Our proposed design is shown below as a block diagram layout in Figure 4.0. As can be seen, it consists of two separate boards connected by an 8 wire cable. The main board, on the right, contains the microcontroller that will be used to control the various devices and compute a direction to be displayed. Also on this board are all the peripherals the team feels are required to accurately gather data and troubleshoot any problems that may arise. The daughter board, located on the left, contains the array of AMR IC chips used to gather data about the direction of the applied magnetic field. It also contains other minimal hardware required to operate the IC’s and a laser to precisely depict and set the orientation of the daughter board.
Figure 4.0: Proposed Block Diagram Layout
The board layout has been broken into two separate boards to serve multiple purposes. First, having the components on separate boards make each board smaller and less bulky. This will become very handy when it becomes time to mount the product to the rotation table for testing. Having the micro on one board the AMR IC’s on the other will enable the Daughter Board to be remained mounted while the Main Board can be removed and have its flash updated or be modified. Separate boards also serve as a safety net. If one boards fails, or has been constructed wrong the entire project is not lost. As an added bonus if the desired orientation/layout of the AMR ICs is wished to be changed a new daughter board can simply be constructed. The design will also allow for variations in number of IC’s as well as orientations without the reconstruction of the main board. Using two boards should also make the group’s testing and configuration processes easier. Having the sensors on a separate smaller board will allow us more easily use reflow soldering on sensor array which will help the sensors align more accurately to their PCB pads. Once the sensors are mounted to their board, the daughter board as a whole can be treated like and individual sensor which will allow for easier breadboard level testing and calibration.
 Main Board
The main board as shown to the right contains the all the equipments to control and collect data from the AMR ICs located on the daughter board. Great lengths have been taken to ensure that this board can handle everything required while at the same time be very versatile and adaptable in case something has been overlooked or requires a change.
The microcontroller is the key component to this board and the entire project. It is the work horse, and will be used to communicate over I2C to the AMR ICs as well as to a computer running hyper terminal or LabVIEW via a serial connection. At the same time it will also be required to display pertinent data to the onboard LCD and status LEDs. To increase the usability of the board a built in programming port has been added along with an area to probe various control lines and pins for troubleshooting purposes.
The parallel connector will require a minimum of 11 data lines to be sent to the Daughter Board. They are as follows: 1 laser power, 1 for all ICs power switched via BJT,1 for the mux power, 1 to sense power voltage, 2 for I2C, 4 (minimum) for the power mux control, and finally a ground connect. The uses for each of these connections will be discussed in a following section. It is important to note that the power mux is used to turn on individual ICs, therefore to support the addition of more ICs on the Daughter Board in the future, more mux lines will be required. For this purpose the labeled 11 data line will probably contain extra wires connected to the microcontroller not used in this configuration.
 Daughter Board
The daughter board, shown at the right, contains the AMR ICs that will be used to determine the magnetic field direction with an onboard laser used to pin point the board direction from which to be calibrated. Along with these it contains its own optional power source and power mux control system.
Because the ICs communicate over I2C they are all connected in series and need to be addressed individually when sending commands. The issue with this is that from the factory they all contain the same name address and there is no way of communicating to each one individually. The power mux allows each IC to be turned on individually in order to reprogramming the IC’s name address such that when they are all turned on each one can be addressed individually. An added benefit to the power mux is that it minimizes the number of lines required to be connected to the micro controller. At the same time, many more ICs can be controlled with the addition of minimal mux control lines if future designs require such modifications.
It can be seen that the power mux also has two available sources. This has been done for multiple reasons. First, having all nine ICs powered by an output of the microcontroller will draw to much current for the mux and the microcontroller to handle. The power for when all ICs are on will be routed directly to the power source on the daughter board and connected by a BJT transistor via a line controlled from the microcontroller. This allows the board the flexibility to operate independently of the main board without worry of voltage drop across the connecting cable. As an additional feature, the parallel connector could also then be plugged into a wireless device, such as one using the ZIGBEE protocol, and be made completely separate from the mother board for uninterrupted orientation swings.
