Understanding the WB5LUA GaAsFET Bias circuit
and using it for PHEMT preamps
or GaAsFET power amplifiers
|
Introduction
Back in 1989, WB5LUA described1 GaAsFET preamps for several microwave bands
which included an active bias circuit for the GaAsFET. Many devices have come
along since then which offer better performance, but require a bias point with different
current and voltages. It isn't obvious how to modify the active bias circuit for
a different bias point, so many folks simply resort to a potentiometer, which usually
works but may drift or fail completely.
Since I've never seen a desciption of how the active bias circuit works, I worked it
out -- it's actually pretty clever. I'll try to describe the operation so that others
can utilize this circuit, not only for GaAsFET preamps, but also PHEMT preamps and
GaAsFET power amplifiers.
Circuit Description
The circuit shown below is the implementation I used for a two-stage2 10 GHz amplifier.
All component references will refer to this schematic. Since it is for a two-stage amplifier,
there are two active bias circuits, each with a 2N2907 transistor and five resistors, R4 thru
R8, which set the bias point of each stage independently.
Everything in the bias circuit is referenced to the input voltage VDD, so it is important
that this is a well-regulated voltage. Normally this is provided by a three-terminal regulator,
U1, which can be a 78L05 to provide 5 volts for a normal low-noise GaAsFET, or an LM317 for
other voltages.
The input voltages, VDD and VSS, MUST be less than the maximum voltage
rating for the GaAsFET or PHEMT or you risk burnout of an expensive device. Some of the
newer PHEMT devices with very low NF have a 3 volt maximum rating, so VDD must be
regulated at 3 volts or less by an LM317, and VSS limited to 3 volts or less by an
appropriate zener diode. Zeners may be hard to find for this low voltage, but a string of
four forward-biased diodes in series would work fine; readily available 1N914 or 1N4148 diodes
are fine.
Next we must determine the desired bias point. A good starting point is the data sheet NF
specification, which usually includes the operating voltage and current for NF measurement.
For instance, an MGF-1302 is measured at 3 volts and 10 ma.
The bias point is set by the resistors in the bias circuit: R4 sets the drain current,
while R5 and R6 determine the drain voltage. The 2N2907 transistor (or almost any small
PNP transistor) acts as a feedback amplifier which adjusts the gate voltage to maintain
the desired drain voltage and current as determined by R4, R5, and R6. Values for these
three resistors to set a particular Vdrain and Idrain may be calculated as
follows:
R4 = (VDD - Vdrain) / Idrain
R5 = R6*(VDD / (Vdrain - 0.65)) - R6
The easiest way to solve the second equation is to plug in an arbitrary value for R6, say
1K ohms, and solve for R5. Then we may change the values of R5 and R6 so that they are
both standard values as long as the ratio of R5 to R6 does not change, and adequate
current, perhaps one milliamp, flows through R5 and R6. (Note: 0.65 is the approximate
emitter-base voltage for the 2N2907).
The gate current in a preamp should be zero, so R7 and R8 may be large; 10K is a
convenient value. Power amplifiers may draw gate current, and amplifiers with a stabilizing
resistor from gate to ground also require current. In these cases, R7 and R8 should
be significantly smaller so that this current does not upset the gate bias.
Example
A simple example may help. Suppose that we wish to operate a GaAsFET preamp at 3 volts
and 18 mA, with VDD = 5V and VSS = -5V. Then:
R4 = (5 - 3) / 0.018 = 111 ohms [110 is a standard value]
if R6 = 1000 ohms, then
R5 = 1000 * (5 / (3 - 0.65)) - 1000 = 1127 ohms [ not standard]
however, if we keep the 1127/1000 ratio, we can find very close standard values
with R5 = 2.7K and R6 = 2.4K.
Spreadsheet
These calculations can get tedious, particularly trying to find standard resistor values.
To make it easier, I made up a small Excel spreadsheet to do the calculations
(hold down the shift key while
clicking here to download), and
entered three different examples as starting points: a GaAsFET preamp, a PHEMT preamp,
and a GaAsFET power amp. With the spreadsheet, it is easy to fiddle things for the
desired bias point with standard or available resistor values. A small adjustment in
VDD is useful in adjusting R4, so R10 and R11 are
included for setting VDD by changing R11.
Power Amplifiers
Use of an active bias circuit GaAsFET (or IMFET) power amplifiers can
help stabilize operation and prevent disasters. Operation is the same as
for preamps, but voltage and current levels can be much higher, so power
dissipation in the components, particularly R4 and U1, is a
consideration. Make sure that power ratings and heatsinking are
adequate.
Safety Shutdown
Most preamps operate at low enough voltage and current so that loss of
the negative bias voltage will not cause damage, but the noise figure
will increase. However, power amps will quickly draw excess current
without gate bias. To prevent failure, a safety shutdown borrowed from
K6UQH3 may be included in the bias circuit. As seen in the
schematic, this is a 2N2222 or other small NPN transistor which forces
the LM317 regulator to its minimum output voltage (about 1.2 volts) if
negative voltage is lost; at 1.2 volts, the power dissipated in the
GaAsFET should be reduced enough to prevent damage. The shutdown circuit
senses the negative voltage through zener diode D2, which should have the
same voltage rating as D1.
Tuning Adjustments
I feel that potentiometer adjustments cause more trouble than they are
worth, unless you are careful to limit the adjustment range, and to wire
them so that intermittent contact doesn't cause a catastrophe. My
preference is to install fixed resistors based on my initial
calculations, then use clipleads to add large resistors in parallel with
R5 or R6 to make small changes in total resistance (adding ten times as
much resistance in parallel reduces the total resistance by roughly 10%).
Optimization is a bit slower this way, but noise figure meters respond
slowly anyway, and I rarely let the smoke out of a GaAsFET. If you use
three pots, you'll probably blow up some GaAsFETS!
Before making these changes, try them in the spreadsheet and see how they
change the bias point. With power FETs, try to adjust for maximum
power with minimum voltage and current; excessive current or voltage may
have unpleasant consequences.
Precautions
The fastest way to destroy a GaAsFET or PHEMT is to apply excessive voltage. Be sure that
the VDD and VSS supplied to the active bias circuit do not exceed its ratings,
and that the SOURCE terminal of the bias circuit is connected to the RF circuit ground.
If you are using a circuit with a built-in negative voltage generator chip, such as a 7660,
remember that the negative output voltage has the same magnitude as the positive input
voltage. Finally, idiot diodes to prevent application of reverse voltage never hurt.
The schematic above only shows the capacitors necessary to keep the three-terminal
regulator from oscillating. Normal RF bypassing and ferrite beads on the wires is are
assumed.
Conclusion
An active bias circuit will provide safe, stable operation even with
varying temperature and supply voltage, with a total part cost less than
any GaAsFET worth using.
Notes
- Al Ward, WB5LUA, "Simple Low-Noise Microwave
Preamplifiers," QST,
May 1989, pp. 31-36.
- Building Blocks for a 10 GHz Transverter (N1BWT),
Proceedings of the 1993 (19th) Eastern VHF/UHF Conference,
ARRL, 1993.
- Bill Troetschel, K6UQH, "Dual Power Supplies for Microwave GaAsFETs,"
Proceedings of the 37th Annual West Coast VHF/UHF
Conference,
ARRL, 1992, p41-51.