The intention of this project was to create a speed detector based on a traditional Doppler shift radar gun. The design should be portable, low-cost, accurate to within 1 mph, and comply with FCC regulations. In order to accomplish this, it was necessary to have a signal source, a receiver, a mixer, a frequency to voltage converter, a micro-controller, and an output display.
Frequency to voltage
Project Block Diagram
The emitter sends out a sinusoidal signal that encounters a moving object and is reflected back to the receiver. The reflected signal has a different frequency than the original signal due to the Doppler shift. The reflected signal and original signal are sent to a mixer where the two signals are subtracted from one another. The resultant output is a signal whose frequency equals the Doppler shift between the original and reflected signals. Thus, the Doppler shift frequency is now represented as its own signal coming out of the mixer. This signal is sent to a frequency to voltage converter. The Doppler frequency is converted to a voltage and sent to a micro-controller which calculates the speed in miles-per-hour that this voltage level represents.
There are several fundamental tasks involved in this design. The emitter and receiver need to have antennas large enough to target a one square foot area, or a high enough frequency to target this small area, or a combination of the two. The signal needs to be strong enough to give the detector a one hundered foot range. The frequency to voltage converter must be properly calibrated to give the micro-controller usable values. The micro-controller must have a software program downloaded into it that will calculate the speed in miles-per-hour that is represented by a given voltage.
The speed detector calculates only the magnitude of the speed, not the direction. This is due to the fact that the Doppler shift between the original and reflected signal is determined without regard to which of the two has the greater frequency. Also, since lateral motion will not produce a Doppler shift, the speed detected is the speed towards or away from the dectector, with no regard to motion in the lateral directions.
The micro-controller calculations can determine the speed of an object to within 1 mile-per-hour. The signal used must not violate FCC regulations and must be powerful enough to give a 100 foot range to the detector. The speed detector is designed to be portable and self-standing.
2. DESIGN PROCEDURE (tjd, jaj)
Many design options were considered in the design of this radar speed detector. The components and issues that were considered were the frequency, the transmitter and receiver devices, the demodulation device, the processor, and compliance with FCC regulations.
Transmitter frequency is a large aspect of the design. A goal was to make this device portable. It also had to be able to detect relatively small objects at a distance, such as any car or motorcycle. Portability was a factor to consider also. The ability to detect any motorized vehicle forced the wavelength to be at most 1.5-2 meters, or the average size of a car. Portability suggested a handheld device. This latter consideration pushed the frequency to the medium to high megahertz (MHz) and low gigahertz (GHz) range.
Next, consideration was given to a device that had a separate receiver and transmitter. Finding either the transmitter or the receiver does not pose a large problem. The biggest drawback to this design is the received signal. There are very few, if any inexpensive devices that will demodulate a GHz frequency directly. A separate mixer would be needed to implement this device. The mixer down converts an inputted high frequency to a more manageable lower frequency. This mixer would mix the received signal from the receiver with the original transmitted signal. This would give an output frequency centered at the base band, or 0Hz. For example, if the transmitted signal is 10GHz, and the received signal is Doppler shifted up 30 Hz to 10.00000003GHz, the output signal from the mixer when the two signals are mixed would be 30Hz, or simply the Doppler shift. This design, with two antennas and a separate mixer can become costly.
Another signal source idea was to have one antenna act as both the receiver and the transmitter. This eliminates one antenna completely, but still requires a mixer. A circulator would be needed to distinguish the outgoing from the incoming signals. A circulator alternates input and output sources rapidly. It alternates from a state of sending the transmitted signal to the antenna to a state of sending the received signal out to the mixer and then alternates back. The average price for a separate circulator is at least $100. This price applies for frequencies too low to fit the project go and is too costly for this project.
Fig. 2. Gunn-plexer Block Diagram
(tjd)Integrated packages were reviewed. This component would have one antenna, a source, a circulator, and a mixer in one package. With the help of Prof. Franke in the ECE department, one type was found. These devices are called Gunn-plexers, named after the man who invented the first Gunn diode about 50 years ago. These devices range in sizes and are available with one of two basic frequency outputs, 10.5GHz and 24 GHz. They range in price from about $20 to $150 depending on the power output. The output power varied anywhere from 7mW to 15mW. The Gunn diode outputs a signal at a HAM frequency, typically either 10.5 or 24 GHz. The signal is output at a low power to comply with FCC regulations and has a range of several hundred feet.
