LED strobe

I have a new project I want to start on.



A high powered LED based strobe.

Will consist of 8 or so XR-Es mounted on a square core inside a acrylic tube. ( much like my video light). The LED's would fire in pairs for a short bust, probably between 1/10- 1/100th of a second then wait for 1/4 -1/2 second before firing the next pair.

The LED's would be over driven by 50-100% using AMC 7135's and powered off a single 18650 or 22650 cell. A capacitor could be added to help smooth the power delivery since a 18650 may suffer at 4 amp even for just 1/10th of a second (???).



If there is anyone out there that wants to help out by programming a PIC for me I would appreciate it. It should be pretty simple requiring 4 outputs from the PIC that pulse. I would like to be able to adjust the timing in some way to get the most out of it.



Now what would be cool is a battery monitoring feature that let you know of the voltage when you turn it on and as it runs down the pulses become shorter and spaced further apart.

This of course complicates the source code some what and would require the use of a ADC.

Im not really sure what length of pulse would be too short but the shorter it is the longer the battery life as well as the ability for the battery/ capacitor to deliver the charge and perhaps the better for over driving the LED's.



Now some quick calculations......



If I drive 2 LEDs at 4 amp for 1/10 of a second twice a second off a 4000ma 22650 cell it will run for 5 hours. Drop it back to 1/100th every second and its 100 hours.

ok, since the forum is back online my post is lost so let's start over
I think I have a PIC floating around somewhere, that I can use to test stuff via software on the computer withouth having to reprogram the PIC. That way the setup can be tested.
The PIC supports measuring 5 channels for a voltage level up to 5V so I can use that to measure 18650 levels on single battery level or series of batteries if I use an extra voltage divider. It also has some digital pins that can be used to generate low frequency low duty cycle pulses.
PICs are probably not really good at driving high currents so I would use this pulse to switch a mosfet (which one? needs to switch fast enough and be able to switch the necessary current). To get alternating LEDs flashing, it may be possible to use a N-channel and a P-channel with the same pulse but I'll need to verify this.

I also have some 556 timers here that I can use to test some frequencies and testing the mosfets even without having to bother with programming things. just a few resistors and capacitors

Once this setup is working, I could get another PIC that doesn't require the PC connection and can run off batteries.

As I mentionned in my lost post, LEDs can be pulsed far beyond their rated currents. How far depends on the duty cycle and frequency. Unfortunately Cree only provides information on the current for 1kHz pulses with 10% duty cycle (1.8A for xr-e). Comparing that with a white osram platinum dragon LED which is also supposed to be run at 1A but produces slightly less lumens, we can see on page 11 that for a duty cycle of 10% (D=0.1) and a Tp of 10-3 (1kHz), we find a current of about 1.6A. I guess we can use this graph for the XR-E as well to get an idea of how far we can push this LED and expect to be able to push the XR-E even further than the Osram LED.

I'll see if I can put something together for testing...

Johan

Excellent Johan.

My thoughts were for the PIC to prive a AMC7135 directly so no MOSFET needed. The AMC will take care of current control. It requires a hi input from the PIC.

I guess if a PIC on a AMC board could be reprogrammed no building of hardware would be required.

EDIT: on closer inspection the AMC boards I have use ATMEL chips but I guess they are similar to PIC's.

Packhorse said:

Excellent Johan.

My thoughts were for the PIC to prive a AMC7135 directly so no MOSFET needed. The AMC will take care of current control. It requires a hi input from the PIC.

I guess if a PIC on a AMC board could be reprogrammed no building of hardware would be required.

EDIT: on closer inspection the AMC boards I have use ATMEL chips but I guess they are similar to PIC's.

Click to expand...

Atmel chips are indeed similar, just need different programming and interface hardware for programming.

Also, for the capacitor to help with the current, 4A for .1s is .4 coulombs of charge, and .4 coulombs of charge resulting in a voltage difference of 1V (from before to after discharge) is .4 farads. So for this to even begin to make a dent in the peak current draw on the battery, youre looking at low ESR supercaps. Something like http://www.cooperbussmann.com/pdf/622e3203-c004-4016-a467-ed7c3b98b5f0.pdf Which are stocked on digikey. I'd go with the biggest you can fit, as I'd guess the batteries would still have a lower internal resistance than the supercaps.

hmm, my amc7135 board don't seem to have a PIC, just some amc chips and a diode.
Can you tell me which boards you plan to use?

ckod, thanks for the info and link, useful.

Johan

The AMC chips I was referring to were the 3 mode ones I got from KD, but any multi mode AMC should use a AMTEL or PIC.

CKOD thanks for the info on caps. I have no idea how to work that stuff out!

I've been sketching on LED strobes to for a while, interesting prospect.

On capacitors, if the solution gets bulky consider put a couple of 18650:s in parallel instead to cope with currents.

Cheers
/O

AW 18650's (2600 mAH) have a 5.2A max discharge rate, and 22650s should be even higher, so you're okay in that respect. That's good, because connecting a battery to an uncharged supercap is even worse than driving the LED. There would be a huge surge of current while charging up the supercap. Dealing with that gets ugly.

Battery voltage sensing is no problem, as most micros have embedded A/D converters. It's certainly not a problem to choose one that does.