The placement of the individual IC sensors on this board are placed every 10 degrees between the interval of 0 to 90 degrees for a specific reason. Consider the following diagram shown below in Figure 5.0 where 9 sensors are laid out 10 degrees apart.
Figure 5.0: Optimal Axis’s for AMR IC Sensors Spaced 10 Degrees Apart
If the AMR IC has two wheatstone bridges inside rotated by 90 degrees from each then they will have two optimal axes where the magnetic field direction reading will be most linear and accurate. These two axes are shown in Figure 5.0 by the blue and green lines for each IC, as also indicated by the legend. As can be seen rotating the IC by 10 degrees provides the smallest interval between consecutive optimal axes while remaining uniform and without doubling up in any one place throughout a full 360 degrees. The same can be concluded for the orientation shown below in Figure 6.0 where the chips are spaced every 20 degrees apart.
Figure 6.0: Optimal Axis’s for AMR IC Sensors Spaced 20 Degrees Apart
Both of these orientations should provide the same data and however, in the second case the optimal axes alternate between a bridge A (blue) and a bridge B (green) where as in the first case there are nine consecutive bridges A optimal axes followed by nine bridge B axes before repeating.
 How It All Works
The two boards shown above will be used to simulate both a single AMR IC sensor compass as well as a multiple AMR IC compass. This will be done in software by only requesting one sensors data or all sensors data and performing similar analysis on each the retrieved data. The results will be done simultaneously ( or consecutively) and display side by side on the LCD display and possibly transmitted to the a computer via the serial connection.
The entire product operates on a fairly simple concept. Use as many sensor as possible orientated such that there optimal, most linear, most accurate regions of measurement are spaced evenly about a 360 degree area. A magnetic field is then applied to the device and the sensor measuring within their reasonably linear and accurate region will be polled by the microcontroller and the direction measured saved. The measurements can then be combined and averaged to make a more accurate reading than an individual IC could do on its own. Currently, the thought is to weight each sensor’s measured value based on how close it is to measuring from within its optimal axis and then perform an average of their readings. The measurements can then be repeated and averages together over multiple readings before outputting a final value to remove any noise and obtain the best measurement possible. Multiple algorithms for performing this have been discussed and are currently in the process of being coded into a Matlab simulation for analysis.
 Setup & Configuration
With the current Main Board proposal setup will be extremely streamlined and easy. The built in programmer will allow the microcontroller to be securely fastened to the board and never need to be removed, removing the risk of breaking pins. It can be programmed on board and then tested and reprogrammed at will. This will allow quick and fast code modifications, changes, or updates through testing and final demonstration if need be.
As hinted at previously, setup consists of being able to address each IC sensor individually over I2C. Due to the fact that each sensor is set to the same address name from the factory, if they are all connected in series and powered on they cannot be communicated with individually. To solve this problem a 1 to 16 MUX has been incorporated onto the Daughter Board and used to power each sensor individually. With the ability to power each sensor individually they can then one by one be powered on and given assigned a new unique address name from which to be communicated with from that point on. After each IC sensor has been give an address they will then be powered by the external power source connecting all of them together and controlled by a BJT transistor controlled by the microcontroller because the mux cannot handle enough current to feed all the ICs at once.
 Data Acquisition & Computation
Retrieving measurements from the AMR ICs will be relatively easy. The current plan involves addressing a single IC over I2C and retrieving its raw X and Y bridge voltage differentials. These data will be saved and then the next IC sensor will be called and the process repeated until all sensors have been called. From this point the data can either be analyzed, but if time permits a it would be beneficial to quickly gather three sets of raw data from each sensor to be averaged to remove noise and jitter.
