Step up with Decent regulation?

[Frank Schwab]

Does anyone have a step-up circuit with decent efficiency (> 70%), decent CURRENT regulation over a 2.8V-1.8V input range ( better than 20%, say), runs down to a reasonably low voltage level (1.6V? 1.0V? 0.8V?), and is capable of driving a Luxeon at 350ma? Schematic is fine; I don't need a pre-built.

The Illuminator Pill? Current regulation is difficult to measure but not very good, and it cuts off at 1.6V.

The MadMax 1674? Set for 375ma at 2.8V, it's down to 250ma by 1.8V.

The ZLT? I just finished putting together the CCR-2 version, and 390 ma at 2.8V turns into 217ma by 1.8V.

The BadBoy? I don't have one of those, but Wayne claims excellent current regulation from something under 2V up to 3V. After building and buying as many circuits as I have, however, I'm hesitant to spend another $35.
I'm not trying to slam anyone here, I'm just really frustrated at trying to get what I perceive as the coolest feature of LEDs (constant light output), and not finding anything that I can build or buy that does that. I know that tremendous effort and money went into each of the ones I've tried (after reading all 30 pages of the original ZLT thread, it's pretty obvious), but I haven't yet found what I'm looking for, so I'm asking for help.

Thanks,

/frank

 

[TripleDouble]

I actually love the ZLT circuit. It's great, but then I again, I have weird tastes. I don't want perfect regulation, if I did, I'd have no idea if my batteries had been previously used for 30 minutes or 2 hrs. And, if you're trying to push a constant 350 ma over that large a range of battery voltages, you'll be in trouble. i.e. if your driving your luxeon at a constant 1w of power, at 1.4v, your batteries will have to deliver twice (actually more than twice because step-up efficiency drops) the current than at 2.8v. Not only is this tough on the batteries, it's certainly not good for battery life. Draining nimh batteries down that low is bad for their health, and alkalines will have a hard time delivering the power you want at the lower voltages.

Perfect regulation might work fine for low current applications, i.e. the Opalec, but I certainly wouldn't want it for my luxeons.

 

[Frank Schwab]

> but I certainly wouldn't want it for my luxeons.

I appreciate and understand that, but...

I DO want it for my Luxeons.

/frank

 

[InTheDark]

Actually, the ZLT does a pretty decent job of doing regulation, much better than I thought it would. It's hard to judge the circuits just going by the specs, because as you probably know LED's get more efficient as the current goes down, so the difference between 390 mA and 217 mA isn't going to be nearly as bad as you think.

I measured the actual light output vs. runtime for the ZLT, and not counting the first and last 5 minutes, it stayed pretty flat, probably within 5-10% (I just found the link here, thanks quickbeam)

http://thelightsite.cruxial.com/mods/jackp_legend.htm

A lot of it depends on the type of batteries you use, i.e. alkaline or NimH. Is there a reason you need perfect regulation? In normal use the ZLT would give you good enough regulation with Nimh batteries that you probably wouldn't be able to tell the difference between a freshly charged battery or an almost dead one. I haven't gotten around to testing alkalines yet, but I suppose there would be a much more noticable difference. Check out dat2zip's webpage, there's a lot of good info on it with some runtime graphs.

Also, if you're running them off 2AA batteries, the input voltage requirements really aren't that important. If the circuit cuts off at 1.6V or .5V, it doesn't really matter because you're batteries are most likely dead and being able to draw that last 1.1 V isn't going to make a big difference in runtime. I guess if you're running them off a constant voltage source, then it's probably more important. Then again, if you had the LS hooked up to a constant voltage source, then you wouldn't really care about the efficiency.

 

[TripleDouble]

I wish you well on your search Frank. In looking over your numbers again, it seems like the current variances for the ZLT seem a bit wide, especially in CCR-2 format. How exactly were you powering the ZLT @ 1.8v?

 

[dat2zip]

Frank Schwab,

Current state of the art technology in step up converters are 90% mainstream voltage regulators. There are a few exceptions of IC's that are meant to drive LEDs with constant current. Of those 99% of them are geared towards backlighting applications and will drive 1-N 5mm LEDs with 20-30mA of constant current.

Sure, there are many other solutions but none are tiny and fit all the requirements we need to drive the Luxeon. I have four or five that would fit on a 5" X 5" perf board.

Since I've spent more hours than you can possibly count on developing both the MADMAX and badboy I'm rather concerned about releasing on an open forum the details of my badboy circuit. I think everyone on CPF is very understanding of this.

You did not mention the exact details of why you need such tight regulation.

In a flashlight most people probably would not be able to percieve small variations in voltage or current.

To get absolutely constant light output would require servoing the temperature as well as current since both can affect the total light a Luxeon or LED produces.

Just for your reference is a typical badboy configured for 415mA of constant current. The data below shows it hold regulation over the 3V to below 2V with only +/- 1mA of variation.

There are many other secondary effects that I have not touched here. Maybe, this is good enough and maybe it is not.

I believe in all honesty that I've published my specifications, and results on my web site so that you can have all the necessary data to determine if one converter or another will meet your needs.

Even then, if you have any addtional questions, I'd be more than happy to answer them.

</font><blockquote><font size="1" face="Verdana, Arial">code:</font><hr /><pre style="font-size:x-small; font-family: monospace;">Vin Iin Iout Vout Eff
3 0.547 0.414 3.355 84.64%
2.74 0.611 0.414 3.35 82.84%
2.33 0.756 0.415 3.35 78.93%
1.96 0.987 0.415 3.35 71.87%
1.69 1.67 0.425 3.36 50.60%
1.57 1.445 0.319 3.268 45.95%
</pre><hr /></blockquote><font size="2" face="Verdana, Arial">This constant current doesn't mean constant light.