Using the Gunn-plexer for this project has many benefits. The design of an antenna for both the source and receiver is taken care of. The high frequency means that the use a very large antenna isn’t necessary, which aids in the portability of the design. The portability of the design is further enhanced by not using two separate components as the source and receiver. The output signal from the mixer gives the Doppler shift as a sinusoidal signal centered on the base band.
The next component needed for this project was a demodulation source. This is a device that will convert the base band output from the Gunn-plexer to a voltage or a digital output. Availability and versatility were key in this device. The EE storeroom has available a National Semiconductor LM2917N frequency to voltage converter. This converter has a "programmable" output. It’s programmed by external capacitors and resistors that select the frequencies that are to be used.
The next component is the processor. Two basic types exist for use, microcontrollers and PCs. This project can be hooked up to any PC using a serial or parallel port. An external interface would have to be designed as well as the program for hardware implementation. Portability is also a concern because a laptop was not available. There is no external interface needed with a microcontroller, however. There are several microcontrollers available in the ECE department. There is an 8-bit MC68HC11, a 16-bit MC68HC12, or a 32-bit MC68332. The chosen processor does not need to be very complex, because there are only a few minor calculations, data manipulations, and inputs and outputs to be used in this project. An MC68HC11 was selected for use. An additional incentive was that a take home unit was available, further aiding portability.
The output display was the last item to be considered. Numeric LEDs would be ideal to display speed data. Standard numeric LEDs need external current pumps because LEDs consume a good deal of current and therefor cannot be driven by the HC11’s outputs. The EE storeroom had Texas Instruments TIL311 numeric displays available. These LEDs have their own internal circuitry and LED current pumps internal to the package. The 4-bit data input is read directly into the display logic. Hexadecimal values from 0 to F are capable of being displayed. The input is the actual value (in binary) that is to be displayed. The obvious advantages to these displays are that the user does not need to control each LED within the display to output the desired number and external power supply circuitry is not needed. In a normal 7-LED numeric display, the user needs to control each LED section. Using these displays cuts down debugging time and complexity of the circuit.
3. DESIGN DETAILS (jaj)
The components used to design the radar speed gun are:
1 (one) 10.5 GHz Gunn-plexer
1 (one) LM2917N Frequency to voltage converter
2 (two) TIL311 Hexadecimal display with logic
1 (one) Motorola 68HC11 evaluation board
The Gunn-plexer is the cornerstone of the entire project. It functions as both a transmission source and a receive antenna. There is an internal circulator that directs the send and receive signals. The Gunn-plexer outputs a continuous wave at a frequency of 10.5 GHz. The received signal is channeled from the circulator to a mixer where it is mixed with the original signal. The output is then at a base band frequency. The output is the Doppler shifted frequency. The Doppler shift is calculated from the 10.5 GHz signal. The Doppler equation is:
Vd=V v/c (equation 1)
where Vd is the Doppler shift (Hz), V is the original frequency (Hz), v is the target velocity (miles per hour (mi/h)), and c is the speed of light (6.818 10^8 mi/h). With V=10.5 GHz, the Doppler shift (Vd) of a target moving at a velocity of 1 mi/h directly toward or away from the radar beam will be 15.4 Hz. This frequency is the magnitude of the Doppler shift and therefore a target moving away from cannot be distinguished from a target moving toward the Gunn-plexer. This is fundamental to the Gunn-plexer.
The LM2917N converts up to a 1 kHz signal to a voltage between 0V and 28V. The supply voltage sets the maximum output voltage. Our supply voltage was set at 5 V. This allows an easy interface into the HC11’s A/D converter. Capacitors and resistors, external to the LM2917N are used to set the voltage step per unit frequency. To calculate the proper values it must be known what digital values the HC11 will convert an input voltage into.
