By Richard H Bauhaus

Transistor upper frequency limits (ft) are now in the gigahertz range and are getting higher every day. To design circuits around these transistors using today's techniques, it's essential to be able to make accurate, dependable measurements of their s-parameters.['H31

High-frequency transistor measurements are very simple if you have this equipment:

■ A network analyzer (HP 841 OA) or vector voltmeter (HP 8405A)

■ A compatible signal source (HP 8690B/8699B or HP 3200B)

■ An s-parameter test set (HP 8745A, described near the beginning of the preceding article)

■ Transistor fixtures for the test set (HP 11600A and 11602A) and calibration sets for the fixtures (HP 11601A and 11603A)

■ A transistor bias supply (HP 8717A).

Typical measurement setups are illustrated in Fig. 1.

8717a Service Manual
Fig. 1. Typical setups for measuring transistor s-parameters. The network analyzer system (top) makes swept measurements from 0.1 to 2 GHz. The vector voltmeter system (bottom) makes fixed-frequency measurements from 30 MHz to 1 GHz.
Fig. 2. Transistor fixtures have snap-on pin identifying masks which tell where to insert the transistor leads. The fixtures swallow leads up to 1.5 inches long and take them completely out of the measurement.

Transistor Fixtures

The business end of the measurement setup is one of the two transistor fixtures, Fig. 2. One fixture is for transistors in TO-18 or TO-72 packages and the other is for transistors in TO-5 or TO-12 packages. Both fixtures attach directly to the front of the s-parameter test set and are usable from dc to 2 GHz (the s-parameter test set operates from 0.1 to 2 GHz).

The fixtures are basically precision 50 Q transmission lines that come right up to the header of any transistor whose leads are fully inserted into them. The fixtures can swallow up any lead length from 0.03 inch to 1.5 inches; this takes the leads completely out of the measurement.

Because the transmission lines are fixed, the transistor must be inserted differently depending on which lead is to be common and on the relative positions of the emitter, base, and collector leads. To make it easy to insert the transistor properly, a pin identifying mask can be snapped into place on the fixture and rotated to one of three positions, one for each of the common-lead configurations. The mask shows exactly where to insert each of the leads. There are standard masks for EBC and BEC bipolar transitor headers and for SGD and SDG field-effect transistor headers, and other masks are available on special order.

There are also calibration standards for the fixtures — a short, a 50 Q through line, and a 50 Q load — which can be used to establish an accurate reference plane for the measurement. The reference plane is where the header of the transistor rests when it is fully inserted into the fixture. The calibration standards can also be used to measure system errors. It then becomes possible to achieve very high accuracies by removing these errors from the raw data; this is done in the HP 8541A Automatic Network Analyzer System.[4'-151

Transistor Bias Supply

The transistor bias supply, Model 8717A, was developed to provide a fast, convenient means for applying bias to the transistors in the test fixtures. Fig. 3 shows the front panel of the bias supply. Bias conditions are selected by means of switches and are applied to the transistor by pressing the BIAS ON button. The button lights when bias is applied.

Accuracy and stability of the bias conditions are maintained by a feedback system in the bias supply. The transistor being tested is an integral part of this system, even when the transistor is as much as 10 feet away.

For use in computer controlled systems, such as the HP 8541A Automatic Network Analyzer, the transistor bias supply can be converted into a digitally programmable instrument by installing a plug-in board.

Fig. 3. Model 8717A Transistor Bias Supply has slide switches for rapid selection of bias conditions. The transistor being biased is an integral part of a feedback system in the supply which maintains the accuracy and stability of the bias conditions.

How to Measure

Here's how to measure the s-parameters of a transistor with the network analyzer setup of Fig. 1.

■ Select the proper transistor fixture and mask for the transistor being tested. Connect the fixture to the s-parameter test set, snap the mask into place, and rotate the mask to the common lead configuration you want to use.

■ With the sweeper sweeping from 0.1 to 2 GHz, calibrate the system. Insert the standard short into the transistor fixture. Push the s,, button on the s-parameter test set and select the proper input port (the mask and the fixture tell which port to select); then set the amplitude controls of the network analyzer for a 0 dB gain indication. Next set the reference plane for the measurement by adjusting the line stretcher in the s-parameter test set and the network analyzer's phase controls until the analyzer shows a constant 180° phase shift.

■ Select the s-parameter you want to measure by pushing a button on the s-parameter test set.

