step up circuit help

[TheFire]

Hello all, I'm working in a step up circuit for a very special LED, and was wondering it it's likely that this step up IC would be able to run the LED on 18.5-20v @300mA. I'm not an electrical engineer, but it looks like that's likely from the data sheet. Does anyone more knowledgeable have any input? I think I'll be able to puzzle out a circuit, but any help would be greatly appreciated.

Thanks,
Paul

 

[INRETECH]

What is your input voltage ?

Your input current is going to be much more than 300ma

 

[MrAl]

Hello Fire,

My guess is that this ic would work in your app, most
likely, because they show an app with 12v input and
24v output at 3 amps.

I do have to ask the same question as INRETECH's though:
What is the input voltage?

The input current will depend on your input voltage,
so that Iin=Iout*(Vout/Vin)*1.3 approximately.

Assuming you want to drive your load at the full 300ma,
and if we approximate your output voltage as 20v, if
you use a 5 volt input this means input current will
be 1.6 amps roughly. You'll have to decide if that
is going to work out with your chosen input power source.

Another thing to think about is the package size.
The Max668 is a very small package to work with.

Good luck with your LED circuits,
Al

 

[georges80]

I think the most important thing to find out first is the I/V curve of your $100 UV LED. If you drive it at constant voltage as against constant current - you may end up frying it very quickly! Looking at the datasheet of the UV LED (and I'm generous calling it that) it shows no I/V curves. I would be very cautious in choosing a powersupply to drive that thing - especially at $100 a potential pop...

george.

 

[TheFire]

Ok, hmmm...
Mr Al: That's about what I had thought too, but wanted a second opinion.

As for input voltage. That is TBD, since I need to figure out what will work with it. I was thinking that it would be optimal if I could do it off of 2x123 (3 volts apiece for 6 total), or perhaps 3 D cells (1.4-1.6v a piece for 4.2-4.8 volts total). I know 123's should be able to supply that much current, and I just measured a fresh D cell at 1.866 amps output, so it should be OK for that too, at least for a while. The beauty of using a circuit like this is that you could exchange bodies and use more cells without any circuit modification necessary.

Package size should be no problem, since I have successfully soldered these before without a great deal of trouble.

Thanks for the advice.

georges: First off, it's VERY nonstandard for an LED sheet to include a I/V curve. I don't think I have ever seen one, in fact. The main reason for this is that it varies greatly from one batch to another. The manufacture of an LED die is still a bit of an inexact science (hence the Luxeon Lottery) and so LEDs vary greatly in these characteristics. The other reason that they aren't included is that it really isn't all that important. Overvoltage or overcurrent doesn't kill LEDs the same way it does ICs. What kills LEDs is heat buildup. The packageing on these things is good for 6 watts, so if I do a good job heatsinking it, and actively cool it while it's being tested until I'm sure the current/voltage range is right, then monitor the current on it in a real-life heatsinking situation, I think I will feel comfortable with not having an I/V curves. I spoke to the a rep from the manufacturer of the die (not the assembler of the package) and he said that you really can drive them at 3 or 4 times the rated current as long as you heatsink them properly. I do appreciate the concern, however, and will make every attempt to not fry the LED /ubbthreads/images/graemlins/smile.gif

Edit: Incorrect comment about zeners removed.

Thanks,
Paul

 

[georges80]

[ QUOTE ]
TheFire said:

georges: First off, it's VERY nonstandard for an LED sheet to include a I/V curve. I don't think I have ever seen one, in fact. The main reason for this is that it varies greatly from one batch to another.
Paul

[/ QUOTE ]

But they do - just check any of the Luxeon data sheets - they plot I/V. Of course it is bin dependent - but does give you a slope that indicates just how severely current will increase vs voltage. That will tell you whether it is safe to drive it with constant voltage or whether you really need constant current. Of course heat is the killer - but if your current skyrockets just because of a 0.5V increase then your junction will cook rapidly - this assumes of course that the power supply can deliver the killer current.

Zener for current limit? /ubbthreads/images/graemlins/confused.gif A Zener can be used to shunt excess voltage, not current. If you really want current limit, then decide on a current you want and put a sense resistor in line with LED - and send that back to the FB pin of the switcher. Then you can just decide on the current, pick the sense resistor, work out the op-amp gain to get the appropriate FB voltage and you're in business - that's what I do on my 1W/5W step down luxeon drivers.

D cells will provide 10A easily when fresh - how did you measure 1.8A - across what load resistor? Maybe you meant 1.8VOLTS for fresh D's ?

george.

 

[TheFire]

I stand corrected. Let me revise my statement. Nobody BUT Lumileds puts I/V curves on their datasheets.

Could you point me to an example of the current limit circuit you were describing? This lack of formal electronics training is really getting to me... /ubbthreads/images/graemlins/smile.gif

As for the D cells, I just checked the voltage. They're actually some rather dead D cells... /ubbthreads/images/graemlins/wink.gif sorry... If D cells will really do 10A, then perhaps we could run it off of 1 or 2 D cells... /ubbthreads/images/graemlins/smile.gif That would be a handier size than 3... or imagine a C cell light cut down to 1 cell... that would be VERY cool... but it probably would start having problems at that voltage, even if we are running it in bootstrap mode...

What about temperature limiting too, while we're at it? It would be really nice to shut it off at a certain temperature... any advice on how to do that?