The HC11 has an 8-bit A/D. The maximum voltage that will be measured from the LM2917N is 5V and the lowest is 0V. These values are set using the HC11’s Vrh and Vrl inputs respectively. These are convenient values because they can be connected to Vcc for 5V and ground for 0V. The resolution of the A/D is given by:
HC11res= (Vrh-Vrl)/2^8 (equation 2)
The numerator is the voltage range, and the denominator is the number of steps that the range will be divided into. HC11res = 19.53 mV/bit. To obtain a total resolution of 1 mi/h for the final output, for every 15.4 Hz increment of Doppler shift, the voltage from the LM2917N needs to increment by 19.53 mV. With the same voltage range on the LM2917N as the HC11’s A/D, the maximum frequency that can be inputted into the converter is 256*15.4 Hz or 3942.4 Hz. This gives 15.4 Hz per 19.53 mV, or 1 mi/h per LSB. The equation to calculate the resistance and capacitance of the external components of the LM2917N is:
Vo=Vcc*fin*C1*R1. (equation 3)
Vo is the output voltage for the given input frequency and supply voltage, Vcc is the supply voltage, fin is the input frequency, and C1 and R1 are the capacitance and resistance values of the external components respectively. The only unknowns are C1 and R1. Vo=5V @ 3942.4 Hz with a supply voltage of 5V. Choosing an arbitrary value for R1=1kOhm, C1=.254 microFarads. The converter has a maximum input frequency of 1kHz. This equals about 65 mi/h, which is adequate for testing and proof of concept for our experiment. At 65 mi/h the converter outputs 1.234V. The entire range of 0V-5V is not used, but it is convenient to keep the conversion of 1mi/h / Least Significant Bit (LSB) increment.
The HC11 does not need to do any calculations to convert the results from the A/D to a usable value because the value in the A/D register is the target’s velocity in binary. It only needs to convert the value into outputs for the LEDs. The LEDs have their own onboard logic, so only the binary value that is wished to be displayed needs to be written to the inputs of the LED. The onboard logic figures out what components need to be displayed inside the unit for a corresponding value. The display is capable of displaying outputs from 0 to F. However, only the values of 0 to 9 are to be of concern.
A divide procedure was used to convert the 8-bit A/D result into 2 decimal values. Only two LEDs were implemented because the maximum convertible Doppler shift by the converter only has two digits. The most significant bit (MSB) for the LED is left in the X-index register, and the LSB is left in the double accumulator after the HC11 completes its integer divide instruction. This is an efficient way of performing this data manipulation because the divide instruction tells us how many times that 10, or the MSB LED, can go into the mi/h value. If the display were of another base, such as base 3 or an octet, the mi/h value would have to be divided by 3 to get the proper input for the 2nd LED.
The value of double accumulator is logical shifted left 4 times. This places the binary value of the LSB for the LED into the upper 4 bits where it will be written out to Port B of the HC11. The value of Accumulator B, which is the 8 least significant bits of the double accumulator is saved to a location in memory, and the contents of the X-index register is read into the double accumulator. Accumulator B is AND-ed with the binary value 00001111. This leaves only the 4 least significant bits which is the value to be read into the MSB LED. This value in Accumulator B is OR-ed with the value just saved into memory, leaving the MSB LED value in the lower 4 bits of Accumulator B and the LSB LED value in the upper 4 bits of Accumulator B. This was done "backwards" because of our protoboard layout. There were not long enough wires to make connections between the upper 4 bits of Port B and the inputs to the MSB LED. The contents of Accumulator B are then written to Port B and connect directly to the LEDs.
The LED is a Texas Instrument TIL311 chip. As previously stated, it has its own onboard circuitry that controls the actual LEDs inside the entire unit. There is no clock or other external timing device. The LED is changed as soon as the logic can calculate the new value. The latch strobe input and the blanking input from both LEDs were both connected to Port C of the HC11, and were controlled through programming. While the latch strobe is low, the LED circuitry changes the display to the current input. If the latch strobe input is set high, the LED will not change regardless of the input. The LED will display the last value before the latch strobe was set high. When the blanking input is high, the LED is blanked (nothing is displayed) regardless of the input. If it is set low the LED is allowed to display an output. Both the latch strobe input and the blanking input were set low for both of the LED’s before the main loop of the program. The complete circuit diagram is shown in Appendix 1.
4. DESIGN VERIFICATION (tjd)
Due to some unforeseen problems, all testing has not been completed on the project. The burnout of the Gunn-plexer before testing has limited the amount of data collected on the efficiency and error of the speed detector. This problem can be attributed to the inability of the Gunn-plexer to operate for an extended period of time at voltages outside of its tolerance range (~ 8.8-9.3 V). When using a 9 V battery as a power source rather than a DC voltage source from the lab, the drop in voltage over time comes into effect. After some operation time, the voltage of the 9 V battery drops to well below 8.8 V. Operating the Gunn-plexer at this voltage level for several minutes caused an irreversible burnout of the device.
