Need Help with LED Stroboscope

[Avalanche]

I just found these forums through the LED museam, and I am amazed! I didn't know there were so many people who (like me) love flashlights and LEDs.
I'm currently working on an LED stroboscope. I fly micro radio controlled helicopters, and it is very important to know the speed of the main rotor. Weight is a huge issue, so I can't put any sensors on the helicopter itself. To this end, I've designed a circuit with a microcontroller that will flash an LED at the correct frequency to stop the blades. The user selects an RPM value, and the circuit does the rest. My problem is finding the right LED for the job. I'm experimenting with AND190 11,000mcd RED leds,and they are GREAT at DC, but are pitifully dim when used in my circuit. I'm flashing at around 33hz, with a minimum 3us on time. I can make the on time much longer, but if I make it too long, the strobed object gets too blurry to see.

I need a LED (or cluster of them) that will be bright enough to see under these conditions. Room lights will be dimmed, but they have to be on at least bright enought for the pilot to see the helicopter. Cost is a factor, as I might actually make a few of these to sell. I'm not adverse to using color LED's - - in fact I was thinking of using Reds coupled with some of those "laser enhancing goggles" often sold with Laser level sets.

Any advice? I'm leaning toward a 1w red batwing luxeon star/o, but they are pretty expensive. I've also been eyeing some automotive bulb replacement clusters, which sell for $5.00 or so. The problem with those is that the LEDS are wired up in series for 12v, and my device is a handheld unit running off of 4AAs or a 9v.

Thanks for any help you guys can give!
-Adam

 

[Doug Owen]

There's a reason this is not a common product, and you're finding it out. Your duty cycle is very low 3 uS out of 30 mS or .01%. This means you need to push 10,000 times the current in that time window. Looked at another way, you'll need to push 30 Amps through that 5 mm part in a peak. Since this is 30 times the maximum allowed for the Luxeon, you've got problems.

Flash tubes are the traditonal solution. But pricey and a hassle due to the high voltage needed.

But I think you're working too hard. Why not use a condenser mike to 'listen' to it? Or alternatively, use a phototransistor to 'see' it's shadow go by.....

Either could easily be made into a 'window comparitor' (that is 3 LEDs, one too slow, one too fast, one 'in the window'). Then it has LEDs.....

Doug Owen

 

[Avalanche]

Thanks for the reply Doug. I've actually looked into the two methods you describe. The microphone system is a hard nut to crack, as model helicopters actually use 4 blades to fly, 2 large main blades, and 2 smaller paddles. The noise of the 2 sets of blades, coupled with the motor, gearing, and the sound of the tail rotor all combine into a mishmash of sound. It is possible, but would probably need a DSP to work.

People are using the photodiode method right now. Handheld Tachometers based on photodiodes are made for the model plane guys to tach their engines. They work great outdoors, but indoors they only do a good job of reading the 60hz of the room lights. People are actually using flashlights as a method of providing a steady light source on the photodiode.

There is one existing product that does work well, it is a shutter system. Basicly a disk with a slit in it, spun by an electric motor. You look through the disk and adjust the speed of the motor until the blades appear to be stopped. The unit then tachs the disk and gives an RPM readout. Again, this does not work well indoors, plus it is pricy at $140.00.

I knew the LED system would be hard to make work when I started the project, but I figured it was worth a shot. I'm not looking for anything near full intensity from the emitters, just enough light to see the helicopter by. The frustrating part is that the system does work - I can quite easily find the speed of a computer fan within 2 or 3 inches of the leds I'm currently using. I just have to extend that to 5 or 6 feet. Worst comes to worst, It will not be too hard to switch to a xenon strobe system. I'm trying to avoid that due to all the RF noise that an arc will generate - not a good thing to have around when there's a radio controlled helicopter in the room.

-Adam

 

[bikeNomad]

[ QUOTE ]
The microphone system is a hard nut to crack, as model helicopters actually use 4 blades to fly, 2 large main blades, and 2 smaller paddles. The noise of the 2 sets of blades, coupled with the motor, gearing, and the sound of the tail rotor all combine into a mishmash of sound. It is possible, but would probably need a DSP to work.


[/ QUOTE ]

I don't know how fast you need this to work, or whether it has to be portable, but a typical desktop PC is quite capable of doing FFTs in real-time while displaying the results.

Have you tried doing some analysis of the sound of these aircraft to see whether you can pick out the frequencies you need? It may be as simple as just looking at the lowest frequency signal within a certain threshold.

If you need better resolution, you could also look at the "chirp-Z" transform, which is a way of using the FFT to get better frequency resolution.

As far as LEDs go, you would need a fairly stiff drive circuit with very low inductance to get reliable fast switching. If you're using multiple LEDs, consider putting a power transistor and driver for the transistor right next to each one to switch the current.

 

[PsycoBob[Q2]]

If you're looking for 5-6 feet of range in a normally lit room, you may need to use a FET to drive a LS/o, which would work great off a 6/9v supply.

