versus three-terminal measurements
versus pulsed I-V curves
How does a
curve tracer work?
all those knobs do?
I-V curve measurements
When you plot the current/voltage
relationship of an element, this is known as an I-V curve. Don't
ask us why "I" is used to denote "current",
that is a great mystery. But the convention is to hyphenate I-V
so it doesn't get confused with Roman numerals.
A resistive element has an
I-V curve that is described by Ohm's law to be a straight line.
Semiconductors have I-V responses that bend, and a a whole lot
more interesting. This is analog electronics at its best!
Before we go any further, you
need to remember the two extremes of I-V measurements: A vertical
I-V measurement means you are looking at a short circuit, and
a horizontal I-V curve means you are looking at an open circuit.
Every other resistor value is somewhere in between.
Two-terminal I-V curves show
semiconductor diode characteristics such as forward conduction
and reverse breakdown. Three-terminal I-V curves show "families
of curves", which is really a two-dimensional representation
of a three-dimensional surface. I-V curves are used every day
for large signal modeling, and bias network design.
There are two categories of
three-terminal devices for I-V measurements: Current-controlled
Bipolar junction transistors
Heterjunction bipolar transistors (HBTs)
Voltage controlled devices
are different types of FETs, and include:
I-V curves are taken so that
we can extrapolate what happens to a high frequency signal when
it is applied to the device that produced the curve. This is known
as large signal modeling, and is a huge topic all by itself. The
leap of faith that must be made is that a high frequency signal,
say 10 GHz sine wave, will experience the same I-V characteristic
that we measure with a curve tracer. This is not exactly the case,
but it is often a reasonable approximation.
versus pulsed I-V curves
"Static" or "DC"
I-V curves refer to curves that are traced at a low rate,on the
order of 100 Hertz (100 traces per second). The problem with static
I-V curves is twofold. First, the self heating that goes on as
the device is swept through the curves is substantial enough to
actually change the curves. This becomes a bigger problem with
large devices that dissipate appreciable power. For power FETs,
the I-V curves start to bend downward at high power levels.
The second problem with static
I-V curves is more insidious than power dissipation. Suppose the
RF curves don't look like the static curves, "just because"?
Semiconductor physicists give us many answers to why this is,
including "surface states" and "traps".
But the goal of the microwave engineer is not to ask why the static
I-V curves are all wrong, but to figure out a way to measure them
in as near as possible to the RF condition. This is known as "pulsed
Someday we will add a page
or two on this topic!
a curve tracer work?
A curve tracer is at the same
time a voltmeter and an ammeter. Here's a block diagram that we
pinched from a Tektronix manual, we hope they don't mind...
There are various models of
curve tracers, but there is only one manufacturer that you should
consider. That is Tektronix. Modern units such as the 370B have
a display that can digitize an image. This is far more convenient
than the old CTs that recorded data with a Polaroid camera. Not
only was this a big mess, but the image was the opposite of what
would be most useful, it had a dark field with white line traces.
Which ended up a smudged mess when you Xeroxed it. Someone we
know once asked Polaroid if they could develop an instant "negative"
film for this type of application. Surely they could have sold
1,000,000 packages of it. But they weren't interested... and now
the company that was founded by Edwin Land, and true Yankee genius,
is all but gone.
The terminals are labeled Collector,
Base and Emitter. For a FET you should pretend Collector is Drain,
Base is Gate, and Emitter is Source. For three terminal parts,
the I-V characteristic is between two of the terminals, with some
control over the curve through the third terminal. There are two
types of three-terminal devices: current-controlled, and voltage
controlled. The xxx knob can be used to generate either voltage
of current steps between the Base terminal (Gate) and Emitter
all those knobs do?
If you really want to become
an expert on curve tracer measurements, there is no substitute
for studying your curve tracer's manual. Here we will review some
of the most important controls for microwave device measurements,
and let you figure out the rest.
These include the horizontal
and vertical axis controls. Here you select the voltage per division
on the horizontal (voltage axis) and the current per division
on the vertical (current axis).
Note that inverting the display
does NOT invert the voltage outputs of a curve tracer. Some engineers
will invert both the display and the voltage polarity during FET
breakdown measurements, so that the curve will start in the lower
left corner and not the upper right.
Step generator controls
Here is where you control either
the voltage or current steps that will generate the family of
curves. You can select current of voltage (use current for a bipolar
transistor and voltage for a FET). You can select polarity, either
positive or negative steps. You can select the number of steps,
and even add an offset (the family doesn't have to start at zero
mA or V) of up to 10 times the step size. You have a choice of
pulse modes, but we recommend you leave the pulse mode off.
Collector supply controls
Here is where you set the maximum
power that your device could see (think of it as a circuit breaker).
You can limit the collector voltage to 16, 80, 400 or 2000 volts.
Unless you are working with GaN, chances are the lowest setting
(16 V) is plenty.
Danger Will Robinson!
don't mess with the 2000 volt scale unless you know how to handle
high voltage! When the little red light is on, you can get the
shock of a lifetime. Our advice is to never exceed the 80 volt
scale, then you can't get killed at work and your kids won't come
in to a pile of money and spend it all on stupid video games...
The maximum peak power is set
by adding a series resistor in front of the DUT (the resistor
is internal to the CT). Older CTs let you choose the resistor
value, on the 370B there are 23 different series resisters, depending
on the max power and voltage that you select (look them up in
the manual if you care). Series resistor setting also provides
a way to "slow down" the I-V curves when you increase
the collector supply; this degree of control which is what makes
curve tracer measurements extremely safe for fragile semiconductors.
Curve traver rule of thumb: Use the lowest power setting (highest
series resitance) that is practical for the device you are characterizing.
The variable collector supply
is located near the output terminals. This is the knob that wither
takes data successfully or does irreparable damage to your hardware.
This is the last knob you set when taking I-V curves, and the
first one you turn down after you capture the data. Turn it slowly,
make sure your data makes sense before you turn it up all the
Output controls and configuration
Here is where you select the
type of curves you are after, such as two-terminal or three terminal.
The configuration knob tells you how the BASE, EMITTER and COLLECTOR
are connected within the CT.
The CT has left and a right
side inputs, these are the same. You can connect two devices at
once, and switch back and forth. Although there are five inputs
on each side, unless you are measuring something with considerable
resistance in the measurement equipment (really long, small diameter
wires), you don't need to worry about using the "sense"
Tektronix newest model is the
370B. This is a great improvement over the older versions in that
it "images" the data for you and stores it in a bmp
file. You can output the file to a floppy disc, and fit maybe
20 I-V curves on a single disc if it doesn't have any other junk
on it. But for $50,000, you'd expect a hard drive and a USB port,
like Agilent is installing on all of their high-end gear. But
this "old school" output has the advantage of instant
bootstrapping, and uses no virus-defeatable Microsoft software,
so it might just be a good trade.
Old models include the 576
and 577. These are just fine to use, but you are stuck using Polaroid
camera to get your image. Or you can use the "Flintstone
bird" to draw the image with his beak:
I-V curve measurements
So far we have three complete
mode MESFET I-V curves
diode forward voltage
Here's some potential topics
we can get into later if there is interest:
PHEMTs (as opposed to MESFETs)
This is not truly a microwave semiconductor, but it has applications
as a current switch.
Also used mostly for its DC switching characteristics