■ Set the bias conditions on the front-panel switches of the transistor bias supply.

■ With bias and RF power off, insert the transistor into the fixture.

■ Apply bias and RF power, and observe the results on the network analyzer and the oscilloscope.

How Much Is Too Much?

The higher the input power in the system of Fig. 1, the easier it is for the network analyzer or vector voltmeter to lock solidly to the signals that are being measured. However, since the s-parameters normally are used to characterize a transistor when it is operating as a small-signal or linear device, care must be taken in testing the transistor to avoid overdriving it and distorting the values of the s-parameters. Most of the difficulty is caused by driving the input current, to levels that result in distortion or clipping. The greatest variations in input cur rent occur when s2i is being measured in the common-emitter configuration, so this is when the greatest care is called for.

The quickest and easiest method for determining whether distortion is occurring is to measure i2i at the lowest possible input power level and compare that to a measurement of at a higher input level. If s21 changes, the transistor is being overdriven at the higher input level. Once a good input power level is found for measuring s2i, in should be measured at about the same level. Typically, s22 and s12 can be measured with 10 to 20 dB more input power without distorting their values.

The maximum allowable input power for measuring an undistorted s21 also can be determined by calculation. The problem becomes one of estimating the peak input current, IL1, where clipping occurs or distortion becomes objectionable. A first order estimate of IL1 is

where Ic is dc collector current, p„ is dc beta, / is frequency of measurement, and ft is /3-cutoff frequency. With this estimate of IL1, the allowable input power for an undistorted measurement of s21 in a 50n system can be calculated using this equation:

Printed-Circuit Slide Switches Save Panel Space

Small front-panel size was one of the objectives in the design of the Transistor Bias Supply described in the accompanying article. Faced with the problem of satisfying complex switching needs in minimum space, the instrument's designers chose a printed-circuit slide switch, also designed at HP.

Fabrication of the switch starts with a printed-circuit board on which contacts and interconnections are made using nearly standard printed-circuit techniques. Molded nylon rails are then riveted to the board. Molded acetal plastic slides ride in grooves in the rails, and a wire spring provides detent action along the molded detail on the rails. Palaney slide contacts are mounted in the plastic slides and make contact to the printed contacts on the board.

The switches can accommodate very complex interconnections. They are small, inexpensive, and highly reliable.

Credit for the original concept and design of the switches goes to Ned R. Kuypers and Larry L. Ritchie.

The value of in can be obtained from a measurement at a very low input power level. This equation should predict Pa within approximately 3 dB.[61


I am much indebted to an inspired design team that tackled difficult objectives and succeeded admirably. My associates in the design of the transistor fixtures and accessories were Kenneth Astrof and Robert McCaw. Working with me on the transistor bias supply were Earle Ellis and Elwood Barlow. Roger Wong provided invaluable assistance on the experimental and theoretical work on transistor s-parameter distortion. S


[1] HP Application Note 91, 'How Vector Measurements Expand Design Capabilities.'

[2] HP Application Note 92, 'Network Analysis at Microwave Frequencies.'

[3] HP Application Note 95, 'S-Parameters — Circuit Analysis and Design.'

[4] R. Hackborn, 'An Automatic Network Analyzer System,' Microwave Journal, May 1968.

[5] HP Application Note 99, '8541A Automatic Network Analyzer Measurement Capabilities.'

[6] HP Application Note, 'Large-Signal Transistor S-Parameter Measurements,' to be published.


Sources of information on transistor concepts and modeling.

1. 'Physical Electronics and Circuit Models of Transistors,' Semiconductor Electronics Education Committee (SEEC), Wiley, 1964, Vol. 2, Chapts. 2 and 9, and pp. 32-40, 121-145, and 149-152.

2. J. Millman and C. C. Halkias, 'Electronic Devices and Circuits,' McGraw-Hill, 1967, pp. 369-378.

3. A. B. Phillips, 'Transistor Engineering,' McGraw-Hill, 1962, pp. 289 and 298-305.

4. J. Millman and H. Taub, 'Pulse, Digital, and Switching Waveforms,' McGraw-Hill, 1965, pp. 7-8, 12-15, and 121-126.

5. R. F. Shea, ed., 'Amplifier Handbook,' McGraw-Hill, 1966.


HP Model 8717A Transistor Bias Supply

INSTRUMENT TYPE: Transistor bias supply with two modes of operation, NORMAL and INDEPENDENT. NORMAL is used to bias transistors. INDEPENDENT switches the two internal supplies into a voltage supply and an independent current supply. OPTION 01: Digital/analog converter for remote programming capability. Switching speed <40 ms (typical).

