Help: Wideband LED Current Source, 1 AMP swing

[Eric_M]

I am attempting to design a wide-band current source that can pass a 20 Mhz PAM signal nearly un-distorted into a string of LEDs. The catch is that I need it to swing from 0 to 1amp to power the Luxeon Rebel brand of LED's.

The circuit I have so far is attached. It is a line driver configured as a Widlar current source with a bias tee.

The device I am using is the OPA2674 DSL Line Driver. It is capable to delivering +/- 500 mA to a 50 Ohm load, and +/- 800 mA to a short circuit load, and has a 200+ MHz gain-bandwidth. This part for sure gives me the 1 Amp swing I need, but it's bi-polar, hence the bias tee.

The issue I am having is understanding the OPA2674's data sheet. It states it has a maximum supply current of 18mA, however if the device is driving +/- 500 mA into the load at +/- 4V (as the data sheet claims), that current has to come from somewhere, and must go through the device itself. So, is this "18mA" only when the device is unloaded?

I ask this because clearly when I probe the supply lines in LTSPICE, they average current being forced through them is +/-250mA for a periodic input, which far exceeds the 18mA max. Aside from this power issue, the circuit works wonders, but I am wondering if its too good to be true.

 

[GUMY]

Supply current is stated for no load condition as You presumed.

 

[FvM]

A bipolar amplifier has no problems to source an unipolar current, if used in a suitable circuit. The bias tee isn't actually necessary.

 

[Eric_M]

Thanks for the help, the amount of current drawn makes sense now. I suppose once I properly heat sink this thing in practice all should go well.

Also, I failed to mention that each LED has a forward voltage of ~ 3V, so I think I need the bias tee for headroom reasons. Does anyone see any potential issues I could have doing this? Without it I would have 1 line driver to 1 LED which is not practical in my case.

My goal is to modulate an LED light bulb (5 to 10 LEDs) with no more than 2 line-drivers.

 

[FvM]

Also, I failed to mention that each LED has a forward voltage of ~ 3V, so I think I need the bias tee for headroom reasons. Does anyone see any potential issues I could have doing this? Without it I would have 1 line driver to 1 LED which is not practical in my case.

AC coupling is O.K. for constant modulation respectively DC balanced modulation signal, but it will cause varying offset with any kind of variable modulation signal average.

I see that the OP won't be able to drive the average 1A LED current, so direct DC coupling won't be an option for OP2674.

 

[Eric_M]

@ FvM

Thank you for the advice. Luckily for me, the standard I am working off uses Manchester Coding =)

Additionally, I will also be using this for testing an OFDM scheme. However, I assume if I choose the capacitance properly, the DC drift will not be an issue as long as it doesn't change drastically during 1 OFDM symbol.