Within the Cambridge lab, the Active Bat deployment effectively fills all of the bandwidth available in the ISM 433MHz band. The PEN project operated in the ISM 418MHz band, but this area of spectrum is of limited us now due to the deployment of systems based on the TETRA system.
To support current and future radio projects, it was decided to target the newer, and less crowded, 868MHz band. This band is valid for the whole of Europe; there is an equivalent 915MHz band in the USA.
Our design goals were to:
- Be battery efficient:
- We should provide fine-grain control over power, turning off as much of the design as possible, yet maintaining fast start up and turnaround times.
- Offer "listen before transmit":
- In order to avoid unnecessary retransmission due to collisions with other radios, the radio needs to be able to detect other transmitters. To be as flexible as possible, a 'Received Signal Strength Indication' mechanism should be provided.
- Control transmit power:
- By allowing the transmit power level to be controlled, we allow devices to make much more efficient use of the 'air transport'. Use of RSSI can detect proximity between the sender and receiver, allowing the sender to reduce its transmit power. As well as increasing battery lifetime in mobile devices, this also allows much greater spatial reuse of the band.
- Have no data rate/coding overhead limitations:
- Many existing radios require 'Manchester' or '4b6b encoding to be applied to the data being transmitted. As well as imposing an additional block of hardware to perform this conversion, this encoding also reduces the amount of data that can be transmitted.
There are also the usual radio interference issues: image rejection, frequency stability, bandwidth efficiency, and so on.
Many commercial companies are also targeting the 868MHz band. However, of the designs currently available, none match our combination of high data rate coupled with low power. In fact, all of the first generation of 868MHz radios from other companies have failed to achieve bit rates above 100kb/s, and many have been delayed by months due to last-minute design re-spins.
By comparison, this design has met or exceeded all of our design goals and is available now for use within the labs.
When attempting to produce a power efficient design, methods to conserve power needs to be considered at all levels of design.
At the architecture level, we reduce the power needed by the radio hardware. Our module offers:
- 50% reduction in power compared with the 433MHz module we use in the Active Bat.
- A five-fold increase in maximum data rate.
- DC balancing of transmitted data is not required.
We also provide a number of control signals that allow control at the application level of which parts of the module are enabled, what the transmit power level should be, and to monitor the received signal strength.
To illustrate the power savings possible with this module, one of our designs using the 433MHz module uses a Manchester code to transmit 20kb/s over the 40kb/s link. With the 200kb/s capacity of the new module, this data could be transmitted in 1/10th the time. As, to a first approximation, the power consumed by the radio is independent of the data being transmitted or received, this would represent a ten-fold reduction in the power used by the radio.
Our module offers:
- Efficient PLL based super-heterodyne architecture.
- Based on low-cost RFMD chip, but with redesigned transmit path.
- Data transmission rates from DC up to 200kb/s.
- Received signal strength indicator (70dB range).
- Output power level control.
- Can target the U.S 915MHz band, as well as any of the European 868MHz bands.
Comparing our 868MHz module with the current 433MHz module:
Former Radio Module Our 868MHz Module Power Consumption 65mW (5V @ 13mA) 32.4mW (3.6V @ 9mA) Data Rate 40kb/s Any rate up to 200kb/s Encoding for DC Balancing Manchester or 4b6b required Not necessary
Simulation was used extensively during all stages of the design. This allowed many different design variations to be explored, prior to committing any of them to a costly hardware implementation.
By anticipating potential problems using the simulation, the number of hardware iterations required to produce a working design was reduced to one. The design worked first time and achieved all of our design parameters.
Following the successful proof of the design, two further design iterations were undertaken: one to port the design to a new set of CAD tools, the second to reduce the size of the module for mobile devices.
All RF simulated was performed using the ADS software from Agilent Technologies. For example, the figure below shows a block design of the whole radio design along with the Circuit Envelope simulation schematic for the PLL.
Design simulations were initially performed at the system level. Once a design was fixed, each element of the design was subsequently refined at increasing levels of detail, until the whole design was modelled at the individual component level.
Our main design changes for this module were to the transmit path. As shown in the block diagram above, we use the input data to modulate the reference crystal oscillator. It is this change that allows us to provide DC to 200kb/s data rate with unbalanced data.
The results below show the effect of the reference crystal oscillator modulation. The first figure shows the output of the PLL locking to the input data. The second shows the input and output 'eye' patterns (showing that data is easy to recover).
The final figure shows the Adjacent Channel Power Ratio for the transmitter. This shows that our module is very well behaved and will not interfere with transmitters in the adjacent channels.
The layout was initially performed using the ADS design tools, but was moved to Mentor PCB tools when we attempted to shrink the module. The change in tools has allowed the size of the design to be halved, making it the same size as the Radiometrix BIM module (48mm x 34mm). There are approximately 100 components on the PCB module and the component cost of the module is around $25.
With the power level control set to the lowest level, the modules obtain a range of 3-5m (line of sight) with zero bit-error rate. At the highest power, the modules can provide a link to almost any point in our building.
For more information, please contact Paul Webster ([email protected]).
The 868MHz radio module was designed and developed by Suat Ayoz and Hatice Tuncer. They are about to start at the Cambridge University Engineering Department where they will both be working towards PhDs in advanced RF design techniques.
Suat Ayoz Hatice Tuncer
These pages were last modified 24/10/01.
For comments, suggestions and further information please contact us.
Copyright © 2001 AT&T Laboratories Cambridge