The biggest problem you've posed is the desire to control the timing. It's the input device and the ability to monitor it and react in a way that provides good feedback to the operator that makes it difficult. If you want someone else to program the PIC and give you the ability to play (sorry, I mean "experiment") with the timing, here's what I would suggest:

PIC (or other micro) with 3 analog inputs, 4 digital outputs.
One analog input is for battery voltage, two are for potentiometers (variable resistors, aka pots).
One pot causes a change in the pulse duration.
The other pot causes a change in the pulse rate.
Four digital outputs drive FETs that control the LED pairs.

Once you figure out what timings you want, you measure either the timing or the pot settings, then the person doing your programing permanently codes the timing into the micro, and you remove the pots.

I'm mostly a PIC guy, so I'd suggest the PIC16F690. It's cheap, low power, can do everything you want, and is available in packages that range from DIP (can be socketed for easy changeout), SOIC (smaller, but still easy to solder) to QFN (really small, but harder to solder). Certainly there are AVRs and others that can compete with PIC in this arena.

If you want to get into the programming, and start working on the firmware yourself, then I would suggest the PIC16F722 instead. While larger, it's cheaper and allows for in-circuit debugging, which is a significant advantage if the design is complex or you are new to programming.

D

Excellent, getting some great fed back here.

I would like to use Trustfire 18650's since thats what I use in all my lights.
Any idea on the max discharge on them? Perhaps a simple resistor to limit draw from capacitor? But then I guess it will also limit draw from LEDs?
I think limiting it to 2 "channels " of LEDs and AMC's dedicated to each channel.
Also a single 18650 to keep the size down.


Now if some one can do the PIC programming I can do the machining!

I'm looking into the programming, but since it's been a while that I programmed a PIC, this might take some time. Since I was just ordering some components, I'll add the PIC suggested by DIWdiver.
If there's someone else also wanting to give this a try, please do so.

Johan

Packhorse said:

The AMC chips I was referring to were the 3 mode ones I got from KD, but any multi mode AMC should use a AMTEL or PIC.

Click to expand...

If an extra PIC is used, I assume it will also need to take care of making sure the right mode is selected. I have no idea how fast you can switch these drivers without them changing modes. This will probably limit the frequency that they can be pulsed to lower than 1Hz.

No in this case I would just remove the ATMEL from the AMC board or use AMC boards with no chip. Just thought if the ATMEL could be reprogrammed it would mean no PCB to be made up.

You might have decided on the design already, but another thought:

There were lumiled emitters with a extreme side emitting pattern, basically an inverted cone on top of the chip with internal reflection spread the light 360 degrees around sideways.

If any of the new high power emitters came with this optics, it would be an even simpler (albeit weaker) strobe setup.

Cheers
/O

DIWdiver, any suggestions on the FETs?
Maybe we first need to get a better idea on the timing before we can decide on the FETs. I assume it needs to be able to turn on faster that the duration of the pulse it is getting.

What are the FETs needed for?

the digital pins of the PIC that DIWdiver suggested can only provide 25mA per I/O pin. therefore it will not be possible I think to connect them directly to the driver. the fet will be used as a switch that is opened/closed by the digital output of the PIC. at least that is how I see it at the moment. I haven't looked into atmel chips.

just a quick thought.

couldn't an astable multivibrator circuit be used instead of the pic?

im unsure if they can be set up for such a short flash, and may be bulkier though.

i'm looking at that as I have the compontents for that more or less available (missing a diode). it's not the standard circuit to get the short pulses but should work. It won't give you a battery indication when the light is switched on though. Not sure if the frequency changes as the battery goes down.

finished repairing my usb oscilloscope today (bad connector) so I can easily watch the pulses on the computer and play with potentiometers before connecting it to LEDs.

A '555 timer chip could easily make pulses as short and intervals as long as you want. If you wanted to skip the programming, you could build the circuit with some '555's, FETs, and some R's and C's. It would be a lot more circuit than with the PIC, but NO PROGRAMMING! It would involve several chips, in order to sequence the LEDs correctly, but it's nothing fancy. Timing would be adjusted by changing resistor values. I think 4 copies of a ciruit using a '556 (dual '555) would do the trick. Each one would do a short pulse and a long one, then trigger the next circuit. Connect them in a loop, and you're on your way! I could post a circuit if anyone's interested.

With some other chips in the design, you could use a single '556, so only set the timing once, and the other chips could advance the LEDs around.

Personally, I prefer the idea of strobing all four LEDs in rapid succession, then one long delay between. This could be done too.

The '555 circuit's timing is highly independent of supply voltage, and the CMOS versions will work well in this voltage range.

There's got to be a little battery monitor out there that could be added, obviating the need for the PIC.

Whatever circuit you use, the FETs will be the same. For big and easy to work with, I suggest the IRL3502. Very few of these big FETs will turn on well at 2.7-3.0V, which is necessary for this circuit. For surface mount, you can get the surface mount IRL3502S, or I really like the Si4838DY. It's smaller and harder to hand solder, but not too bad. There's no question it will perform very well here, but it has a MAXIMUM allowable gate voltage of 8V, so you can't use it in 12V lights without some extra care.

I could post a circuit if anyone's interested.

Click to expand...

im interested :wave:

im very interested in the pic programming too though, as i feel that is the best way forward