 

[georges80]

[ QUOTE ]
TheFire said:
I stand corrected. Let me revise my statement. Nobody BUT Lumileds puts I/V curves on their datasheets.

Could you point me to an example of the current limit circuit you were describing? This lack of formal electronics training is really getting to me... /ubbthreads/images/graemlins/smile.gif


[/ QUOTE ]

I think any High Power LED manufacturer does/should provide I/V curves.

Ok - here it is - pdf file (88k)

Just pick an opamp that can run single rail and has an operating voltage range that meets you requirements. The one I use runs to 2.7V. You don't need rail-to-rail capability. Another choice (free samples...) is the National Semiconductor LM7301. I've used it in the same circuit, works fine.

george.

 

[TheFire]

I agree that they all should provide the curves. For that matter, any LED manufacturer should include the curves (although it is true that they are all pretty similar for low power devices).

The FB threshold for the one I'm looking is also 1.25v, so that should be good.

I was thinking of using a temperature sensor that increases R3 linearly as the temperature goes up 70°C. Would that be an appropriate temperature solution?

 

[MrAl]

Hello again,

georges:
Thanks for bringing up the I/V curve thing, because i
assumed the end user would already know about this and take
steps to limit the current. I dont believe you have to
bother looking up the I/V curve however, because you can
usually assume a high di/dv for LED's or diodes. This
basically means they all behave about the same with
regard to their I/V curves, so the same precautions have to
be used for any LED, unless of course it has an integral
high value resistor or built in current limit :-)

With this in mind, the Max668 chip has a nice feature
for regulation. It happens to have a 100mv sense input!
This could mean an easy way to regulate current to the
LED, something every LED owner wishes for :-)
Of course you'll have to make sure it's stable once it's
built up, but more then likely there is a way to do it
if not directly with a simple series resistor.

To avoid banging the LED with a high pulse during turn on,
you'll have to check into it's soft start capabilities, of
which i didnt get a chance to read up on for this chip.

Good luck with your LED circuits,
Al

P.S. Almost forgot: I hope you'll keep us informed about
how this project goes for you?

 

[TheFire]

Al: Thanks again for the thoughts. A careful perusal of the datasheet revealed this line:

Once the peak inductor current is selected, the currentsense
resistor (RCS) is determined by:
RCS = 85mV / ILPEAK
For high peak inductor currents (>1A), Kelvin sensing
connections should be used to connect CS+ and
PGND to RCS.

Will this be enough to set the current? Will the peak current be around the amount I want to drive it at (so I just calculate that with the ILPEAK = my peak desired current and plug a standard resistor in with a value that's similar)?

It says this about it's soft start:
The MAX668/MAX669 feature a “digital” soft start which
is preset and requires no external capacitor. Upon
start-up, the peak inductor increments from 1/5 of the
value set by RCS, to the full current-limit value, in five
steps over 1024 cycles of fOSC or fSYNC. For example,
with an fOSC of 200kHz, the complete soft-start
sequence takes 5ms.

It seems that that will probably fill the bill nicely for soft on. Thanks again for the input.

georges: Thanks for the circuit. I'll implement that if we can't get the on-chip current sensing to work. Your reason for recommending the LM7301 is also my reason for choosing a maxim IC /ubbthreads/images/graemlins/smile.gif...

Al: Of course I'll keep you updated. Hopefully there will be a big splashly thread in the modified lights forum for it in the semi-near future /ubbthreads/images/graemlins/smile.gif...

Thanks again,
Paul

 

[MrAl]

Hello again Fire,

[ QUOTE ]

Will this be enough to set the current? Will the peak current be around the amount I want to drive it at (so I just calculate that with the ILPEAK = my peak desired current and plug a standard resistor in with a value that's similar)?


[/ QUOTE ]

I took a look at the full data sheet and found out that
the spec they were quoting for the 'sense' input
voltage threshold is pretty much the same as for the
Zetex circuit: it's for the inductor current alone,
not for the output current.
There is an input for regulating, but it's not 100mv,
rather, it's 1.2 volts! This is a bit higher, but
should still work under the same principle. It will
require a 4 ohm resistor (Iout=300ma) and will eat up
5 percent of the efficiency, but that's not too bad
i guess. Otherwise, you would have to build up a small
resistor network like the ZLT-CCR circuit has, and do a
little experimentation with resistor values.

If you decide to use the 4 ohm resistor (probably a good
idea at least for the first run up) you connect it
in series with the ground end of the LED, then connect
the junction of the resistor and LED to "FB" on the chip.
This provides the chip with about 1.2 volts when the
current reaches 0.3 amps. You'll have to do some stability
testing on the prototype.

If you are using the 669 chip it's a good idea to read up
on the bootstrap section too :-)

Good luck with this LED circuit,
Al

 

[INRETECH]

With that expensive LED, I would keep it VERY simple and very reliable, such as using a LOT of 2AA and a LM317 in constant-current mode

 

[TheFire]

Yeah, but the brightest flashlight in the world is useless if it sits in your toolbox. I know you can fit that many AA's in a 3D maglight, but it's really clumsy, and AA rechargeables are painful to charge in that configuration, and it would be overall nicer if it ran on standard batteries. btw, Wayne (dat2zip) says he's gonna see if he can modify a badboy to run these things... I'm still going to see if I can get this circuit running, though...