From the non-averaged, or averaged (if applicable) raw data mathematical operations will be performed to determine to overall magnetic field direction. The data for a single IC sensor will first be analyzed to determine the most accurate possible direction by means of checking which of the two on board wheatstone bridges are most within its linear range and making a weighted average. The result will then be saved and send the LCD screen. Once the single IC direction has computed, the saved raw data from all the chips will be combined in a similar fashion and the result displayed to the LCD screen next to the single IC result. The process will then be repeated until the device is shut down.
 Meeting Requirements
In order to meet the remainder of the requirements stated above, testing procedures need to be determined to compute the specs for both the individual IC and multiple IC compass readings. The following test plan describes preliminary ideas as how to determine accuracy, precision, and repeatability.
Definition: Accuracy is most easily understood and is usually reported as maximum possible difference from the true value. In the team’s case with a compass it can be expressed as how near to the true reading does it point, example +-5 degrees.
Plan: A firm plan has yet to be determined to compute the accuracy of the team’s device. In order to do so the team needs an instrument with greater accuracy than the team’s device. Future talks with Andy should determine what device the team can use. Once a higher accuracy piece of equipment is obtained it can be compared with that of the group’s laser pointer output which can be sent a significant distance and broadcast on a wall where any minute error will be amplified.
Definition: Precision can be defined as the smallest unit increment a product can measure. For example, if the team’s compass is rotate a known 10.0 degrees, and the compass readout changes by 10.1 degrees it could be said to have a precision of 0.1 degrees. This is different form accuracy in the sense that it may have a very precise reading, lets say within tenths of a degrees, but may have a large offset representing bad accuracy.
1. Create a uniform field in the correct range for the group’s sensors using the Hemholtz Coils. 2. Using a laptop/LabVIEW interface to the High Precision Rotary Table the team can get a precision angular reading for when the device reads 0° 0' 0" or some other recordable value. 3. Rotate the table, again very precisely, and compare the known change in the table to the change in the team’s devices output. 4. The device can output a reading calculated from the 9 sensors and also a reading from a single sensor for comparison.
Definition: Repeatability may seem as if it is a function of accuracy and precision but it is not. The best way to describe this is with an example. Assume the compass at hand has an accuracy of +-5 degrees and a precision of 0.1 degrees. Now assume that when positioned at 0 degrees the compass reads 4.1 degrees, very inaccurate but with moderate precision. Now rotate the compass around and then place it again facing 0 degrees. If the compass were to now read 4.2 degrees it would have good repeatability of 0.1 degrees, if it now read 2.1 degrees it would have bad repeatability, of 2 degrees.
1. Create a uniform field in the correct range for the team’s sensors using the Hemholtz Coils. 2. Using a laptop/LabVIEW interface to the High Precision Rotary Table the team can get a precision angular reading for when the device reads 0° 0' 0" or some other recordable value. 3. Take a reading from the constructed compass. 4. Rotate the table around to multiple positions. Wait a little bit and then return the table, again very precisely, to the previous position 5. Take another reading from the constructed device. 6. Compare the readings. If the readings are the same the device is very accurate. 7. The device can output a reading calculated from the 9 sensors and also a reading from a single sensor for comparison.
Once the specifications have be determined a conclusion can be made. If the the proof of concepts passes a little research into the area of nanowires should indicate whether the same setup should significantly increasing accuracy with nanowires. Specifically, one might want to see if they have a linear and non linear region of measurement like the AMR ICs do, as this is basis of the group’s averaging routine.
If the sensors are provided by Honeywell the team’s simple circuit design will easily be within the department’s 200 dollar price range. Beth Stadler and the Electrical Engineering Department will provide the team with the necessary testing equipment.
With this design the team hopes to get nine fold the accuracy of an individual sensor. If the original hypothesis is correct and more sensors can improve the accuracy then Honeywell might be inclined to look into adding more axes to their sensors or more sensors to their navigation systems. Research may also be done to begin implementation of nanowires into the wheatstone bridge in place of the currently used magnetoresistive elements to further decrease size.