Switching 800mA at 10Mhz?

 

I am attempting to switch an 800mA load at 10Mhz (e.g.with a square
wave). I started with an assortment of BJTs and FETs placing a 10 ohm
carbon resistor load in the collector, (common emitter config). The
squarewave gets pretty sloppy. I added a "speed up" cap to the base
with some improvement however not good enough.
  I am lost in a pleathora of differential drivers, totem poles,
cascodes and so forth, but before I proceed I would be interested in
any designs that may already perform this basic task.

Thanks,
Fritz

 

Probably the biggest problem is that you're letting the transistor
saturate, which will severely limit your speed.  Try using a resistor
in the emitter to set the current.  By controlling the base voltage,
you can set the emitter current via the emitter resistor.  Then, you
can select a collector resistor to ensure that the transistor doesn't
saturate.  This should get you a lot closer...

 

sonnichs@berkshire.net wrote:

You're gonna be wanting to disclose your minimum requirements for:
risetime
falltime
edge linearity
delay rise
delay fall
symmetry
transient response
voltage compliance
load impedance
A description of what you're driving might be helpful.

What are you using to measure the current waveform?
Taking a wild guess that you're not measuring the current at all,
but rather measuring the voltage across the resistor.  How are you
hooking that up.  Probing technique can mess up your measurement.
What's the bandwidth of your measurement system?
mike

--
Wanted, Serial cable for Dell Axim X5 PDA.
Return address is VALID but some sites block emails
with links.  Delete this sig when replying.
FS 500MHz Tek DSOscilloscope TDS540 Make Offer
Bunch of stuff For Sale and Wanted at the link below.
MAKE THE OBVIOUS CHANGES TO THE LINK
ht<removethis>tp://www.geocities.com/SiliconValley/Monitor/4710/

 

Thanks for the replies. Here are some more details. I intend to
modulate a "bright" LED which draws about 600-800mA at 10Mhz. (At some
point I will convert an ethernet signal too OOK and use this as a
transmitter). Conventional devices for laser diodes usually move less
that 50mA.
  This LED has a fast response time of about 10nS so we are comfortable
that this will not be a limiting factor. The vendors of course don't
give a lot of information for these devices. The forward voltage is
3.7V. Dynamic resistance is 0.8ohms. (I presume this is the slope of
the forward voltage). See: http://www.lumileds.com/pdfs/DS46.PDF
  At 10Mhz I am assuming that I can transmit a reasonable square wave
if I can pass frequencies of 100Mhz thus I am looking at rise/fall
times of about 10ns. These comply with some of the FETs and BJTs out
there.
   I am measuring the waveform with a 10x probe via the voltage accross
the resistor. Also picking it up from a PIN diode set up as a
reciever-they seem to comply. Since I get a clean square wave using the
same probe at the sig-gen and base I assume that this is not the issue?
   At any rate I expect it will be easier to operate the resistor or a
diode first as a starting point. I presume driving 800mA at these
frequencies is not uncommon and probably there are some circuit designs
already available as a starting point. I can't find much in my library
-Horowitz & Hill alludes to some things but I would prefer a complete
circuit to start with. I think I have tweaked the most from my present
simple design-a  Darlington driven by a buffer-it gives a much better
waveform than a simple common emitter transistor circuit.

Thanks,
Fritz

 

sonnichs@berkshire.net wrote:

The secret is not to let the voltage move.
Bias the LED  on with the most current you can tolerate for the minimum
intensity.  Then drive the thing with the collector or drain of a
transistor.  The voltage will only have to move far enough to drive the
dynamic resistance.  The base/gate drive you need depends on the
reverse transfer capacitance of the device.  For bipolars, I'd look into
the improved versions of the 2N3866.  Don't remember the numbers of the
newer parts.  Low Cob is what you want.  You might find a cascode
connection beneficial depending on the turnoff edge.  I've achieved an
amp/nanosecond slew rate with 4 IRF530s in parallel
and 100 ohm gate drive resistance for a power supply dynamic tester.

mike
--
Wanted, Serial cable for Dell Axim X5 PDA.
Return address is VALID but some sites block emails
with links.  Delete this sig when replying.
FS 500MHz Tek DSOscilloscope TDS540 Make Offer
Bunch of stuff For Sale and Wanted at the link below.
MAKE THE OBVIOUS CHANGES TO THE LINK
ht<removethis>tp://www.geocities.com/SiliconValley/Monitor/4710/