Controlling Luxeon brightness

StarMan

Does anyone know of a chip which can be used, and controlled by a Basic Stamp, to control the brightness of a Luxeon 1 watt emitter.· These units·draw 350 mA at full brightness and·brightness should be·adjusted by·a current controlling scheme·as opposed to a voltage control.

The MAX1698 is good up to 125 mA.· I can't seem to find anything beyond that.

Would it be possible to use multiple MAX1698s?· What the best approach to controlling them?


Chris Isaacson

dbpage

I don't see in the spec sheet http://pdfserv.maxim-ic.com/en/ds/MAX1698.pdf·where the max output current of the MAX1698 is 125 mA. I see a max CS trip current of 2.5 A, with typical at 1.5 A. The current that flows through the LEDs flow from Vbatt thru an inductor,·thru a zener diode and thru Rfb. The amount of current that flows through the LEDs is determined by maintaining 300 mV across Rfb.· Set Rfb to 0.857 ohms, and you'll get 350 mA.· Your task is to choose a zener that handles 350 mA, if you plan to add secondary LEDs in parallel with the primary LEDs.· The optional zener·prevents overvoltage across the secondary LEDs (if any) in case the LED in series with Rfb opens. The zener is not needed if there are no secondary LEDs.

StarMan

Thanks for the reply, Dennis.

The 125 mA is a non-traceable mental error on my part.· Thanks for pointing that out.· It prompted me to take a better look at the spec sheet.

At this time I will not be running parallel LEDs but I may be running up to four in series.

I have some additional questions:

I don't understand the Shutdown function. Is it optional?

Given the forward voltages from the Luxeon Emitter data sheet (http://www.lumileds.com/pdfs/DS25.pdf), what are the appropriate voltages for Vbatt and Vcc?· Could they both be supplied from the same power supply?· If so, do they need to be individually regulated?· Suggestions on regulators?

Any suggestions on transistor selection?

Any suggestions on diode selection?

How do I interface a BS2 with ADJ (preferably without using PWM which is not in the public domain)?· Is there a digital pot available for this?

Thanks for your help.

Chris Isaacson

dbpage

The shutdown feature is optional. The shutdown·input can be controlled by a stamp.· I assume that the ADJ input can be controlled by a stamp, if you build circuitry to insure that the voltage never exceeds REF, which is typically 1.25 V.·Maximum brightness is when ADJ is at 1.25 V. A white·Luxeon has minimum brightness at 100 mA and maximum at 350 mA. According to the specs, an Rfb of 0.857 ohms sets the maximum current to 350 mV. Minimum brightness is when ADJ is at 356 mV. Your stamp output would need to be adjustable between 356 mV and 1.25 V. To drive ADJ with a stamp, you might consider the ADC0831 analog to digital converter as described in http://www.parallax.com/dl/docs/books/edu/baad.pdf , or·I suppose you could use AD 5220 digital pot as described·http://www.analog.com/en/prod/0%2C2877%2CAD5220%2C00.html·. The forward voltage for a white Luxeon is typically 3.42 V. Vbatt for four white Luxeons in series would need to be less than ·13.68 V. Vcc must not exceed 5.5 V. An appropriate voltage for Vcc is 5 V. You could try Radio Shack MOSFET catalog number 276-2072··http://support.radioshack.com/support_supplies/doc31/31806.htm. I did not find a schottky diode at radioshack.com. Pick one that has a reverse voltage that exceeds 14 V.

I hope I answered all your questions. Consider what you are paying for this 'advice,' which is zero. I calculated all values from the spec sheets. Actual values may vary. You may want to adjust some values, for example, to use a 1 ohm resistor instead of a 0.857 ohm resistor,·which would·reduce the maximum brightness somewhat in exchange for a readily available component.

