By Gordon McDonald, VK2ZAB
As of 2002, the BF981 mosFET has been replaced by various surface mount mosFETs of similar performance. Some electronic parts distributers may still stock the BF981, for example, Mainline Electronics, UK.
This article originally appeared in the Australian, June, 1984 issue of Amateur Radio. This article may be useful for constructors more comfortable working with larger size mosFETs. Also, some hams may still have spare BF981 transistors.
Introduction
There is no doubt that the Philips BF981 dual gate mosFET is an excellent
choice for two metre preamps. On a performance for dollar basis it was probably
the best device (as of 1985) available. However, circuits published in some
overseas magazines have failed to perform as well as expected. This article
suggests a possible reason for this and provides information on how to
realise the device's potential.
Performance claimed
The Philips data sheets give typical noise figures for the BF981 as 0.7 dB
at 200 MHz and 0.6 dB at 100 MHz. Curves are provided for determining the
source admittance necessary to obtain these figures.
The noise figures claimed for some circuits published in European magazines
are from 1.2 to 1.5 dB and although this is not to be sneezed at, it isn't
as good as the manufactures data sheets claim.
The reason for this probably lies in the type of input circuit used. The
European circuits use the standard coil and parallel capacitor combination
with the input tapped down the coil. It is difficult to adjust this
combination to the point where the input gate sees the source admittance
required for optimum noise figures.
Input circuit
Any component placed in circuit between the source of signal and the control
electrode of the first amplifier will cause a reduction in noise figures.
This applies to any type of amplifier whether FET, Bipolar transistor, or
Vacuum Tube. This fact must be weighed against the necessity to provide that
impedance match demanded by the amplifier for best noise figure and any
requirement to guard against strong adjacent channel interference by
limiting the input bandwidth.
The latter requirement has not been considered in the amplifier described.
Reference to the circuit diagram will indicate that the number of
components is minimal. The bandwidth is as broad as the proverbial bull's
foot.
The sizes of L1 and C2 have been calculated to enable G1 to see that source
admittance which will give optimum noise figure at 144 MHz. This value was
interpolated from the circles of typical constant noise figures for 100 MHz
and 200 MHz given on the data sheets.
Circuit description
C1 should be a low loss, low inductance capacitor. The author's prototype
uses a leadless mica type soldered directly on to the input connector pin.
RFC must be low loss and preferably self resonant just above the two metre
band, ie, it must exhibit high impedance.
L1 is most important. Although a former and slug are used in the author's
prototype, it would be better to use an air cored coil provided that the
necessary equipment and patience are available to enable it to be adjusted
for best noise figure.
C2 must be a high Q trimmer. Best noise figure will be obtained with about
1.3 pF and this is close to the minimum capacitance of any trimmer.
The source and gate, two bypass capacitors, must present an impedance which
is to all intents zero at 144 MHz. There are various ways of achieving this,
but the author's approach is to use two capacitors to make up the necessary
values and different types of ceramic discs. The leads should be as short as
possible and soldered as close to G2 and S as possible. If the lead length
is such that the capacitor can be removed from the circuit and used again,
it is too long.
The same applies to the decoupling capacitor at the cold end of the drain
coil L2 although these are not as critical.
The effectiveness of the source and gate two bypass capacitors may be tested
by running the amplifier while observing the noise figure or by observing
the level of a very weak signal and placing the metal end of a screwdriver
on S or G2. If the noise, noise figure or signal changes, the bypass is
ineffective. In other words the element should be "dead".
No attempt has been made to match the output for optimum gain. With the
circuit shown, the author obtained 23.5 dB, which was considered sufficient.
Attempts to obtain more gain may result in instability.
The values of resistors R2, R3, and R4 are not criical. The requirement is
to be able to adjust the voltage on G2 to 4-7 volts. A seperate source of
voltage with appropriate adjustment may be used if convenient.
Adjustments
First adjust the voltage on G2 so that the BF981 draws 10 mA as indicated by
1 volt drop across R5. If a noise figure meter is available adjust L1 and C2
for best noise figure. Adjust C10 for maximum gain. Note that the gain is
not as important as the noise figure. For example, a typical second stage
noise figure may be, say 5 dB, (optimistic for most transceivers) and you
have a preamp for 1 dB noise figure with 20 dB gain. The overall system
noise figure would be 1.07 dB. If your preamp noise figure now deteriorated
to say 1.03 dB, the gain would have to be increased to 2.25 dB in order to
maintain the same system noise figure. Thus in this example, 0.03 dB noise
figure is worth 2.25 dB gain.
Next try adjusting the current by setting R3, thus G2 voltage for best noise
figure. Up to 13-14 milliamps maximum would not be unreasonable. Readjust L1
and C2 afterwards.
If a noise source is not available, set the current to 10 mA, the slug in L1
flush with the top of the former and C2 just in mesh. You could then try
readjusting each for best signal to noise ratio using the weakest steady
signal available, ie S1 or S2. The frustration caused by this method should
make your next project clear. Build a noise source.
Results obtained
The author obtained an indicated noise figure of 0.61 dB with a gain of 23.5
dB at 144 MHz using an HP 8970A automatic noise figure meter.
At 0.61 dB the slug was almost flush with the top of the former, C2 was just
in mesh and the current was 11.87 mA. As these results and settings are
almost exactly as predicted by the data sheets, there is no reason to
believe that they are not repeatable. However, all this assumes that the
generator is 50 ohms, ie your VSWR must be low; but that's another story.
Good luck and best of DX.
Caution
Some earlier BF981s were sold in a symmetrical X pack. With these it was
difficult to tell which way was up. If your amplifier doesn't work and you
have on of these earlier components, try running it up the other way - it
might still go.
Parts list
TR1 BF981 Philips.
S1-S2 Coaxial sockets to suit your system.
C1 220 pF Ceramic or leadless mica capacitor.
C2 0-3 pF Trimmer - see text.
C3-C4, C5-C6, C8-C9 Ceramic discs. Parallel Combination to make 500-1000 pF. Say one 330 pF and one 470 pF. See text.
C7 1000 pF Ceramic capacitor.
C10 1-10 pF trimmer.
C11 220 pF Ceramic capacitor.
L1 0.331 uH. 6.25 turns 26 guage tinned copper wire on a Neosid Type A former with an F29 slug. No screening Can is used on the former. Space the turns by winding the winding wire on double and then removing one lot. Note that F29 is the only suitable slug.
L2 4 turns 22 guage air spaced 1 cm long on an 8 mm mandrel.
RFC 0.47 uH. See text.
R1 33 ohm.
R2 2.7 K.
R3 5 K trimpot.
R4 4.7 K.
R5 100 ohm.
R6 8.2 K.
BF981. SOT-103 pack.
