Announcing the uFlex Luxeon driver

Announcing the availability of the uFlex driver board.

Edit: New V1.00 firmware allows the user to reconfigure uFlex from 1W to 5W or back again.

This is the driver board that will also go with the FlexiLED (originally Stalklight) - CPF FlexiLED thread.

Technical details of the driver can be found at FlexiLED/uFlex Page . There you'll also find a PDF document that describes the interface and features.

Refer to the PDF for all the details, but a quick summary follows:

uFlex runs from 3V to 20V and can drive 2 series connected 1W luxeons, 2 series connected 3W Luxeons or a single 5W Luxeon. If the input voltage drops below the Vf+0.5V (approx) uFlex will go into direct drive mode.

Drive current is a nominal 380mA for 1W and 760mA for the 5W configuration. The drive current is set in the firmware and can be set up to 1000mA. It is not user programmable since it is in the firmware.

There are a total of 8 brightness levels, ranging from very dim (5-20mA) all the way to full brightness. The current brightness setting is stored in EEPROM for next power-on. Autosleep mode is available so that the unit will start dimming after 45 minutes and turns off about 2 hours after that - autosleep can be user disabled (mode is stored in EEPROM).

The board is rectangular 1.6" x 1.25".

uFlex is intended to be offer a driver solution for reading lights, maplights, tasklights and other fixed or semifixed lighting. I've used it in semiportable FlexiLED prototypes and found the variable light output to be a great feature. Of course being a switching powersupply, the battery life is increased dramatically when the Luxeon is dimmed.

Each brightness level appears equal step sized (approximation of human eye response), unlike other products I've been told about, i.e. you WILL see a distinct light output change as you cycle through each level!

george.

This is an EXCELLENT product.

Sounds like you really put alot of thought and work into this.

This thing is jam packed with features.

Are you planning on having these for a while? Or are they not going to be avalible after this run?

jtice, thanks! Yeah it has a bunch of features and took some time to develop, that's for sure! Darell was my alpha site so any complaints about features or user interface should go straight to him /ubbthreads/images/graemlins/grin.gif and he I'm sure will immediately forward them to me /ubbthreads/images/graemlins/whoopin.gif

All I can say is that it's very nice to be able to dim a Luxeon - just one step down from full bright will nearly double the battery life (human eye's response to light intensity mostly tracks a 10**x curve).

They'll be around for quite a while, I did a run of 500 PCB's and I have about 90 assembled and complete and another 100 close to complete. If they're popular then nothing stops me running more PCBs etc.

The drivers are intended to go with my flexiLED product but I will sell the drivers on their own for folks that want to use them in different applications.

george.

Wow -- color me impressed. That is exactly the circuit that I've been wanting for a while now. I have two questions:

A) When will the bare boards be available to the public?

B) I'm building a desk light but it's also one that I would love to be able to run off of batteries if I so choose. In other words I'm looking for it to be interchangeable. Forgive me -- I'm new at this. Would I be able to unplug the circuit from a 12v 500mA wall wart and directly hook up 4 to 6 "D" batteries? Ticey was attempting to explain internal reistance and the intrinsic qualities of a circuit to me but it didn't find it's way into my thick skull.

I got $ burning a hole in my pocket here and a project in mind that would be great for... Price/availability? (Possibility of a sample for review for my site? /ubbthreads/images/graemlins/blush.gif )

oO

Lukek: boards are available now - see my website for pricing etc. There's also a thread in BST Lights.

Yes, it will happily run on 12V and 4 or 6 D cells. It will drive a 1W Luxeon if fed with >4V - 20V DC.

Oddone: yeah right a sample - I'll mail you a PCB and a bag of components and 6" of solder ok? /ubbthreads/images/graemlins/grin.gif oh, and a copy of the software development tools (they're free).

In Summary - assembled/complete/working boards available NOW. Price on my Website - my led webpage

george.

Yeah, anybody should feel free to complain to me about the interface. I'll quickly tell you why you're wrong... and even how many times you're wrong. /ubbthreads/images/graemlins/wink.gif

LukeK - this is your lucky day. As George says, this'll do exactly what you want. The flexibility of the Vin makes this thing really versatile. Works just as well in an automobile/boat/RV as it does on a wallwart, or a few AA batteries. Regulation is a beautiful thing.

[ QUOTE ]
georges80 said:
Oddone: yeah right a sample - I'll mail you a PCB and a bag of components and 6" of solder ok? /ubbthreads/images/graemlins/grin.gif oh, and a copy of the software development tools (they're free).

