Current source

[GagoX]

Hello everyone,
I would like to share my ideas about the project I am working in right know...may be you have cool ideas too!

I am trying to design a circuit as simple as possible that allows me to control the current that flows through a LED array but at high frequencies (for communication purposes).

for example, using a DC voltage with a sinusoidal voltage source of 1 Vpp in series, a resistor of 10 ohms and one white LED (everything in series) I can drive the current up to 20 Mhz (measuring the voltage in the resistor) (-3db cut off frequency) (basically the speed at which blue light can follow).

Because one LED has a voltage drop of 3.5V typically when I set the DC voltage to 4.5 and a signal of 1Vpp , I get a modulation depth between 100mA and 0mA. (Although white LEDs are rated to work with 20mA, for pulsed operation the stand 100mA)...at least that's what I calculated.

In reality LEDs are non-linear elements, thus increasing the current to 100mA, will increse the voltage drop to 4 volts may be...and therefore with 4.5 volts DC supply I have a very distorted signal...so I can go up a little bit and try..and error, and try, and error, etc.

Of course my problem is that the current depends on the voltage drop of the diode. So I start looking at circuit alternatives.

I am pretty sure you know this circuit...signal comes to the positive pin of a OPA, the the output of the OPA goes in the base of a transistor, which its collector is attached to the cathode of the LED (or anothe...how was it?..never remember the names), the othe pin of the LED to supply voltage, the emitter is conected to a resistor and also to the negative pin of the OPA...a nice controled current source.

now my problems come:

Imagine that I have 10 of this LEDs in paralell...or in some other series/paralell configuration. I would loove to modulate 100mA passing to every one of them...or in other words..modulate 1A..(actually 1A x 3.5 V =3.5W)
(ina series configuration would be less current but more voltage=the same power)

now...CAN I DO THAT?...whith the controlled current source, can I modulate 1 A at 20 Mhz??? Can you think of more options to drive high currents at high frquencies...
Know that I am writing...I think...may be one current source per LED and then conect every positive terminal from the OPA together...(not very elegant ;-)
but then the question is a normal DC source capable to able to give that current at high frequencies??? (I have acces to a very big, modern, cool looking DC source with 0-5V/5A,0-25V/1A in the lab)

Actually the best would be to drive a LED Array of 12, 4 lines in parallel of 3 LEDs in series, whith 166mA. And even better 6 of these matrices in parallel, to have an modulating current of 1A.

I search in google and I didn't find nothing. (May be I should look in the books of electronics?..but I don't have any!

Any help would be very appreciated.

Chau

 

[Analog]

Perhaps a schematic would help cut through the verbage. But from what I understand, you need a voltage controlled current source, that operates at high frequency and high load current? A power MOSFET on the output of the opamp should give you 1A easily. But the question is parasitics at that frequency. Why are you keyed in at 20MHz? Is this frequency important? You'd have to find a decent op amp to work at that frequency, which is possible but also problematic. Please elaborate on the needs and add a schematic to help. Thanks.

 

[GagoX]

Thank you for the reply!

well... hereby I attached the 2 schematics I thought of. The signal is 1 Vpp
and yes...frequency is very important...you see, I want to use the LEDs as communication devices, and getting the most of them is the idea. 20 Mhz would be great! the higher the frequency, the more data I can transmitt. And as you may guess..the more current I am able to modulate, the further the distance I can transmitt. (and I don't want to use on-off keying)

Do you have an idea, what OPA and what transistor I could use...specially the transistor...I think that is the main problem, isn't it?

Is there any other configuration for voltage controlled current source that might be better, in order to overcome the parasitics problems...I am also not very sure how to model the circuit and where I should put a compensation network (lead/lag)...(I think in the feedback path...)...and as I said at the beginning....I would really love to have something simple...and not write a thesis about this.
And by the way...do you know if someone has done something like this?..I cannot believe that I am the first one doing something!

Thank you again for your time

 

[Analog]

I would choose the op amp carefully - a video speed op amp would be the ticket, such as a max477 (300MHz GBWP) or the like.

The transistor choices abound, MOSFET or bjt etc.

 

[audioguru]

Is the voltage of each LED exactly the same? I doubt it, unless they were all made at the same time. You don't know if they are all exactly the same. You could buy extra ones and measure them all then sort their voltages.

