What is in this article?:
- How to Use Lumped Constants in Microwave Circuit Design
- Transistor parameters
- Design example (continued)
| Download this article in .PDF format |
This file type includes high resolution graphics and schematics when applicapable.
Transistors are obtainable in all types of packages ranging from TO-18 cans to bare uncased chips. Several manufacturers supply units in TO-50 cans or other packages compatible with strip-line or microstrip circuitry. Data on such transistors for use at microwave frequencies are often minimal. Although some manufacturers supply y-parameter curves over an extended frequency range, such parameters are generally given for cased units only and do not reflet the characteristics of the bare chips. The packaged transistor parameters, however, are not bad approximations to determine device performance when wire bonding the bare chips to substrates, because wire lengths are uncased or in a can.
A two-stage transistor amplifier, using lumped-constant passive components was designed to illustrate the techniques involved.
A type TI 2N3570 transistor was selected on the basis of cost and availability, and good datasheet information covering pertinent characteristics at the frequency of interest, 1.0 Gc. Parameters for a TO-18 cased transistor are given on the data sheet, but an assumption was made that the chip bonded to a thin-film substrate exhibits similar y parameters.
The amplifier design is considered an extension to the microwave region of lower frequency design techniques. Thus, the Linvill method was chosen as a starting point. This design technique consists basically of using transistor two-port parameters (in this case, the y or admittance parameters) to determine values of gain for different input and output terminations. After determining the terminations required for a specific gain, matching networks are designed to conjugately match the transistor with its input source and output load impedance.
By calculations, the maximum available gain for this transistor at 1.0 Gc is 8.8 dB; the load termination required for this gain is (2.15 – j7.83) x 10-3 mho. A matching network is designed to transform a 50-ohm load to this admittance. The input source is then conjugately matched to the input admittance of the transistor, which is determined on the basis of the transistor y parameters and the load admittance. Two stages of gain are used to obtain a maximum theoretical gain of about 17 dB. The interstage matching network is designed to match the output admittance of the first stage to the input admittance of the second stage using a 50-ohm load and source.
The completed circuit of the two-stage amplifier is shown in Fig. 1. The power-supply by-pass capacitors are the largest possible for the available space on the thin-film substrate.
In packaging the amplifier, the holder or box was designed with the following criteria in mind:
- small size,
- adequate grounding of the thin-film substrate,
- minimum length of signal leads, and
- reduction of stray capacity.
A cut-away view of the brass case is shown in Fig. 2. Notice that the area directly under the substrate is machined to separate the ground plane from the bottom of the substrate. This is done to minimize capacitive coupling that could significantly alter the performance of the amplifier due to the small values of capacitances used in the design.
The box is machined from solid brass and accepts the rf connectors directly. The thin-film substrate rests on machined ledges, and the ground plane around its periphery is then soldered to the ledge. This operation is performed by heating the box to a controlled temperature and flowing solder around the ledge. Excellent electrical and mechanical contact is made.