ECE375 Final Project

Final Project

File Types pdf and asm
ECE375 Final Project
Your submission should consist of two items:
1. 100 points - a single .asm file containing your AVR assembly source code
2. 56 points - a PDF document containing your typed report
Your grade will be reduced if you do not use the assembly code template file
(https://canvas.oregonstate.edu/courses/1798830/files/84423190/download?download_frd=1) that is
provided.
Please take the time to read the entire documentation.
Introduction
In July of 1969, the Apollo 11 mission departed earth and headed to space for an eight day visit. The
spacecraft consisted of three distinct components: the command module, the service module, and the
lunar module. The command module and lunar modules each contained an AGC (Apollo Guidance
Computer).
The specifications of the AGC were as follows:
16-bit word length (divided into 15 data bits and 1 parity bit)
2048 words of RAM
36,864 words of ROM (program memory)
2.048 MHz clock oscillator
The AGC was designed in a similar manner to the CPUs which we have discussed during lecture. The
instruction format of the computer utilized 3 bits for the opcode and 12 bits for the address. An
accumulator register was available for mathematical operations. The AGC provided five vectored
interrupts which could be used for timer support and remote communications.

AGC Assembly Code
The Apollo Guidance Computer was developed specifically to control the spacecraft on its way to the
moon and to manage the many details that were required to make the mission successful. Suffice to say,
this was not an easy job. Perhaps most fascinating, the Apollo 11 assembly code has been transcribed
and uploaded to GitHub for public review (see here (https://github.com/chrislgarry/Apollo-11) ).
The programmers included various annotations throughout the source code, some of which would not be
out of place in ECE375 assembly code. The assembly commands are different, but it's easy to
recognize the basic structure of comparisons and conditional branches.
For example, consider this snippet of AGC assembly code (pay special attention to the comments):
KILLTSK2 LXCH ITEMP2 # SAVE CALLER'S BBANK
[...]
ZL
ADRSCAN INDEX L
CS LST2
AD ITEMP4 # COMPARE GENADRS
EXTEND
BZF TSTFBANK # IF THEY MATCH, COMPARE FBANKS
LETITLIV CS LSTLIM
AD L
EXTEND # ARE WE DONE?
BZF DEAD # YES -- DONE, SO RETURN
INCR L
INCR L
TCF ADRSCAN # CONTINUE LOOP.
Other areas of the code are more specific to the world of rocket engines...
IGNYET? CAF ASTNBIT # CHECK ASTNFLAG: HAS ASTRONAUT RESPONDED
MASK FLAGWRD7 # TO OUR ENGINE ENABLE REQUEST?
EXTEND
INDEX WHICH
BZF 12 # BRANCH IF HE HAS NOT RESPONDED YET
IGNITION CS FLAGWRD5 # INSURE ENGONFLG IS SET.
MASK ENGONBIT
ADS FLAGWRD5
CS PRIO30 # TURN ON THE ENGINE.
My personal favorite section of the code is located in this file (https://github.com/chrislgarry/Apollo11/blob/master/Luminary099/THE_LUNAR_LANDING.agc#L245) which apparently involved proper
alignment of the landing radar antenna:
BZF P63SPOT4 # BRANCH IF ANTENNA ALREADY IN POSITION 1
CAF CODE500 # ASTRONAUT: PLEASE CRANK THE
TC BANKCALL # SILLY THING AROUND
CADR GOPERF1
TCF GOTOPOOH # TERMINATE
TCF P63SPOT3 # PROCEED SEE IF HE'S LYING
P63SPOT4 TC BANKCALL # ENTER INITIALIZE LANDING RADAR
CADR SETPOS1