Uh, re-reading bikeNomad's comments, a capacitor on the supply-side of a FET right next to the LS would keep inductance and rapid-discharge current in those small cells from affecting anything. If you grab a bunch of fairly beefy 16v capacitors, and a few power resistors, you can play around a bit. Obviously, start with low pulse-currents. I'd recommend doing a test with a 100ohm ¼ watt resistor before going to normal Luxeon current ranges.

 

[LEDMANIA]

HI Adam,
If you have tried to use the led in measuring the speed of your helicopter using your designed circuit,and it didnt work, why not try to use NE555 timer IC configured as astable multivibrator w/ adjustable frequency and duty cycle.I hope you are familiar w/ this very popular IC. You just connect the LED to the output of the 555 timer,adjust the frequency,aim the light on your chopper and see at what frequency the prop will stop by the light.I use this circuit many times in determining the rpm of my 4 stroke motorcyle engine as well as the rpm of my electric fans..but ofcourse
I also have a handheld frequency counter to check the timer's pulse rate..the 3microsec. pulselength of your circuit is too short for the led to lit. If your circuit has a duty cycle adjustment then increase to at least 50% or more......

 

[Doug Owen]

[ QUOTE ]
PsycoBob[Q2] said:
If you're looking for 5-6 feet of range in a normally lit room, you may need to use a FET to drive a LS/o, which would work great off a 6/9v supply.

[/ QUOTE ]

I still don't think this will work all that well. Even with a wiz bang driver. I seem to have made a mistake in my 'in the head' math above, if you are looking for the same sort of level of light you get from a 5 mm LED at 30 mA (may or may not be enough for the job, depending), and you have a 10,000:1 duty cycle (like we do) you'll need to drive that LS with *300 Amp* pulses. 300 times the maximum the bond wire will take. I bet it doesn't even flash once as bright as needed.

Mind, 300 Amp pulsers are rare indeed, but in a one ohm environment (like we need to drive the LS) you're up against it. Only out I know of is a serious pulse transformer. Hit the primary with a very fast hundred volts or so at a dozen or two amps and you might start getting close. You need pulses in the few kW level to deliver an average .1 Watt to the LED.

Small radar modulators (the bit that drives the magnatron) could do this, as could the conventional strobe drive circuit (replace the flash tube with an appropriate transformer, rectifier and LED).

At least that's how I see it.

Doug Owen

 

[gandbag]

you could try using a fan that spins at your desired RPM with an LED behind it to overcome the 3 microsecond pulse rate. Keep the LED continuously on, and spin a fan how fast you want your rotor to spin. you should be able to make the fanblades thinner or thicker as needed to change the pulse duration.

 

[idleprocess]

Use an IR LED and IR photodiode? Just so long as the IR photodiode doesn't key off any other heat sources...

 

[TheFire]

I guess the other thing that hasn't been brought up is that you could try keying off multiples of the frequency, if you know that you're in the correct neighborhood, since you'll get resonances around those as well. And really, don't feel bad about hitting a 1.5 volt part with 9 or more for a fraction of a second - buy a bunch and see just how hard you can hit them at your required frequency.

 

[LED-FX]

Can get quite acceptable brightness out of an LED strobe....

As Doug has said need to increase current through LED because average power is so low.Don`t entirely agree that you need 300 times the current though...
Winding up pulse width to increase brightness without blurring will help as well.

Loads of schematics about for LED Strobes, here`s one off hand:

http://members.shaw.ca/novotill/LedStroboscope/index.htm

HTH
Adam

 

[Doug Owen]

[ QUOTE ]
LED-FX said:
As Doug has said need to increase current through LED because average power is so low.Don`t entirely agree that you need 300 times the current though...
Winding up pulse width to increase brightness without blurring will help as well.



[/ QUOTE ]

It's true you can get acceptable brightness from *some* LED strobes.

Actually, if you check, you'll see the number I first used (in error) was one thousand. I then realized this was a mistake, the correct number is 10,000 times.

You're right, increasing the duty cycle allows lowering of the level, but the original design constraint forbids that solution, right?

The circuit you cite depends on 10% duty cycle, not .01%, and nearly ten fold overdrive to compensate. We need a thousand times that. That circuit will need to drive at 150 Amps, not 150 mA, with 'our' duty cycle.

The problem here is a well understood one. I happens in radar. If you want miliwatts average and have to live with low duty cycle, you need kilowatt pulses to even get modest (like small boat radar) performance.

A 5 mm LED at 30 mA and 3.6 Volts is about a tenth of a Watt. If we want that average level (meaning as many photons out there to see) delivered in a ten thousandth of the total time, we need 10,000 times .1 Watt, or a kiloWatt of drive for the on time.

This is why conventional strobe lights use flash tubes with very short, high level, pulses. If we wish to do the same with LEDs, we'll have to live with the same rules.

And we haven't even started looking at how we'll deal with the lead and device inductance (which will fight us the whole way).

Doug Owen

 

[tonyb]

I tried using a led for a timming light. It just dosn't have the light output to overcome ambient light, its much easier to use a xenon flashlamp maybe even a neon bulb would be better.