[/ QUOTE ]

Muahahahahaha, beware making such remarks to me - I've been known to -welcome- baggies of SMD parts and whatnot. And I keep a special 0.5mm tip on a propane iron just for soldering 0803 resistors. /ubbthreads/images/graemlins/grin.gif

At any rate, I'll be hitting your site shortly to see what the prices are. My aforementioned project involves five-watters, so I'm definitely looking for a driver stout enough to handle individual 5-watt Luxeons.

That reminds me... Is it possible to chain commands among multiple uFlex boards from one pushbutton? If so, you're gonna make a lot of modders' day! (And my life a helluva lot easier, lemme tell ya. /ubbthreads/images/graemlins/grin.gif )

oO

[ QUOTE ]
OddOne said:
At any rate, I'll be hitting your site shortly to see what the prices are. My aforementioned project involves five-watters, so I'm definitely looking for a driver stout enough to handle individual 5-watt Luxeons.

That reminds me... Is it possible to chain commands among multiple uFlex boards from one pushbutton? If so, you're gonna make a lot of modders' day! (And my life a helluva lot easier, lemme tell ya. /ubbthreads/images/graemlins/grin.gif )

oO


[/ QUOTE ]

The standard firmware flavours of uFlex are either 350mA or 700mA (nominal) - at 8V maximum (output cap rating - no temp derating necessary). With the 0.15 ohm sense resistor (which could be made lower) the firmware can be modified (by me...) to drive up to 850mA. Lowering the resistor value would allow me to increase the drive current to 1A easily - it just wasn't in my 'master plan' /ubbthreads/images/graemlins/thinking.gif

The switch could be wired to more than one unit - it just grounds a pin on the uC - assuming they all started in 'sync' then they should be able to 'track each other'. I've never tried it - it might work - but there might be 'timing' corner cases (the uC's are using their internal oscillators) where releasing the button at just the 'wrong' time might make one board 'see' the button being down a bit longer than another board - and then the brightness will be one step out of sync. You might just have to buy 2 or 3 uFlexes and 'stress test' the concept for me /ubbthreads/images/graemlins/grin.gif Seriously, I'll try it out in the next few days. The autosleep definitely wouldn't sync up though - over the long 45/120 minute time the internal clocks would DEFINITELY slip out of sync.

george.

An idea, but not knowing the actual circuit design and code involved I don't know how feasible it is:

Is it feasible to provide some jumper that can be replaced with a simple single wire to disconnect each board's oscillator and take a feed off one driver? What popped into my head was something along the lines of a low-profile two-pin header for microjumpers where one pin is tied to the osc output and the other heads off to the clock input to the microcontroller. Jumper the two and the internal clock is used, but connect the out from one to the in for however many others and they all sync up clock-wise. Then, in theory at least, they should work in sync in every way, unless powered up in different states.


Probably not doable, but it sounded good. /ubbthreads/images/graemlins/grin.gif

As for buying a few, we'll see...

oO

[ QUOTE ]
OddOne said:
An idea, but not knowing the actual circuit design and code involved I don't know how feasible it is:
........

oO


[/ QUOTE ]
Nope - the oscillator is internal - the attiny15 only has 8 pins and I needed them all... so I'm not using an external Xtal. The AVR is pretty neat in that it has an internal oscillator option - that I used, sorry...

george.

Awwwwwww, okay. /ubbthreads/images/graemlins/grin.gif

I keep seeing and hearing of good things from that little Atmel chip. I might just have to slip a few of their part #s into my next Mouser order.

oO

Hmmm. Surprisingly, some of this is going over my head. /ubbthreads/images/graemlins/thinking.gif

[ QUOTE ]
OddOne said:
I keep seeing and hearing of good things from that little Atmel chip. I might just have to slip a few of their part #s into my next Mouser order.

oO


[/ QUOTE ]
Check the digikey prices too - unless you specifically like to order from mouser. I'm pretty sure (at least in 100 piece qtys that I ordered in) that Digi was cheaper.

george.

George,
This things are looking GREAT.

I need one thing cleared up.

I am planning a 2 1W blackout/emergancy light, that this circuit would be Perfect for.

If I have 2 1W LED's what is the max mA to each LED your board can be set to?
You say 1,000mA,,,, is that to each LED? OR, 500mA each?
I would like to drive around 700mA or more to each LED if possible.

[ QUOTE ]
jtice said:
George,
This things are looking GREAT.

I need one thing cleared up.

I am planning a 2 1W blackout/emergancy light, that this circuit would be Perfect for.