If you heat up the LED or cool it down you would get more or less light. My run time graphs on my web site show this. There is a dip after initial turn on and I attribute this to the LED and flashlight warming up for the first 5 minutes or so. After that it stabliizes and the curve is faily flat.

-WayneY

 

[0tr3b]

Try a bigger battery. Compare the regulated outputs of a D-size (or even a C-size) to a AA at he same discharge voltage level.

Getting the juice out of a battery is like squeezing the juice out of an orange. Say you want one teaspoon every second. Towards the end, you might still have 20 teaspoons left inside the orange, but it is getting harder and garder to get it out, to a point that you can squeeze out only half a teaspoon every second.

A D cell at 1 volt has a lot more energy left than a AA cell a 1 volt. The D- cell can also discherge this energy faster than the AA can because of lower internal resistance.

 

[MrAl]

Hello Frank,

Have you tried adjusting the 1150 ohm resistor
on the Zetex CCR-2 circuit?
According to theory you should be able to
get some pretty darn good regulation with
this simple setup, but you will have to
dedicate some time for trying different
values for the 1150 and the short #32 gauge
wire.
You might replace the 1150 ohm resistor with a 2k pot
but be careful not to adjust down below 500
ohms. If you connect a 470 ohm resistor in series
with the pot you can then adjust the pot any
way you like to test for current regulation
over input voltage range.
The 1150 affects current regulation over
input battery voltage range, and the
short wire affects the output current level
mostly, although there is some interaction
between these two.
I just tested the circuit today and i noticed
that without adjusting the 1150 ohm resistor
the current regulation is better then with the
original zetex circuit, but still not optimum.
This means the resistor has to be adjusted
to get better current regulation, and may or may
not have to be adjusted for each circuit built.
My guess is that if the 5 ohm resistor value
isnt exactly 5 ohms for each and every circuit
built up, the 1150 ohm resistor will have to be
adjusted anyway.
This of course means there are two adjustments
to do now instead of one, but once done, it
shouldnt have to be adjusted again unless you
decide to use a different battery type.

I tested mine today using four D size
alkaline cells, two of which were older
(to get lower input voltage for testing)
and two of which were brand new Rayovacs.
The circuit was run with two new cells or
two old cells to test high input and
low input.

Good luck with your circuit,
Al

 

[Frank Schwab]

Wayne -
I know you've spent a lot of time and money developing your circuits. I have no intention nor desire to step on or steal that. Unfortunately, as a part-time business, you aren't a perfect supplier. For example, last I looked you were out of BadBoys, which you just started offering. Until recently, there was no information on your site comparing the MadMax to the BadBoy, at least specifically related to the quality of regulation between the two. Up until your recent major revamp of your site, all the information on your site referred to the brightness of your two solutions, and how much current they could provide - almost nothing about brightness or current over the lifetime of a battery. Based on the new information you have provided, the BadBoy is exactly the kind of circuit I was looking for. And, at the moment, the form you have chosen to offer the BadBoy in (fully-assembled pill) is outside of my budget for the 4 or 5 lights I want to build now. Believe me when I say that I appreciate the work you've done, the money you've spent, and the knowledge you've made public on this forum allowing me and others to do some cool stuff. Thanks.

How am I testing? With an adjustable bench-top power supply, with four meters measuring guzinta and guzouta voltage and current. I have no access to a light meter, and shouldn't need one.

Why do I want such tight regulation? I didn't think 20% regulation was all that tight, but...
I hate half-bright flashlights. I know I can fix the problem by throwing away the half-dead batteries in the light, but the tightwad in me doesn't want to do that until they're dead. With an electronics package attached to the light, I shouldn't have to worry about it.

Mr. Al -
I haven't spent any time adjusting resistors yet. I wrote my question in a fit of frustration after laying out, etching, and soldering together a board and getting:

I was just a bit depressed. I'll work on the board based on the math you developed in the threads to maximize it's flatness; I've got five sets of parts so this is what I'll likely build my lights up with.

Thanks for everyone's help, from this SW engineers perspective this task just shouldn't be this difficult.

/frank

 

[dat2zip]

Frank Schwab,

Sounds like you have a possible solution to your quest.

-WayneY

 

[MrAl]

Hi again Frank,

I was just a bit depressed. I'll work on the board based on the math you developed in the threads to maximize it's flatness; I've got five sets of parts so this is what I'll likely build my lights up with.

Sorry you had some disappointments there.
I guess i should have made it clearer on
the web site that the 1150 ohm resistor
needs to be optimized for each and every circuit
built up, to make up for variances in the
sense threshold voltage level as well as the
possible variance in the 5 ohm resistor.
I found a pot on DK that would be ideal for
replacing the 1150 ohm resistor with, and then
you can test this by adjusting the value and
see what provides the best current regulation.

TC76B-202-ND [2k pot, 0.1 watt] (for testing for value of 1150 ohm resistor)
470QBK-ND [470 ohm, 1/4 watt] (use in series with the pot above)

Using the 470 ohm resistor in series with the pot
ensures that you wont accidentally adjust down
to too low a resistance with the pot.
The 5 ohm (or 4.7 ohm as i used) should be
SM only.


Thanks for everyone's help, from this SW engineers perspective this task just shouldn't be this difficult.

By now you should have realized that lighting
a light bulb is the most difficult engineering
task on the face of the earth, mostly because
all engineers assume it's going to be so easy:-)

Good luck with your LED circuits,
Al
PS say hi to Thomas and Sophia for me :-)