What would happen if three 3.2V LEDs are in series and three more LEDs are in series but the second string of LEDs are 3.7V, then the two strings are connected in parallel? The lower voltage string would hog most of the current and would burn out, then the other string would get all the current and it would also burn out. Poof, poof.

That is why LEDs should not be connected in parallel without series voltage-equalizing resistors.

Your circuit has a Mosfet turning on the LEDs but it has nothing turning off the LEDs. They have capacitance that would probably remain charged during the very short time before the next 20MHz pulse, so will not be modulated.

I don't think anybody has tried communicating with visible light, except the old war ships at night. IR is used instead because visible light can be filtered out so it works in daylight.

 

[GagoX]

Thank you for the answers Analog and audioguru.

Thank you Analog for the tip about the max477...it is hard to find a suitable op amp with so many options
And what transistor could I use?....everytime I look for "power transistors" they don't specify the modulation Bandwidth at high currents...(and if so they always refer to on-off switching)...with the circuit I want to use I am always using the transistor on (because of a dc value on the signal)...and the modulation of the current in the collector/source is the one I am interested in.
any suggestions?

and thank you also audioguru for your comments, you are absolutly right about the resistors you saved me some money jeje...
and, about the comment..

"Your circuit has a Mosfet turning on the LEDs but it has nothing turning off the LEDs. They have capacitance that would probably remain charged during the very short time before the next 20MHz pulse, so will not be modulated."


that is exactly myproblem.

well, actually I don't want to turn the LEDs on and off (no pulses)...I really want to control their intensity gradually and at high frequency. The fact that their are Visible LEDs is irrelevant I think, they could be IR too...but my problem (or at least I think) is that I have never seen a circuit that actually does what I want to do now..1A, 20 Mhz.

Anyway, having parasitic capacitances will lead to a decrease in modulation bandwidth(or even worse, not a flat frequency response). If it happend that with a fast OPA and a nice transistor I still couln't get the 20 Mhz...what could I do?...compensating?...how?

the question is again 1A at 20 Mhz..is it possible?

and well, I am surelly not the first trying to comunicate with visible light...try

https://en.wikipedia.org/wiki/Visible_Light_Communications

there is a lot of info about it (in google scholar for example)...very theoretical though...


Thank you very much!!!

 

[audioguru]

I think you will be modulating the charge in the capacitance of the LEDs and their wiring, not modulating the current in the LEDs. Even if something is added to discharge the capacitance, I think the waveform of the current in the LEDs would be distorted.

 

[Russlk]

The question: why 20 mHz has not been answered. Why linear modulation instead of PWM? Light communication is limited to line of sight, you could reach the moon with the power you are trying to generate. Some optics to focus the beam would reduce the power needed.

 

[neon]

I am answering your first question . my lights your lights are modulating at what 50-60 cycles and i cannot see the difference. at hi freq it will be dc they LEDS are diodes put ac across and will rectify.

 

[Nigel Goodwin]

I am answering your first question . my lights your lights are modulating at what 50-60 cycles and i cannot see the difference. at hi freq it will be dc they LEDS are diodes put ac across and will rectify.

You might try reading the thread dates before you post, this thread is well over a year old!.

 

[audioguru]

Neon doesn't understand that the brightness of the LEDs are modulated which produces AM. The DC current that feeds the LEDS is modulated. The polarity across the LEDs is never reversed.

 

[neon]

Neon doesn't understand that the brightness of the LEDs are modulated which produces AM. The DC current that feeds the LEDS is modulated. The polarity across the LEDs is never reversed.

NO YOU DON'T UNDERSTAND how can you modulate a diode for light at those frequency for what purpose? how can you detect modulation at those freq. i ask.

 

[audioguru]

It would be difficult to modulate an LED at 20MHz. It would also be difficult to detect the 20Mhz modulated light or IR.

But TV remote controls are modulated at 40KHz and language translation at a conference uses 100kHz.

 

[Hero999]

Using a sine wave is pointless as it'll just increase the power dissipation.

You're better off with a square wave because it'll save power, then you can just modulate the frequency of the pulses; this is known as PFM (Pulse Frequency Modulation).

And forget 20MHz, I'm pretty sure that white LEDs can't turn on and off that fast enough, for a start the die capacitance prevents this and the phosphor on the surface of the die is far to slow.