TC POSTJUMP # OFF TO SEE THE WIZARD...
Your Assignment
Navigating in space requires a lot of physics and math. In order to get to the moon, you need
trigonometric equations, a bit of calculus, lots of arithmetic, and a reliable way to propel and steer your
spacecraft. Of course, you also need to design and build a sturdy spaceship.
The AVR microcontroller on your lab board is arguably more powerful than the CPU that drove the
Apollo 11 mission. In honor of this early computer, we will use our ATmega128 board to solve some
simple mathematical equations that are related to spaceflight.
What do you have to do?
In this final project you will write AVR assembly code to solve some equations as described below. You
will store the computed results into data memory. A template file is provided to you with details
about the various operands. You must write your source code within the template file. Please be sure
to read this entire document in order to understand how the values will be provided to your code. Each
time your program runs, you will solve the equations shown below.
All math will be performed using integers. No floating point operations are required!
Satellite Velocity (problem A)
Calculate the velocity of an artificial satellite orbiting an object in space. Your answer must be in units of
kilometers per second.
You will be provided the radius of the orbit ( ) and the standard gravitational parameter ( )
corresponding to a particular planet. Round your answer at all stages of the equation. For example,
should be rounded to the nearest integer when performing division. If the square root of is not
already an integer, round it downward.
 Math Example
Please look at the template file as you read through this example. Suppose that the source code file
indicates that we have an OrbitalRadius of 0x64, 0x19. Assume that it also specifies SelectedPlanet as
0x02.
Since the values are stored in little-endian format, the OrbitalRadius has a decimal value of 6500. With a
SelectedPlanet value of 2, this is indicating that we need to use the GM value that is located at index 2
of the array (explained within the template file). Based on the example file, this corresponds to the
v =
GM
r
−−− √
r GM
GM
r
GM
r

hexadecimal data 0x08, 0x15, 0x06, 0x00. Interpreting this as an unsigned 32 bit value in little-endian
format, it's equivalent to the decimal value 398,600.
We now compute
When you perform division, you will round to the nearest integer. That's why we rounded to 61 even
though 61.3230 is a closer approximation. You need to store this value into memory at address
"Quotient". This allows us to evaluate your code for partial credit.
When the grader inspects the Quotient in data memory (using Atmel Studio) they will see the 24 bit
representation in little-endian format:
0x3d, 0x00, 0x00
Next, we need to compute the square root: (note that we round down for
square roots)
Our final answer is 7 km/s. The value 7 should be stored into the "Velocity" location as a 16 bit signed
representation:
0x07, 0x00
Period of Revolution (problem B)
What is the period of revolution for the satellite described in problem A? Provide your answer in units of
seconds.
As before, you will be provided the radius of the orbit ( ) and the standard gravitational parameter (
). Round to the nearest integer when performing division, and round downward when computing square
roots.
For the sake of simplicity, your program should approximate as 10.
Math Example
For this example let's use the same parameters as we used for part A.
We first need to compute
Plugging in the orbital radius value of 6500, we get
 (big-endian)
(note that we used the approximation of )
We need to store this intermediate value into data memory at the "Product" location. As before, this is
used for the grader to consider partial credit. The number will be stored as a 7 byte little-endian value:
= = 61.3230 = 61
GM
r
398,600
6500
61 = 7.8102... = 7
−− √
r GM
π
2
T = 4π
2r
3
GM
−−−− √
4π
2r
3
4 ⋅ 10 ⋅ 6500 = 10,985,000,000,000 = 0x9FDA505DA00
3
π = 10
2

