The Design of a MMIC Based Oscillator

A SURE Program Research Synopsis

Michael Colwell


The design of a MMIC based oscillator was the goal of my ten week summer research project under the SURE program. To accomplish this goal, I first learned about MMIC technology and microstrip circuit fabrication. Then I applied this knowledge in making basic circuits. These circuits provided the foundation from which I designed an oscillator. Once the oscillator was designed and built, it was tested and its results were examined.

MMIC technology is very new, and is paving the way for communication at high frequencies. MMIC is an anagram for Monolithic Microwave Integrated Circuit. This refers to the way in which the microwave circuit is fabricated. Monolithic, or one-layer, chip design has all the active and passive components fabricated on the same semiconductor substrate. The circuits are very small, and support frequencies upwards of 30 GHz where the wavelenth is measured in millimeters. Applications of this technology include satellite communications, wireless LAN, and automobile radar.

The first learning step taken towards MMIC oscillator design was to build a simple 50 Ohm transmission line. This was done in order to learn how to fabricate microstrip circuits. At Clemson, we used a relatively simple wet etch process. The first step was to design the circuit layout using a CAD tool. I used the PUFF Computer Aided Design for Microwave Integrated Circuits software tool developed at the California Institute for Technology. After the circuit layout was completed, the artwork for the circuit was printed out. Then a mask was made of the artwork. This involved taking a picture of the artwork, and developing the film, or mask, which resembled a photographic negative, where the negative was clear where the circuit was to be, and black elsewhere. A duroid substrate which consisted of two plates of copper separated by a dielectric, was then prepared with chemicals. Photoresist was then applied to the substrate. Then the substrate, along with the mask was placed in a UV box and exposed to ultraviolet light. The substrate was then placed in etchant, where the photoresist coated copper that was exposed to the UV light through the clear parts of the mask remained, while the remaining copper was etched away. The result was a 50 Ohm transmission line microstrip circuit. The transmission line was then connected to a network analyzer and its s-parameters measured. Instead of the ideal 0 dB across the frequency range for the s21 parameter, some slight variations were observed. These losses were due to the radiation effects at high frequencies experienced in microstrip circuits, and also to non-homogeneous connections in the measuring equipment.

After designing, building, and testing the simple 50 Ohm transmission line circuit, I moved on and designed an amplifier circuit. This circuit was very similar to the previous one, except that a break was inserted in the transmission line in which to place the amplifier. The input and output pins of the MMIC amplifier package used were connected to the microstrip transmission lines, while the amplifier itself actually sat in a hole drilled through the substrate to facilitate the connection of the two ground pins to the copper plate on the back of the substrate. This amplifier circuit was then connected to the network analyzer, and its s-parameters were measured. The measurement of its s21 parameter showed that the gain of the amplifier was constant at about 12 dB until 5 GHz, where it dropped to 10 dB, and after that the gain started to decrease rapidly.

Now that two simple microstrip circuits had been successfully designed and built, I moved on to the design of the oscillator. In order to design an oscillator one must keep to one main criteria, that the loop gain must be set to unity with the phase change equal to zero. In the design of my oscillator, a simple transmission line feedback path was used, with the length of the feedback path adjusted in order to set the total phase change equal to zero. I neglected to design the feedback circuit to compensate for the positive gain of the amplifier and bring the magnitude of the loop gain to unity, and instead drove the amplifier into saturation. In the transmission line feedback circuit, a bandpass filter was also inserted in order to ensure that the oscillator would be stable. This filter was constructed using the transmission line coupling of a half-wave resonator strip to set the center frequency at 5 GHz. This filter circuit was designed using PUFF and fabricated on a separate substrate from the amplifier circuit.

As stated before, the feedback circuit length was to be adjusted to set the phase of the circuit to zero. This was done by connecting both the amplifier circuit and the feedback circuit together on a common ground plane. The oscillator was then measured and checked for oscillation, and the filter circuit removed and its length shortened. The filter circuit was then reconnected to the amplifier circuit and checked again for oscillation. Oscillation was first observed at a frequency of 4.682 GHz. Then the feedback loop was shortened again, and oscillation was again observed, this time at a slightly higher frequency. At a length 2 mm less than the feedback length at which oscillation was first measured, no oscillation was again observed. The oscillation frequencies obtained in that 2 mm margin ranged from 4.682 to 4.714 GHz.

The filter's s21 parameters were then tested using the network analyzer to see if it was the cause for not reaching the 5 GHz design goal. It was shown that the center frequency of the filter was in fact 4.7 GHz. This was because in my design I did not take into account the fringing effects of the open-ends of the filter, which was in effect an added capacitance which made the electrical length of the resonator strip longer, thus making the center frequency lower.

Thus the MMIC oscillator design goal for the summer was accomplished. This goal, however, did not come to fruition without the taking of many learning steps necessary to reach that goal. It also did not come to pass without leaving any room for improvement. Thus at the end of the summer, I was left with a great amount of knowledge which I gained through the program, and a challenge to take that knowledge further and improve upon what I have already done.