VAHA&_E CAPAOTOJtS IV3ICA7E TUNE" Cr^OjiTS
F,XE£> CAPAC*TORS ARt BYPASSES
Fig, 1. Grounded source RF amplifier
Fig, 1. Grounded source RF amplifier
neutralizing indocto« mere «ill gam. decrease stability
TZ SHOULD HAVE HIGHER Jbsi THAI« Ti
step-up), limited of coarse by gain requirements. The cascode circuit is more stable, since the grounded source first stage is driving the low impedance grounded gate second stage. It has the obvious disadvantage of using two of these expensive devices.
Probably the best RF amplifier configuration for all frequencies is the grounded gate. The optimum source impedance for both highest gain and best noise figure is about UK) ohms. Since cross-modulation is caused by excessive gate-to-source voltage, the low impedance level gives better signal-handling capability.
The 2N3823 in this circuit is stable and requires no neutralization through 500 mc. Power gain is between 15 and 20 db. Output impedance is above 50K ohms through 500 mc.
For each circuit, highest gain and best noise figure are obtained at zero gate source bias. Because of the P-N junction contact potential, the zero-biased FET will handle signals up to a few tenths of a volt before gate conduction becomes appreciable. For best resistance to cross-modulation, the FET should be biased so that the gate-source DC voltage is half the cutoff voltage.
To prevent cross-modulation in later stages in the receiver, interfering signals must be attenuated by tuned circuits before reaching stages capable of causing cross-modulation. If the FET RF amplifier is followed by a mixer susceptible to cross-modulation, and the intervening timed circuits are not capable of lowering the interfering-signal amplitude sufficiently for the mixer to handle the signals, the advantage of the FET will be lost. Hence the desirability of an FET mixer. Any of the usual vacuum-tube circuits (other than those using screen-grid injection) can be used; Fig. 4 shows a typical circuit. For maximum conversion gain, the local-oscillator voltage should be close to one-half the FET cutoff voltage (peak) with the mixer biased to half the cutoff voltage. However, for good resistance to cross-modulation, the instantaneous sum oi oscillator voltage and signal voltage should neither drive the gate into conduction nor cut off the FET. Hence it is mandatory that the local-oscillator voltage be as low as is practical, at the expense of conversion gain and noise figure. High-selectivity tuned circuits can then
cut down interfering signals to minimize cross-
modulation in the IF stages.
While all FET's will provide the high resistance to cross-modulation ol the 2N3823, there is no other FET capable of equalling the 2N3823's low-noise high-gain VHF performance. The greatest disadvantage of the 2N3823 is the price: S 12.90. There is a less-expensive version, the i IS34, which is a 2N3823 in a plastic capsule and without guaranteed VHF specifications. A considerable part of the cost of the 2N3823 is in the testing of parameters. DC parameters can be tested very rapidly by machines; RF parameters have to be laboriously tested by human operators, an expensive process. The TIS34 sells for $7.80—still expensive, but a 40% savings. The 2N3823 guarantees a noise figure under 2.5 db at 100 mc, and transconductance minimum 3500 nmho at 100 mc. The probability that a TIS34 will provide RF characteristics equal to those of a 2X3823 is better than 90%.
Most of the competitive N-channel FET's have lower transconductance than the 2N3823, which will result in inferior noise figures. For applications below 50 mc some of these devices may do well enough to be below the atmospheric noise level. It has been stated that with respect to noise an FET is approximately equivalent to a pentode with 3)4 times the transconductance of the FET. Devices fabricated at Texas Instruments with very high transconductance—15,000 umhos and up-had inferior high-frequency performance. Apparently the 2N3823 has about the optimum geometry for RF applications.
Finally, if von intend to try out a TIS34, get a data sheet for the 2N3823 as well, since the latter lias data and graphs not on the TIS34 data sheet.
Was this article helpful?