0x00, 0xDA, 0x05, 0xA5, 0xFD, 0x09, 0x00
Our next step is to compute the quotient:
 = 27,558,956.347 = 27,558,956 (after rounding to the nearest integer)
Finally, we need the square root:
 (after rounding downward)
The square root will be stored into data memory at the "Period" location (represented as a 24 bit signed
number):
0x81, 0x14, 0x00
Special Cases
Your code will also be capable of indicating four special error cases:
If then you must store a value of -1 into "Velocity" to indicate the error.
As a 16 bit signed value (in little-endian representation) this is: 0xFF, 0xFF
If your code detects a computed velocity value of 0 km/s (after rounding), store a value of -2 into
"Velocity" to indicate the error.
As a 16 bit signed value (in little-endian representation) this is: 0xFE, 0xFF
If then you must store a value of -1 into "Period" to indicate the error.
As a 24 bit signed value (in little-endian representation) this is: 0xFF, 0xFF, 0xFF
If your code detects a period of revolution that is less than 25 seconds (after rounding), store a value
of -2 into "Period" to indicate the error.
As a 24 bit signed value (in little-endian representation) this is: 0xFE, 0xFF, 0xFF
After handling any of these special cases, your code should halt (you do not need to continue computing
the math equations).
Dealing with Parameters
In order to compute the correct values, your program needs to know the orbital radius and standard
gravitational parameter (which depends on the satellite's height and the particular planet you are
orbiting). These values will be provided to you within specific program memory locations. All values are
provided in little-endian format.
Inside the template code, there is a location named "OrbitalRadius". That value contains the radius (in
kilometers). The orbital radius is an unsigned 16-bit value (implying that the maximum value is 65,535).
10,985,000,000,000
398,600
27, 558, 956 = 5249.662 = 5249
−−−−−−−−− √
r ≤ 1000
GM ≤ 1000

The template file provides you with a list of gravitational parameter values. The units are (km * km *
km)/(sec * sec). You must retrieve the correct GM value by utilizing the "SelectedPlanet" information that
is specified at runtime. The template file provides you with 9 different GM values, each of which
corresponds to a particular planet. As an example, if SelectedPlanet has a value of 2, then your code
must use the GM value that is stored at index 2 within the array. From looking at the template file, we
can see that this is the GM value corresponding to Earth. Please see the template file for more context.
Since each GM is an unsigned 32 bit value, the maximum value is 4,294,967,295.
Your solution MUST read the provided 32-bit GM values at runtime. You are not allowed to hardcode the
GM values into your code! These values will be changed during testing!
Some Ideas to Consider
The AVR does not have built-in instructions to divide numbers or compute the square root. As a result,
you will need to implement your own approach to compute those values. There are lots of ways to do
this (feel free to implement any algorithm of your choosing). Since you are specifically working with
integers (and not floating point numbers) both of these mathematical tasks are simplified.
In order to divide two numbers, you can simply keep subtracting the divisor, until the number is too
small to subtract further. This is similar to the division approach that is sometimes taught to children.
For example, "How many times does 5 go into 14?"
Subtract 5 from 14
14-5 = 9
Subtract 5 again
9-5 = 4
Subtracting a third time would result in a negative number. Therefore, we are done. 14/5
generates a quotient of 2, and a remainder of 4.
The remainder is useful because it tells us whether we should round the quotient up or down.
A small code snippet is available here
(https://canvas.oregonstate.edu/courses/1798830/files/84422730/download?download_frd=1) . It
demonstrates the concept using 8 bit values.
The simplest way to determine the square root of a number is to start from 0 and work your way up.
You can compute the square of a number, check to see if the result is correct, and continue trying.
For example, suppose that you want to determine the square root of 5,123.
We try all options beginning from zero. Eventually, we will work up to the correct value.
0 * 0 = 0 (too small)
1 * 1 = 1 (too small)
2 * 2 = 4 (too small)
3 * 3 = 9 (too small)
...

