ANNCXO: Ahead Of The TCXO Curve
By Paul Kruczkowski, Editor
Many remote and mobile platforms have limited power supplies, and thus require components designed with low power consumption in mind. This equipment often requires precise frequency control over its operating temperature, so one component that can significantly impact power consumption is the oscillator.
Oven-controlled crystal oscillators (OCXOs) offer the best frequency stability over temperature. However, the power hungry oven circuit that provides this stability can be problematic for products with power consumption limitations. Temperature compensated crystal oscillators (TCXOs) have found a home in these applications, since they replace the oven circuit with a compensation circuit that draws little current. The trade-off is that TCXOs have never measured up to the frequency stability over temperature of an OCXO.
The primary issue is temperature compensating the AT cut crystal used in a TCXO. The frequency vs. temperature curve of the crystal is approximately a third order function. Despite the advances made in compensation techniques using fifth-order polynomial generators, it is not possible to fit the curve perfectly. These residual errors in compensation limit the temperature stability on the order of ±200 parts per billion (ppb) over -40 to +85°C, until recently.
Engineers at Greenray Industries have devised patent pending technology that replaces the TCXO’s polynomial generator with an artificial neural network to generate the temperature compensation voltage in what they call an artificial neural network compensated crystal oscillator (ANNCXO). In case you are wondering, an artificial neural network is a machine designed to model the way in which the brain performs a particular function or task.
“The artificial neural network is capable of learning the shape of the frequency vs. temperature curve of a crystal, yielding a curve fit with little residual error,” John Esterline, a senior design engineer at Greenray, told me. “The resulting temperature stability is ± 50 ppb over -40 to +85°C, or better, depending on the option selected. That rivals the temperature stability of OCXOs.”
Greenray offers the ANNCXO in a Euro package for frequencies from 10 to 50 MHz and has plans to expand to higher frequencies. The company is also looking to design smaller packages, like 1”x 1” square and DIP, in the future.
While the ANNCXO cannot replace the OCXO in all applications — it can’t match the OCXO’s phase noise performance — it does combine the low power consumption of a TCXO with the temperature stability of an OCXO. The ANNCXO delivers substantial improvements in temperature stability for applications like mobile communications and smart munitions. It also opens the door for system engineers to apply the technology in novel ways.
How would you apply the ANNCXO technology? What new applications do you think would benefit from the ANNCXO? Please share your thoughts in the comments section below.