Slashdot Mirror
New Full Duplex Radio Chip Transmits and Receives Wireless Signals At Once (ieee.org)
Wave723 writes: A new chip by Columbia University researchers uses a circulator made of silicon transistors to reroute signals and avoid interference from a transmitter and receiver that share the same antenna. This technology instantly doubles data capacity and could eventually be built into smartphones and tablets. The chip enables them to work around the principle of Lorentz Reciprocity, in which electromagnetic waves are thought to always travel along the same path both forward and backward. Traditionally, electronic devices required two antennas -- a transmitter and receiver -- that took turns or operated on different frequencies in order to exchange signals.
The gist of what is clever here is the canceller which removes the transmitted signal from the receiver. Circulators have been around for donkey's years (not just in military systems) but they are bulky (especially at lower frequencies such as those for mobile comms). The are often used to allow a single antenna to operate at both transmit and receive either alternately (e.g. radar) or on different frequencies (e.g. satcom). Making a solid state one is clever, but this isn't the first one.
However, some of your transmit signal will always end up in the receiver for three reasons; (a) the circulator isn't perfect, (b) the antenna doesn't have a perfect match so some of the transmit energy sent to it bounces back again and (c) energy can reflect back from the immediate environment. Cancelling schemes exist, and invariably consist of some mechanism for sampling the transmitted signal and feeding just the right amount back into the receiver exactly out of phase. In theory this works, but in most practical circumstances the extremely high level of cancellation needed requires a completely unachievable precision.
For added pain, the solution tends to be very narrow band and the cancellor's settings have to be continually updated as the transmit interference changes (particularly in a mobile environment due to (c)).
If they have managed to make this work in a practical and useful way then it will be very impressive, but I would need to see some real world experiments to be convinced of its practicality.
Active circulators with an arbitrary number of ports have been known of for decades but they have to operate within the noise, power, and linearity of their active devices. When used to separate a transmitter and receiver on the same frequency, their limited isolation will cause major problems on the receive side. Their big advantage is that they can operate over wide bandwidths which makes them very useful for test instrumentation.
This design was published in 1991 and would be useless in this application. I ran across it when it was included in one of my microwave engineering books.
I suspect the refinement they came up with involves a second adaptable stage to cancel feedthrough and near end echo but in order to do this, the output of the final amplifier has to be sampled to include its high levels of noise and even if it all works perfectly, all of the noise contributed in those circuits will get added to the receiver. I looked at doing something similar by sampling an earlier stage but just the noise contributed by the final amplifier was enough to overload the receiver. You can see this effect when powering up a SSB transmitter with no modulation which promptly raises the noise level for the entire band.