I recently had the opportunity to meet John Odell, the man behind the family of Dielectric Resonant Oscillator (DRO) Phase Locked Loop (PLL) frequency synthesizers sold under the Herley/CTI name. I told him that hams are big fans of his designs and wondered if he’d share some of the story on their design, modification, and history.
First, for those who aren’t familiar with these units, check out the family of products at Kratos’s site (the new owners of this product line). A search on eBay shows that surplus units are available in a range of frequencies.
Historically, hams started in the 1950’s with klystrons, then in the 1980s moved to Gunn diodes, and “brick” oscillators (cavity type diode multipliers) and some managed to get phase locked sources with that arrangement. Those brick oscillators provided exceptional low noise performance at the cost of long term frequency stability, as most people did not phase lock their ~100 MHz references. In the mid-late 1990s, silicon based synthesizers became decent, (from Qualcomm, National, etc.) at least for small multiplications (from 2.5 Ghz times 4 to 10 GHz). In the early 2000s, the Herley/CTI (and competitors) oscillators became available with much better pahse noise performance and the ability (with dual loop versions) to lock to convenient frequencies like 10 MHz which could be disciplined to a GPS or Rubidium source. This solved the problem of low phase noise and long term frequency stability. Builders are using Herly/CTI DRO oscillators in radios from 10 GHz through 144 Ghz with multipliers, sub-harmonic mixers, etc.
Before getting into the technical details and history of the PDROs, a bit about John Odell himself. I asked John to tell us a little bit about himself and how he ended up designing the Herley/CTI oscillators:
I started in oscillators at Sperry Corporation in Phoenix Arizona in 1985 at the age of 24. I worked on a 9.345GHz Electrically Tuned DRO used in a LNB for weather radar. I was also fortunate enough to be in the advanced development group at Sperry where we designed some of the first successful MMIC’s using the new Triquint foundery service. I designed monolithic oscillators, mixers and amplifiers with a great group of young engineers. From Sperry I went to a GaAs start up in NJ and when it folded, I went to CTI to develop the PDRO product line. I always liked analog electronics, and oscillators because of the nonlinear challenges.
In regards to the origin of the DRO business for CTI:
The DRO business was a natural evolution from the cavity oscillator business, It offers comparable phase noise, at a smaller size, lower cost, lower power with higher reliability. Customers where and continue to be Satcom providers, Military for communications, EW sytems and radar, and Point to Point radio customers. The largest use by far is for Point to Point radio. Back in the 1999-2000 time period there was a big push for point to multipoint communications, and CTI sold 10’s of thousands of these PDRO’s to several companies. Two of the main suppliers got out of the business and some of the old stock got sold to surplus and is available on Ebay.. The 2.25” x 2.25” x .625” became the industry standard back in the early 90’s. it was a good size for frequencies in the 6 to 40 Ghz frequency range.
One of the questions that frequently comes up among hobbyists is how to modify PDROs for different frequencies. On that subject, John says:
The tuning adjustment can be used to lock to additional comb frequency ( multiples of the reference input) for example a unit that is built for 12.3 GHZ locked to an external 100MHz reference can probably be tuned approximately -100MHz to +200MHz and lock to between 12.1GHz to about 12.4GHz at multiples of the 100MHz reference. This should be done while monitoring the phase lock voltage pin on the front of the DRO. As tuning screw is adjusted the phase lock voltage will move from 1V to about 12V for a 12V unit. Once the tuning screw is tuned outside of the lock range the unit will sweep and you will see a clipped sign wave signal at the phase lock pin, continue to adjust the tuning screw until the unit relocks and a DC voltage is present on the phase lock voltage pin. Adjust the tuning screw further to set this voltage at VCC/2 (6 volts for a 12V unit, 7.5 V for a 15V unit). With the dual loop version, you would need to reload the PLL chip on the crystal and change the crystal frequency to change the internal crystal loop, and I would not recommend doing this as it is out f the scope of most Ham hobbyists.
I also asked John about how to modify the oscillators to operate beyond the normal range of the DRO.
The DRO frequency is determined primarily by the Dielectric Resonator Puck. These pucks have a dielectric constant of around 32 to 36 so it is hard to change there frequency with material and the material must be of a high Q, or it will degrade the oscillator phase noise. Some users have used a diamond file to file the edge of the puck towards the back of the unit (opposite the output connectors) to raise the frequency about 2 percent. This should be done on a constant 45 degree angle. You must be careful to not apply too much pressure to cause the puck to come unglued and you must clean and discard of ground resonator powder from the unit using a Q-TIP wetted with alcohol. I MUST WARN YOU THAT THE RESONATOR POWDER IS A KNOWN CARCINIGEN AND SHOULD NOT BE BREATHED IN OR INGESTED, SO I DO NOT RECOMMEND THIS PROCEDURE FOR TUNING THE FREQUENCY.
I solicited a couple of microwave email lists for questions. Several people asked for schematics and some very detailed design information which is obviously proprietary and not available. Here are the remainder of the questions and their answers:
3. Does he have a suggestion of the best way to add varactor fine adjustment to the internal crystal osc so that I can phase lock it to an eternal reference?
I don’t recommend this. I recommend buying an external reference unit and locking it to an external reference that is a sub multiple of the frequency you want to run at and then adjusting the tuning screw to lock the unit to the external reference. Look at this crystal osc. available from Digikey. This and a PLL chip can go a long way. Again Have Fun!
4. Sampling phase detector parts as monolithic assembly of parts have become difficult to find. I’d be curious to know if he’d experimentally compared the performance of discrete and monolithic SPDs.
Discrete SPD’s have been built that work up into the 16GHz range with the correct layout and PLL circuit. Good luck with this and have fun.
5. What would the performance hit of a discrete one would be not using chip and wire?
Most of the SPD PLL’s are made with surface mount components and work up to 18GHz if designed correctly using a monolithic SPD.
I would like to thank John Odell of Kratos for taking the time to answer these questions and provide us with a look into one of our favorite pieces of surplus electronics.