In order to prevent this problem from reoccurring, it is necessary to avoid operating the Gunn-plexer at voltages below 8.8 volts. However, to keep the portability of the project, a 9 V battery is an ideal source. Since voltages above the tolerance range of the device are not a concern when using a 9 V battery as a power source, one solution to the problem is to design and build a voltage turn off switch. The switch would cut off the current to the Gunn-plexer when the power source falls below 8.8 V. In order to do this a special Zener diode circuit, designed to have a reverse breakdown voltage of 8.8 V can be used. Another possible implementation of the voltage cut off switch would be to design a BJT circuit that turns off the current to the emitter when the base current falls below a value corresponding to an 8.8 V source.
Before the burnout of the Gunn-plexer, debugging of the entire project was performed. The output values were determined to be well outside of the desired accuracy range. This has been attributed to two types of miscalibration. First, the hardware values used did not exactly match the theoretically calculated values. To calibrate the frequency to voltage converter, a 28.57 F was needed but a 25 F capacitor was used. To remedy this situation, a variable capacitor can be used. Also, the program initially used to calculate the speed of the target contained an incorrect value in a calculation, producing erroneous results. The program has been corrected and will now output the correct data.
To show the functionality of the frequency to voltage converter, the linear relationship of the input frequency and the output voltage is shown in Fig 3. The frequency does not reach the desired 1000 Hz due to the miscalibration of the capacitor value. The figure does show that the properly calibrated converter will rise fairly linearly to 5 V at the maximum frequency value (1000 Hz when properly calibrated).
Fig 3. Linearity of Frequency vs. Voltage (tjd)
5. COSTS (jaj)
LED output display (x3)…………………………….$27
(50 hours x $35/hr) x 2.5……………………………$4375
(50 hours x $35/hr) x 2.5……………………………$4375
The major cost of the project is labor. The labor for the project is broken down as shown in Appendix 2.
6. CONCLUSIONS (tjd)
While the project as a whole is non-functional, the problems have been determined and addressed. With the purchase of a new Gunn-plexer, the design of a voltage switch or reliable 9 V source the signal can be sent and received and the Doppler shift frequency determined from the mixer output without worry of burnout. The use of a variable capacitor carefully calibrated to 28.57 F will remove the error from measurements taken. The computer code has been debugged completely to perform its desired function. With these changes, the speed detector should be fully functional and meet the goals of being portable, low-cost, accurate to within 1 mi/h, and complying with FCC regulations.
Port E Pin-0
APPENDIX 1. CIRCUIT DIAGRAM (jaj)
HC11 Port C2
HC11 Port C0
HC11 Port C3
HC11 Port C1
HC11 Port B5
HC11 Port B4
HC11 Port C6
HC11 Port C4
HC11 Port C7
HC11 Port C5
HC11 Port B5
HC11 Port B4
APPENDIX 3. HC11 PROGRAM (jaj)
* Define parallel port names as actual port addresses
PORTB EQU $1004
PORTC EQU $1003
PORTE EQU $100A
* Define registers needed for serial communication
SCDR EQU $102F
SCSR EQU $102E
SCCR1 EQU $102C
SCCR2 EQU $102D
BAUD EQU $102B
OPTION EQU $1039
DDRC EQU $1007
ADCTL EQU $1030
ADR1 EQU $1031
LCD EQU $0000
LDS #$F000 ;sets top of stack
LDAA #%00110000 ;sets 9600 baud rate
LDAA #%00100000 ;turns on A/D lines
LDAA #%11111111 ;sets Port C to all outputs
LDAA #%00001100 ;sets blanking and strobe on LCD's
BEGIN LDAA #$00
LDAB ADR1 ;set the double accumulator to the A/D value
IDIV ;divide MPH by 10
LSLB ;moves value to LCD2
ANDB #%00001111 ;selects only LCD values
ORAB LCD ;moves value to LCD1
STAB PORTC ;sets LCD
APPENDIX 4. REFERENCES (tjd)
C. Elachi, Introduction to the Physics and Techniques of Remote Sensing. New York: Wiley-Interscience, 1987.
R. Uribe, "Digital Systems Laboratory," class notes for ECE 249, Department of Electrical and Computer Engineering, University of Illinois at Urbana-Champaign, Fall 1998.
National Semiconductor, LM2907/LM2917 Frequency to Voltage Converter, National Semiconductor Corporation, 1995.
Texas Instruments, Hexadecimal Display with Logic, Texas Instruments Incorporated, 1997.
D. Neaman, Electronic Circuit Analysis and Design. Chicago: Irwin, 1996.