If I have 2 1W LED's what is the max mA to each LED your board can be set to?
You say 1,000mA,,,, is that to each LED? OR, 500mA each?
I would like to drive around 700mA or more to each LED if possible.

[/ QUOTE ]

Ok, with the 0.15ohm sense resistor I used the max current is about 850mA. That maxes out the input to the a/d converter. To go higher I have to put another 0.15ohm in parallel with the 1st one. I haven't worked the numbers to decide what the real maximum current that the board is capable of before saturating the Inductor etc. Something I guess I'll have to do for you power fiends /ubbthreads/images/graemlins/grin.gif

For now let's just say the max current the board can provide - with suitable firmware is 1000mA.

So,

Current is 1000ma Max at 8V (output cap voltage rating).
So, 1000mA to one 5W Luxeon
or 1000mA to 2 series connected Luxeon III's
or 1000mA to 2 series connected 1W Luxeons.

I should rename my the uFlex flavours - 1W/5W is a misnomer since it's really current drive.

Basically I have 2 versions (off the shelf)
350mA nominal (actually about 380mA - a bit of overdrive)
700ma nominal (actually about 750mA - a bit of overdrive)

Since they are constant current regulators - they will provide that current level into whatever load is there - assuming sufficient input voltage, since the uFlex is a buck down converter. The 8V max output is to protect the output capacitor. No temp derating is required since the output cap is a high quality aluminium polymer device.

goodnight,
george.

Hi George

I am thinking if it can be sandwiched in the MAG 2D switch (with 8AA in it) and cycle through the levels of brightness.

Another member did it by cutting the MAG switch in half and put the Willie Hunt's board inbetween the two half with switching cabability. I can remember the thread.

That was for the incan and yours is for LED and dimmable.

If would be a dream come true if it could be put inside the MAG light. Whether it can or cannot, I'll get one to play with. Great job.

Henry

Hello George,

I hope you don't mind that I posted your website address to a German bicycle mailing list where currently a hot discussion regarding LED lights for bikes is going on /ubbthreads/images/graemlins/wink.gif

On a more serious question: As I understand it, you boards work as buck converters, and they're limited currently to 8 V output. What would be involved (and would it be possible at all) to modify them for <= 11V output, i.e. 3 1W/3W in series?
Oh, and will you (or the docs) tell me which resistor I would need to change in order to get other current values (for both Driver and uFlex)?
And what would be your minimum qty for a modified firmware (2..3 power levels at specified current for the uFlex)?

Bye
Markus

[ QUOTE ]
markus_i said:
Hello George,

I hope you don't mind that I posted your website address to a German bicycle mailing list where currently a hot discussion regarding LED lights for bikes is going on /ubbthreads/images/graemlins/wink.gif

On a more serious question: As I understand it, you boards work as buck converters, and they're limited currently to 8 V output. What would be involved (and would it be possible at all) to modify them for <= 11V output, i.e. 3 1W/3W in series?
Oh, and will you (or the docs) tell me which resistor I would need to change in order to get other current values (for both Driver and uFlex)?
And what would be your minimum qty for a modified firmware (2..3 power levels at specified current for the uFlex)?

Bye
Markus

[/ QUOTE ]

My god - not the germans /ubbthreads/images/graemlins/grin.gif

It would certainly be possible to change the output capacitor to a higher voltage rating. I've got some Luxeon III's on order from ElectoLumens so when I get them I'll be able to play with various combinations - and like I said in a previous post, I need to work out the maximum power that the uFlex can provide. Something will be the limiter, either the inductor or the fetky I'm using. At least being external devices I can probably find pin compatible higher power versions.

To have custom software it would of course depend on the severity of the change. Having just 2 or 3 levels isn't too hard - I would classify it as a qty 5 type job. The current drive of each level can be customer specified.

george.

Just some tests I did this morning - prepping a 1000mA version of uFlex.

Driving a Q2H (1W) at 1000mA (Vf at 1000mA was 3.54V) I measured pretty flat 80% efficiency from 6V, 8V and 12V inputs. This is with the series protection schottky and EMI beads still in circuit - they contribute to about a 5% efficiency loss at the high current flowing through them to the board to drive 1000mA into the Q2H.

Another 2% efficiency is lost across the 0.1 ohm sense resistor I used for the 1000mA version. That one you can't do much about.

So, in summary (comparing against my previous tests), the uFlex loses about 6% in efficiency when driving 1000mA through a 1W/3W Luxeon versus 350mA. Losses are most likely in the inductor and Fetky - due to higher currents.

george.