Hi Gang,
I'm designing a 160m monoband amplifier using three GU-74B tetrodes and
thought a few of you might be interested in reading about a clever screen
grid voltage regulator circuit I stumbled across. The circuit is the
creation of Gordon Hardman W0RUN, chief engineer for R.F. Concepts (Alpha
amplifiers), and is used in the Alpha 8410 amplifier, which uses a pair of
4CX1000A tetrodes. (For background on the general topic of screen regulator
design, see the excellent article by Ian White G3SEK in October 1997 QEX,
(http://www.muenster.de/~dl5qe/qexartic.pdf ).
Briefly, the Alpha 8410 screen regulator uses two power MOSFETs, one as a
series regulator and one as a shunt regulator, both controlled by a single
op-amp. The series regulator fixes the screen voltage when the tubes draw
positive screen current, and the shunt regulator fixes the screen voltage
when the screen current is negative. The circuit runs very cool. drawing
only about 30 mA in quiescent conditions (resting plate current). Also, a
desirable feature of the design is its relative immunity to internal tube
flashovers. Like active shunt regulators, the Alpha design requires an
external Vcc supply to power the op-amp and provide the reference voltage.
In exchange for this complexity, the circuit offers excellent voltage
regulation and reliability.
To explore the design any further, you'll need to look at the simplified
schematic diagram which I've redrawn for clarity and uploaded to
http://www.w8zr.net/homebrew/Screen%20Voltage%20Regulator.pdf.
(There are some changed component values plus other small modifications in
my version of the Alpha circuit.) The circuit assumes an unregulated input
voltage of 400-450V, and a regulated output voltage of 300V. Note that Alpha
doesn't use a screen current trip circuit or screen current metering in
their tetrode amplifiers, which is why neither is shown in the diagram.
Referring to the schematic, Q1 is an N-channel power MOSFET, rated at 600V
and up to 8A of current, and is used as a series pass regulator. This device
is normally used as an on-off switch, with an "on" resistance of about an
ohm, In this circuit, however, Q1 is used as a voltage controlled resistor
with a typical value of several KOhms. It is biased right at its conduction
threshold, where it is beginning to turn on. This requires a positive
gate-source voltage of 2-3V, depending on the particular device. Most of the
time, Q1 loafs along at under 50mA and dissipates under 10W.
The gate voltage on Q1 is determined by Q2, which is an identical power
MOSFET. Q2 is controlled by the op-amp U1a, which samples the regulated
output voltage from the voltage divider R9 and R10 and compares it to the
reference voltage set by R5. Typically, the voltage at both inputs of U1a is
about +5.1V. Like Q1, Q2 is also operated at its conduction threshold. With
a positive screen current, Q2 is barely kept alive with its drain current
coming only from the 47K dropping resistor R2, which is about 3 mA. Note
that with zero or positive screen current D4 is reverse biased (slightly),
so is effectively out of the circuit.
Let us follow the current path from the input to the output, under
conditions of positive screen current. The screen current flows through R1,
then through Q1, then through D3, and then through R8 and R11 to the screen
grids. The purpose of the 3.3V zener diode D1 is to raise the voltage at the
gate of Q1 by just a few volts. Without D1, the gate-source voltage at Q1
would never rise high enough to reach its conduction threshold. Thus, when
the circuit is supplying positive screen current, Q1 acts as a simple series
regulator and Q2 loafs along to set the gate voltage for Q1.
Now consider what happens when the screen current is negative (i.e., flowing
out of the screen grid). Now the screen voltage tries to rise above 300V,
but Q1 is shut off because D3 becomes reverse biased. Instead, the screen
current flows through D4, into the drain of Q2 and thence to ground. The
op-amp continues to regulate the screen voltage at 300V, which in effect
means Q2 is acting as a shunt regulator. Since the negative screen current
is under 100 mA, the dissipation in Q2 never exceeds (300V)x(0.1A)=30W. What
all this means is that Q2 plays two roles in the design: for positive screen
currents it regulates the base voltage on the series pass transistor Q1, and
for negative screen currents it acts as a shunt regulator.(Recall that
negative screen current is caused by electrons knocked off the screen grid
by the impact of charges flowing to the anode and is equivalent to excess
current flowing out of the screen grid into the regulator circuitry)
Note that the diagram shows a 3 Amp-rated 1N5408 at D4, which may seem odd
since D4 normally doesn't conduct more than about 100 mA. (Actually Alpha
uses a 1-Amp 1N4007, but I changed it to a higher rated diode.) The reason
for the heavier rated diode is so that D4 can handle the current surge in
the event of an internal tube flashover. In a flashover, the screen grid of
a GU-74B can rise instantaneously to the plate voltage, 2.5 KV, sourcing a
negative current pulse (flowing out of the screen grid) of several amperes.
If this happens, the op-amp will immediately switch Q2 into its "on" state
of about an ohm and shunt the current surge harmlessly to ground. Q2 can
easily handle instantaneous surge currents of tens of Amperes, which is the
reason Alpha specifies such a robust device in the application. Z1 clamps
the voltage spike to 400V, which is well below the voltage rating of the
other components. If the current pulse is massive, Z1 could in principle
self-destruct, but if so it fails in a short-circuit mode, which safeguards
the other components from damage. (Z1 costs about $0.50, so it is cheap
protection.) Resistors R8 and R11 limit any current surge to under 40A, and
probably less than 10A, depending on circumstances. Finally, zener diode D2
protects Q1 in the event of a flashover by clamping the gate-source voltage
to +7.5V, which is well below the maximum rating of the device. The 10 uF
electrolytic capacitor C1 stores charge to improve dynamic regulation of the
circuit during SSB voice peaks.
I've breadboarded the circuit and it seems to work quite well. I've been
sufficiently impressed with its performance to give it a try in my homebrew
amplifier. In my opinion, it's quite a clever design. I'll let you know how
it works out.
73,
Jim Garland W8ZR
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|