Unfortunately, the notes I kept on this job are
scattering is an essential problem to solve. As a first approximation, in this
kind of circuit, BJT collector efficiency is about 60%. As the wished output
power is 100W with 10W drive then the heat peak power to scatter is
((100+10)/0.6)*0.4= ~75W and let say that the average power is about 50W
in SSB. I didn't perform any calculation regarding heat scattering and I let me
guided by experience and intuition ! My investigation leads me to buy a
190x120x40mm heat sink that seemed suitable to me.
q Test of
a MRF5590 or equivalent as first stage, it provides 6dB gain at
q The first tests were carried out with the transistor bases
grounded through a VK200, at 12VDC and with drive reduced to
q Measuring : SWR between TRX and PA and output power on a suitable
50R dummy load.
q Determination of the adjustable capacitors values and
also the values of the essential UNELCO or SEMCO chip
q Implementation of two biasing circuits from a regulated
For SSB linearity purpose at weak signals
transistors have to be biased in AB class.
q Each biasing circuit is a
classic bridge of two resistors supplied by a +5VDC regulated voltage provided
q A negative temperature compensation of the transistor base
current is obtained from a 1N4007 diode directly mounted on the transistor top
package (photo). The reason of this device is that the transistor base current
naturally increases according to the temperature, so it would induce a collector
current increase that in turn would produce a temperature increasing from which
there is a potential risk of thermal run away as well as a linearity
degradation. The 1N4007 diode sees its current increasing according to the
temperature and consequently it's going to shunt a part of the base current. A
perfect temperature compensation is not easy to get, it would require several
trials because it involves the bridge current, the base current and the diode
current. Other methods could be used as well.
q Each bridge resistors are
calculated to have a going through current much higher than the related
transistor base and diode ones and to obtain a collector quiescent current of 12
to 15mA max for T2 and 120 to 150mA max for T3. There is every chance to have
to adjust the resistance connected to +5VDC, this is due to resistor
tolerance and transistor characteristic dispersion.
q An accurate
biasing adjustment requires.
As said previously
linearity requires transistor biasing but also to control input power. Here also
a two-tone test generator and an oscilloscope can show that input power
has to be limited to avoid transistor saturation and therefore splatters
production. To me, according to data sheets, 3W input is a recommended
Results & Comments
some instabilities, FT290R PA (2SC1947) was destroy and replaced by a MRF237. A
220R resistor placed in parallel with the input has been providing the PA with a
reliable stability in the duration.
q Printed air striplines
proved to be operational because the adjustable capacitors I have used gave
distinct tuning regarding minimum input SWR and maximum output
q Adjustable capacitors and chip capacitor value optimisation really
lead to get the expected figures. Actually, when PA and TX are 14.5VDC
supplied, then drive being 3W, the amplifier deliver 115W (CN630 Daîwa) and
overall consumption is about 15A. In this configuration, as the driver
stage current is about 1.4A and other circuits one about 0.4A, therefore the
MRF245 current is about 13.2A this give about 60% for collector
Motorola informs that the MRF245 capability to withstand a 20:1 SWR is decreased
when voltage supply is increase above 12.5VDC and when output power is more than
100W. Consequently make sure that the PA is correctly loaded before
transmitting. The antenna must be connected !!!
q I haven't
tested an intensive SSB operation but at 14.5V the peak power to be scattered is
about P=U*I*40%=14.5*14.6*0.4=~85W and therefore an average of 50 to 60W, so I
think the heat sink I used should be suitable.
q In FM operation,
continuous 85W have to be scattered, and actually after a few minutes, a heat
sink cooling seems advised. A 12VDC fan controlled by a NO thermal probe
(about 60°) both mounted on the heat sink top and the whole enclosed in metal
sheets acting as an air sheath should do the job.
Schematic / Components
q T1 and T2
transistors together with RLY1 form the T/R switching device.
q RLY1 is not
particularly dedicated to VHF and for sure there is better
internal RF links are made by means of 3mm - 50R coaxial.
adjustable capacitors are Philips Teflon and air TRONSER at
q Others capacitors are ceramic 63V, chemical or Tantalum 25V or
q Switches are miniature type, the On/Off one is a triple
model because of the 15A constraint.
q Monitoring lights are 12V
model but as they are 14.5V supplied a serial resistor is used to protect their
Case, PCB and Transistor
q As I kept
in stock a heat sink bar, 100x10mm, I used two pieces for front and rear faces,
they take a small part in heat scattering but they do give a sympathetic look to
the amplifier case.
q PCB is 1.5mm epoxy, double-sided, 35µm copper coated
and its size is 150x100mm. As recommended in the rules of thumb concerning this
kind of unit both board sides are connected together by means of several rivets
and there is no copper beneath printed striplines. The PCB is connected to the
case at four points (photo).
Motorola's transistor SOE
(Stripline Opposed Emitter) packages like the MRF5590 stud type and the MRF245
flange type are straight in contact with the heat sink and tightened in order to
ensure a good heat scattering. Motorola's AN555 note give advises
concerning SOE transistor mount, in short:
- heat sink surface
has to be completely clean and flat
- tightening hole edges have to be
- a thin coat (0.01in) of thermal compound has to be
- recommended tightening torque is 5 to 6 lb-in max
must not exist any mechanical stress on transistor leads.
q To avoid
any mechanical stress on transistor leads, the best way for positioning them as
regard PCB is firstly to tighten them moderately on the heat sink before
soldering them and with the same objective jump at the
opportunity to determine the size and the type of metallic spacers that
have to be used between the PCB and the heat sink.
two soldering lugs (photo) are grounded at MRF245 package tightening
q It goes without saying
that this PA construction applies to experienced and prudent OM
q I wouldn't certify that this PA is perfectly
reproducible as today it's a unique example, however another one is under
construction this would allow to better circle the subject.
q In my
opinion, the output power optimisation is thoroughly linked to the values of
UNELCO chip capacitors, which tolerance could be 20%, so I would recommend to
purchase several additional judicious values.
q Finally, during
adjustment, the chip capacitors will not be definitely soldered otherwise
unsoldering them will be a hard job that will require a 100W iron with the risk
of local overheating causing damages.
q Operation beyond 100W
implies a thoroughly respect of all the above