Paul Baker

Any digital pot should work in with the 1698, you just dont want a total resistance so high that the -50 nA to 50 nA input bias of the ADJ pin is on order of the current flowing through the pot. A general rule of thumb is to have the current flowing through the pot to be at least 2 orders of magnitude (100x) more than the current flowing through the wiper. The maximum the 1.25 votage reference can source is 150 uA. With R=V/I, the minumum pot resistance is 8 1/3 k, so don't get a pot less than 10k. The maximum digital pot resistance commonly available is 250K -> 5 uA or a factor of 100 over maximum input bias resistance. Depending on how the reference is constructed nonlinearities can exist if it is over or underdriven (having to source very low or very high currents), typically shooting for a middle of the road current is a safe bet. I personally would choose a 100K pot (the 250K pot from AD is > $9 while most 100K's are < $2), but the data sheet shows no problem with choosing the 250K (they use a 500K pot). All other factors are a choice of personally preference, but the more taps a digipot has, the more levels of brightness (in an U/D controlled digipot this also means a longer wipe up/down time).

The shutdown function is optional, this could potentially be used as a PWM method of adjusting brightness, but it wasn't designed for this purpose so I wouldn't recomend it.

Vbatt and Vcc can be the same, the only restriction on Vbatt is that it must be less than the cumulative voltage drop of your series of LEDs for the shutdown function to work.·The Vf of the Luxeons has no bearing on required Vbatt and Vcc (that is the whole purpose of the 1698). The 1698 is a charge pump that boosts the output voltage so that a current determined by Rfb is maintained. Both Vcc and Vbatt can be run by a couple of standard cells (3V) or a 5V supply so long as your source is between 2.7V and 5.5V and can drive the current your running the Luxeons at plus a margin for the current consumed by the 1698 (not much compared to the LED drive current).

Any power mosfet capable of handling 3A should work (this is the max CS current), the sheet gives MBR0540 as an example for the diode.

I would choose a digipot which is compatable with BS2's serout command, I think SPI is compatable but don't take my word for it. Your choices of interfaces are 1,2,5-Wire, SPI and·I2C. Ask somone in the Basic stamp forum which would work with serout.


Post Edited (Paul Baker) : 11/19/2004 11:58:28 PM GMT

StarMan

Paul, Dennis,

Thank you.· Your input is greatly appreciated.

I've ordered samples of the MAX1698.· They also offer an evaluation kit (MAX1698EVKIT) which is very expensive but the data sheet has a parts list which is very interesting.

I'll definitely be busy for a while studying data sheets and digesting all of this



Chris Isaacson

pjv

Hi Starman,

I believe the best way to control and limit the current to your intense LED is to use a tiny fly-back switcher. It dissipates only neglible power, and, if properly designed, short circuit proof.

Several things are critical however, and if not done right WILL smoke the circuit. So you need to be comfortable with hardware and measuring techniques.

You need to have the processor give a consistent pulse train into a swithing transistor or FET which alternately charges the primary winding of a tiny ferrite 1 to 1 flyback transformer for an "ON" duration, and then let the built up magnetic field decay into the transformer secondary during the "OFF" duration.

The transformer and LED polarities MUST be connect in the fly-back manner, meaning that during the charge (ON) cycle, the LED on the secondary is not conducting. Only during the discharge (OFF) cycle, when the secondary voltage reverses may the LED be permitted to conduct.

I would start with measuring the input charge current by observing the voltage accross a small (one ohm) resistor with an oscilloscope·with a 50/50 ON/OFF duration of 10 microseconds. Then I would slowly bring the speed down and watch the primary current increase, and go no slower than observing the onset of transformer saturation or the desired secondary current. In a 1 to 1 transformer, the secondary current will be nearly the same as the primary current.

It is absolutely imperative that your controller software cannot stall in the conducting (charge) state, thereby over-currenting the transformer input inductance and "popping" the switching FET.

Also, the LED polarity MUST not conduct during the charge cycle, as that also could pop the FET.

As with all flybacks, be careful of high spike voltages when the charged field collapses without a suitable secondary load connected.

When done right, these things are not scary, but hardware knowledge and caution are always advised.

Have fun,

Peter

StarMan

Hi Peter,

Thanks for your reply.· However, it's a bit over my head.· Do you have any reference material I could read to educate myself further?· I have a 50Mhz scope.· Would that be adequate?

