficiency
900mA..........13.25V........8.94V............ 1.37A...... 94%
994mA..........13.41V........9.46V............ 1.44A...... 95%
31mA........... 11.0V........ 9.64V............ 43mA........80%
163mA..........11.73V........9.61V............ 0.20A...... 96%
163mA..........11.73V........4.12V............ 0.50A...... 90%
312mA..........12.13V........4.06V............ 1.00A...... 90%
442mA..........12.40V........4.00V............ 1.50A...... 88%

Edit: Table updated after measurement & compensation of 0.10ohm sense resistor change post soldering.

Again, all voltages were measured at the pins to the Fatman driver with a Fluke 87III. Output current was
measured by measuring the voltage across the 0.1ohm 1% resistor (on the mV range of the Fluke). Power
supply current was taken from the HP power supplies current display.

It's great to have the dimmer pot hooked up - I can vary the output current to whatever I desire from
30mA to 1A (well, 994mA...).

I only run the high current test for short periods at a time since the Q2H stars are only on their 'panel' so they get /ubbthreads/images/graemlins/mad.gif

george.

 

[HarryN]

Hi George - Is this unit rated at (-40 C)? - Yes, my favorite question.

 

[dabiscake]

Nice work! /ubbthreads/images/graemlins/clap.gif great efficiency numbers to. BTW, you wouldn't want to add a battery low voltage cut-off in your circuitry, designed towards rechargeable LiIons... would you? /ubbthreads/images/graemlins/naughty.gif /ubbthreads/images/graemlins/grinser2.gif

 

[georges80]

Harry: -40C is the rating of the semiconductor devices. The rest are R's & C's (the C's are X5R) so yes, -40C operational should be just fine.

dabiscake: Thanks, the efficiency numbers are very good - I'm quite tickled!

Here's another test:

Led V Led I In V In I Efficiency
12.01V 252mA 2.53V 1.46A 82%

This is pretty well the worst case - very low input voltage.

All the tests max out the input current at around 1.5A - the limit of my powersupply.

Low voltage cutoff would be nice - but that's more components and it would need to be adjustable since I can't 2nd guess the chemistry or # of cells folk will use.

I'll put a PAD on the active low Shutdown pin of the switcher and if someone wants to build a separate low voltage sensor board - they can then drive a low into the Shutdown pin of the Fatman and turn off the switcher.

george.

 

[Doug S]

George, fantastic performance. /ubbthreads/images/graemlins/thumbsup.gif I predict that your boards will be highly appreciated. I've a couple of questions. I presume that the minimum voltage is based on a UVLO on the switcher IC. What is the datasheet min-max range of the UVLO? What I am wondering here is whether the existing IC UVLO is adequate for single cell Li-ion applications. Also, when the IC is in shutdown what is the board residual current draw [sum of IC shutdown current and your current sense circuit]. Different question. In doing your efficiency measurements did you calibrate the .10 ohm 1% resistor *after* board assembly? The reason I ask is that I have found that the low ohm 1% SMT resistors I have been using lately have been increasing 3-4% in value post reflow. To get good efficiency data on the circuits I have used them in for current measurement, I have been driving a known current through them and measuring voltage to calculate their actual post reflow resistance.

 

[georges80]

[ QUOTE ]
Doug S said:
George, fantastic performance. /ubbthreads/images/graemlins/thumbsup.gif I predict that your boards will be highly appreciated. I've a couple of questions. I presume that the minimum voltage is based on a UVLO on the switcher IC. What is the datasheet min-max range of the UVLO? What I am wondering here is whether the existing IC UVLO is adequate for single cell Li-ion applications. Also, when the IC is in shutdown what is the board residual current draw [sum of IC shutdown current and your current sense circuit]. Different question. In doing your efficiency measurements did you calibrate the .10 ohm 1% resistor *after* board assembly? The reason I ask is that I have found that the low ohm 1% SMT resistors I have been using lately have been increasing 3-4% in value post reflow. To get good efficiency data on the circuits I have used them in for current measurement, I have been driving a known current through them and measuring voltage to calculate their actual post reflow resistance.

[/ QUOTE ]

UVP is 1.95(min) to 2.2(Max). It's actually the opamp in the current regulation path that is 2.7V min. Though on the bench it works just fine to 2.5V.

I haven't shutdown the converter yet - the production board will have a pad wired to the shutdown pin - so I will be able to measure that. I'm expecting the boards back this week and then I can take more measurements. But I'll be out camping the w/end and early part of next week - so further testing will have to wait till after that. By bringing out the shutdown pin to a pad someone (else) can build a small UVP circuit for li-ions & rechargeables protect them from severe overdischarge.

The shutdown currents due to Fatman should be in the <100uA range (based on the datasheets) - BUT this is a boost circuit and there is a direct path through the inductor and schottky between the battery and the load - so most of the shutdown current will be due to the Luxeons (which will depend on their total Vf and the battery voltage).

You are RIGHT - I just measured the output current with a Fluke 87III and it is just under 3% lower than the V/R of the 0.1ohm sense resistor. That brings the efficiency down about 2-3% in line with the datasheet (actually still a bit better). I'll have to update my efficiency table /ubbthreads/images/graemlins/blush.gif

cheers,
george.

 

[Illuminated]

Very nice work George!

Can't wait to buy at least one...still basking in the delightment of the uFkex.

John

 

[jdriller]

George makes very cool boards /ubbthreads/images/graemlins/bowdown.gif /ubbthreads/images/graemlins/clap.gif

 

[moraino]

Hi George,

Can Vin higher than Vf of the Luxeon. Badboy will be fried if this is the case.

I wonder if Fatman is suitable for my 18650 Li-ion with Twok Lux3 setup in a ASP Triad conversion. It draws about 690mA at 4.0 volt. I also need low voltage protection of 3 volts.

 

[georges80]

[ QUOTE ]
moraino said:
Hi George,

Can Vin higher than Vf of the Luxeon. Badboy will be fried if this is the case.

I wonder if Fatman is suitable for my 18650 Li-ion with Twok Lux3 setup in a ASP Triad conversion. It draws about 690mA at 4.0 volt. I also need low voltage protection of 3 volts.

[/ QUOTE ]

Fatman has a 1.5A rated diode that is essentially in series with the LED - so if your input voltage is high enough to push more than 1.5A throught the diode and into the LED then you are stressing out the diode. I just tested a Fatman and pushed 4.2V into it with a Lux1 (Q2H bin) with a Vf of 3.3 volts (at 350mA) and saw about 1.3A from the power supply.

For undervoltage protection, being a boost and running down to 2.5V it will definitely over discharge a li-ion. Fatman has a thruhole pad that can be used to shutdown the converter. One could use an external circuit to drive the active low shutdown pin when necessary to turn Fatman off.

george.