By Dr. Robert L. Howald, Manager, TNS Systems Engineering, General Instrument Corporation
Traditional HFC systems have implemented the return path, a 5-40 MHz bandpass in North America, using bidirectional RF amplifiers in the coaxial plant, and a return path laser in the fiber optic node, driving a return fiber for the optical trunking. At the hub or HE, the optical power is converted back to RF. The technology is the same analog AMbased optics approach used to transport the broadcast forward path video signals. There are numerous design and implementation issues that make this approach difficult and costly. These include analog laser specifications, laser second order response characteristics, optical link length constraints, optical receiver specification, and testing of the components. All of these contribute to the overall cost issue of developing high performance analog optics. Node laser issues are exacerbated by the fact that this component must operate in an outdoor environment, specified over a very wide temperature range.
An improved approach implements a digital transport method at the point that the RF plant terminates, and the optical trunk begins inside the node. To do this, the analog laser technology is replaced by a high-speed baseband digital technology of the type that has been used in the telecommunications industry. As a digital signal, immunity from the troublesome analog laser impairments is obtained and longer distances can be covered, potentially avoiding the need for hub repeater hardware required in analog systems, among other benefits.
This paper describes analytical and design issues associated with digitizing the return path at the node. It can be shown that the analog-to-digital (AID) converter is a mathematical analogue to the AM modulated laser technology traditionally used. We can treat the AID converter quantization noise as the effective "analog" optical link noise. This can be correlated with the known performance capabilities of the lasers currently used. Additionally, the distortion performance, in particular the laser clipping aspect, is replicated identically by AID input thresholds. Conveniently, a complete technology upgrade can be achieved, while the key concepts of the mathematical analysis remain virtually unchanged. The second order and third order distortions also can be kept very low in AID's and, furthermore, the second order distortions do not degrade to poor values as analog laser can. This can be an important issue in broadband applications. Measured performance has been taken of NPR and of BER on a loaded return, showing how the dynamic range analysis above can be applied to the digitized return as it is used in analog systems. Ultimately, the strength of the digitizing technique is furthered by the functions and processing that can be applied once the information is represented completely as bits. Some of these concepts will be introduced.