SPECIFICATIONS: See Table at right,

PRICE: $1295.00

OPTION 01: Programmable D/A Converter, add $500.00






continuously variable

variable in 0.25 V steps

dc voltage

VCE(VDs) 0—31.75Vdc 0-31.75Vdc @ 500mA

Same as MANUAL

continuously variable

step = 3.13% full scale current range

4 ranges 3 ranges

Same as MANUAL

0.01- 1mA 0.01- 1mA 0.016- 1mA 0.016- 1mA

dc current

0.1 - 10mA 0.1 - 10mA

D.16- 10mA 0.16- 10mA

1.0-100mA 1.0-100mA

1.6- 100mA 1.6-100mA

10- 500mA @±10Vdc

16- 1000mA @±10Vdc (500mA max output)

Voltage Accuracy

±4% of meter full scale

±(0.2Vdc +2% of programmed value)

Current Accuracy

±4% of meter full scale

±(5/tA + 2% of programmed value)


lE(isj Current


5, 50, 500mA 5, 50mA (+ 20% -0%) (+ 20%,-0%)

Same as MANUAL



Same as MANUAL

1. 3, 10, 30, 100V

Same as MANUAL


(|E' l'C M Current

Same as MANUAL


0.1, 0.3, 1, 3, 10, 30, 100, 300, 1000mA

Same as MANUAL


<0.2% + 20 mV when current varies from 0 - 500 mA

Constant Current

<1.0% when voltage varies from

0 - 10 V (for current ranges <100 mA)


<0.05% + 30 mV with +10% power line change

<0.001% + 1 mA with -+-10% power line change



20 mV

Constant Current

<100 /¿A



Recovery to within 1% of final value <500 /xs (typical)


Recovery to within 1% of final value < 20 ms (typical)


HP Models 11600A and 11602A Transistor Fixtures

FUNCTION: Holds devices for s-parameter measurements in a 50-ohm coax circuit. Either fixture provides common emitter, base, and collector for bipolars. and common source, gate, and drain lor FET's. Other devices also fit the fixtures (tunnel diodes, diodes, etc.).


MODEL 11600A: Accepts TO-1B/TO-72 packages. Will also accept any 3 or 4 lead package with leads that lie on a 100 mil circle and whose diameters are 16 to 19 mils. MODEL 11602A: Accepts TO-5/TO-12 packages. Will also accept any 3 or 4 lead package with leads that lie on a 200 mil circle and whose diameters are 16 to 19 mils.

LEAD LENGTHS: Up to 1.5 inches long. FREQUENCY RANGES: dc to 2 Ghz nominal. IMPEDANCE: 50 J2 ±2 !i. VSWR:

11600A <1.1 from dc to 2.0 GHz. 11602A < 1.15 from dc to 2.0 GHz. CONNECTORS: APC-7* precision connectors for input and output are standard.

OPTION 01: Precision Type N connectors tor input and output.

'Amphenol RF Division, Danbury, Connecticut. PRICE: $425.00.


MODEL 11601A and 11603A Calibration Kits for Transistor Fixtures.

FUNCTION: To calibrate the 11600A and 11602A respectively. Each kit has three calibration reterences. REFERENCES INCLUDED:

1. Short circuit termination.

2. 50 ii through section.

3. 50 S3 termination.

Each is mounted in plastic, leads are compatible with the fixtures. PRICE: $75.00.

MANUFACTURING DIVISION: HP MICROWAVE DIVISION 1501 Page Mill Road Palo Alto, California 94304

W Rich Bauhaus received BSEE,

^jjk MSEE, and EE degrees from ¡SI "Sv H Stanford University in 1959,1960, •<*«* mm and 1961. He has been with HP

jr. ^ P since 1961. After two years in * ' spectrum analyzer design, Rich lit,.,„ f served as project leader for the ¡MW 8400 Series Microwave

Spectroscopy System, the 8745A S-Parameter Test Set and the 4Transistor Fixtures for the Test X Set, and the 8717A Bias Supply.

Xi, Tau Beta Pi, the Involvement Corps, the National Initiative Foundation, and the Sierra Club. He enjoys hiking, backpacking, and sailing.

+1 0

Post a comment