fastpll James A Crawford

JTIDS Fast Frequency-Hopping Synthesizer

-2-	Ultra-Fast Type-I PLL

One of the very first PLL projects that I had the privilege to work on was for a hybrid spread-spectrum L-band communication system. Even though the system used a carousel of these PLLs, each PLL was still required to hop within about six microseconds. Even by modern standards, this was very fast, but in the circa-1980's, this kind of speed made this synthesizer the fastest indirect synthesizer of its kind in the world.

A limited description of this PLL can be found in the here. Chapter 6 in my first book discussed this PLL in some detail. The material presented here is nevertheless, different.

This synthesizer was the basis for two articles that were written in Microwaves & RF magazine in 1984:

Crawford, J.A., "Understanding the Specifics of Sampling in Synthesis," Microwaves & RF, Aug. 1984.

___________, "Extending Sampling to Type II Phase-Locked Loops," Microwaves & RF, Sept. 1984.

This project was also the genesis that has fueled my 20+ year involvement with PLLs, and the authoring of two technical books on this general subject.

In the 1980's and earlier, a company's proficiency in the frequency synthesis area was a closely-guarded capability. Subcontracts would often be let to outside suppliers for a myriad of different functions, but the frequency synthesis piece was almost always retained inside.

One of the more funny things that happened early in my career at Hughes Aircraft Company was that I had become fairly well known within Hughes for frequency synthesis. I led a small design group at the time that designed most of the frequency sources for the communication systems being built by Hughes Ground Systems Group. After almost 9 years with HGSG, however, I left the company and joined TRW in El Segundo, CA. Shortly after my arrival at TRW, I was asked to sit in on a high-level meeting that was to be held at Hughes Space & Communications Group. TRW was the prime contractor for the MILSTAR satellite whereas Hughes had been subcontracted to provide the frequency synthesis capability for the satellite. The synthesizer project was behind schedule, and I was asked to attend the meeting in order to provide some technical assessment to TRW. The conference room at Hughes was filled with perhaps 30 people, including many high-level managers. Near what turned out to be nearly the end of the meeting, the Hughes technical folks were pressed further about solving the remaining technical problems, and TRW offered assistance. Hughes responded by saying that they had a top-notch expert in frequency synthesis that they were planning to bring into the project themselves to help out; his name was Jim Crawford. Everyone on the TRW side of the table remained absolutely quiet; you could have heard a pin drop. I had no choice but to announce to the room of people that I was that Jim Crawford...and I obviously worked for TRW now. With that shocking bit of transparency, the meeting ended fairly abruptly, and I was warmly invited by all parties involved to help out in any way that I could.

	*Dove (Prototype) Board* A breadboard of the fast-hopping PLL is shown here to the left. The gold rectangular components on the board at the far left (Divide-by-N) and far right (L-band VCO) are chip-and-wire hybrid circuits which had already been miniaturize to this form-factor. All of the other circuitry on the board was eventually shrunk down to the hybrid form-factor. The ground keep-out area near the center of the board was purposely void of ground plane in order to keep parasitic capacitance as low as possible. A very high-speed sampling-bridge occupied this portion of the layout.
	*Timing Sample & Hold Hybrid* The TS&H hybrid was the heart of the synthesizer. The small toroids in the upper right-hand corner of the hybrid formed part of the sub-nanosecond sampling bridge electronics, an area that was patented as USPas 4810904, 4771248, 4668922, and 4634998.
	*Divide-by-N Hybrid* The divide-by-N hybrid is shown here to the left. The large die was a current-hungry bipolar device. The prescaler that was used was generally an 11C90 device which was external to the hybrid. Considerable effort was expended with interfacing the 11C90 to the hybrid. The input power compliance range of the 11C90 was very restricted compared to modern devices. In normal operation, the complete divide-by-N function only had to operate over a frequency range of 340 - 600 MHz.
	*L-Band VCO* The VCO was a fairly straight-forward Colpitts design, except that the tuning-port needed to have a very small time constant. A Wilkinson power divider was used to provide two outputs from the VCO. The main resonator used an air-coil inductor as shown.