Chris Isaacson

pjv

Hi Chris;

I composed a one hour response to you·two days ago, detailing my suggestions to you. As I am currently on a cruise ship in the Caribbean, internet coverage via satellite has been spotty and problematic. I lost the whole response, and there is no ability to save, cut and paste etc on these computers.

So I'll have another go when I get back ashore in a week or so.

Peter

Paul Baker

Peter,

What the heck are you doing cruising the forums while your on a cruise ship? Log off, get drunk and get burnt in tribute for all of us not fortunate enough to enjoy the mild Caribbean breezes.

Paul

(Misses his Florida like crazy from October to March every GD year)

StarMan

OK.

Here's where·I am.· I ordered and received some samples of the MAX1698.· They are incredibly tiny and still in the bag they arrived in.· I also took the plunge and bought the Evaluation Kit (http://pdfserv.maxim-ic.com/en/ds/MAX1698EVKIT.pdf) (to the tune of $100).

The Eval Kit works fine with the onboard LEDs in the upper current range but there is a cutoff as the LEDs dim.· I find this undesireable for my project. I'm·looking for·dimming all the way down.· So far, I've only been dimming with the pot.

Next, I removed the 15ohm resistors and connected a 1W Luxeon star and a 1ohm resistor.· The Luxeon is connected to Vout and GND and the resistor is across FB and GND.· The Luxeon comes on full bright and does not dim at all.· There is a little flicker·as I turn the pot but no change in brightness or current.· I measure 550mA and the Luxeon gets warm real fast so I don't leave it on for more than a few seconds.· Just for grins I replaced the 1ohm with a 15 ohm and I get a steady 90mA.· The pot still has no effect.

Any ideas would be appreciated.

Chris Isaacson

pjv

Hello Chris;

Well, I'm back ashore, and·two days ago·started to re-write my resopnse. It's quite a bit of work, because I'm trying to explain it in sufficiently simple terms so that I don't again·go "over your your head" as I apparently did last time.

In the meantime, I see you've been busy trying things, and that's good. We all learn from each experiment we do.

Although you did not state the voltage you were using with the evaluation board, I suspect the reason why you are observing the inadequate dimming results is due to having too high a VBAT voltage. You see, the switcher on the board is a BOOST switcher, meaning that it cannot deliver a voltage lower than VBAT. If you lowered the VBAT, the series·on-board leds will probably control down to the level you want. Alternately, if you added (seriesed) some more LEDs in the primary string, you would also obtain·the proper·result.

This is doubly the case when you redirected the switcher's output to·a single·Luxeon diode. It's forward voltage drop, I understand, is in the 3.5 volt range (as compared to 3 leds of maybe 1.5V in series), so a VBAT supply in excess of that will allways cause current to flow, whether the switcher is running or not. Clearly, the switcher cannot control the current in this situation.

The fly-back concept I referred to in my previous post, albeit somewhat more complicated, does not suffer from this issue. It can be made to operate properly from a relatively wide range of voltages.

If you think that you have now already selected the path you want to follow, please let me know, so I don't spend a lot of time completing my "fly-back" response in vain.

And yes, your 50 Mhz scope should be fine for experimentation, I just hope it has more than one channel.

Have fun experimenting,

Peter

StarMan

Hello Peter,

Welcome back from your cruise.· I hope it was enjoyable.· Thanks for your reply.

I'm using a LM2940 to provide 5V to the board. The on-board LEDs are Nichia NSPW500BS with a Vf of 3.6V @20mA.

I will gladly pursue the "fly-back" approach.· It sounds promising.· I will also continue to experiment with the Max1698 since I am already somewhat committed.· I'm definitely having fun.

My scope is a Philips PM 3215, 2 channel, 50MHz.

Chris

pjv

Hi Chris;

If you are also using the same 5 volts for the LED supply (that is the VBAT going to the inductor), then I'm somewhat stumped.

It's hard to believe the board or chip has a problem!

Peter