70 * 70 = 4900 (too small)
71 * 71 = 5041 (too small)
72 * 72 = 5184 (too large)
Since the instructions tell you to round downward, we conclude that
For our purposes, the correct value is 71 (even though a calculator says 71.575135347)
We are only working with integers, hence the rounding.
Additional Details
When you are rounding quotients, use the convention that a value of 0.5 rounds up (and anything
lower rounds down). For example, a value of 153.4999 would round to 153. A value of 536.501
would round to 537.
When you are rounding square roots, you must round down (unless the value happens to be a
perfect square). For example, and
I understand that this isn't the most accurate approach, but it allows us to simplify the assignment
and make it faster to implement.
Your submission with be tested with a variety of parameters. Do not hardcode the answers.
All numerical values that are provided to your program will be arranged in little-endian format.
Furthermore, all of your answers must be stored in little-endian format.
This is an individual assignment. You are not allowed to work in groups (and submissions will be
compared for signs of plagiarism).
You are encouraged to re-use your work from earlier assignments. It is perfectly fine to incorporate
code that you wrote previously in this course (including any lab code that was written in collaboration
with a partner).
Please, design your project on paper and try to envision any obstacles before you begin
programming. It is much easier to fix bugs at design time (rather than realizing the problem after
you've written the code). Design twice, code once.
My number one suggestion... design your code in a modular fashion. For example, write your square
root code inside a procedure that you can call from anywhere. On a similar note, if you implement a
multiplication procedure, write it in such a way that you can reuse it for other parts of your code.
You will not receive credit during testing if your code fails to reach the "Grading" label (or gets stuck
inside an infinite loop).
Materials
Required template file - available here
(https://canvas.oregonstate.edu/courses/1798830/files/84423190/download?download_frd=1)
Outside Resources
While working on the exam, students are welcome to use code from the ECE375 slides, previous
homework, or previous labs. You may recycle code that you have written previously.
5123 = 71
−−−− √
25 = 5 −− √ 24 = 4 −− √

It is acceptable to adapt an algorithm from an outside source but you must write the assembly code
yourself. Do not copy assembly code from others!
Extra Credit
There are two options for which you can earn extra credit. In total, you can earn up to 10 additional
points by implementing these features.
4 pts - Design your code in such a way that is approximated as 9.8696 (rather than 10). This will
make your answers more accurate. As with the standard implementation, you will still round your
final answers. There are multiple ways to tackle this challenge.
6 pts - Write your code so that the maximum allowable orbital radius can be as large as 120,000km.
This means that your code will need to accept a 24 bit unsigned value in the "OrbitalRadius" field.
See the template assembly file for more context. The original project implementation can only handle
a maximum radius of 65,535km but your extra credit implementation will be tested with values as
large as 120,000km. As with all parameters, the orbital radius will still be provided in little-endian
format.
Note: even though the radius will be provided as a 24 bit number, you can assume it will
never be larger than 120,000 (in decimal).
Written Report
Document your work in a well-written report. Be sure that your report includes all of the content that is
listed in this section. Present your work to the reader in a professional manner. I suggest that you use
images, charts, diagrams or other visual techniques to help convey your information to the reader. Make
your report something that is intriguing and interesting.
Explain how you implemented your source code to compute the satellite details. You should provide
enough information that a knowledgeable programmer would be able to draw a reasonably accurate
block diagram of your program.
This project includes division, multiplication, and square root operations. In your code, what is the
maximum number of bits that each stage of your program can handle? For example, what is the
largest number (in bits) that your code can divide? What is the largest number (in bits) that you can
multiply? When you determine the square root, what is the largest number (in bits) that your code
can handle?
Describe the algorithm that you used to determine the square root of a number.
What were the primary challenges that you encountered while working on the project?
What features of the program took longest to implement?
Is there anything you would design differently if you were to re-implement this project?
Draw a pie chart and give estimates of the time that you invested into this final project. In particular,
be sure to label these three categories:
time spent doing design and research work
π
2

time spent programming productively
time spent debugging bugs or unexpected obstacles
Proofread your report! Better yet, ask another person to proofread your work. You will be marked
down for spelling or grammatical mistakes.
If you are implementing extra credit work, you MUST include an explanation of your additional work
within the report.
There is no arbitrary minimum length requirement on the report. For example, if you are able to
document your work and answer all of the questions in a two page document, that is perfectly fine. If you
need five pages, that's perfectly fine too.
I will review your assembly code separately, so there is no need to include the source code within your
final project report (unless you are describing certain snippets of the code).
Where to get help?
If you are working on the design of your final project, feel free to stop by office hours and ask questions.
Since this is the final project, I expect students to demonstrate initiative while testing and debugging their
code. We are happy to provide feedback and answer questions, but if you encounter a bug in your code,
you will need to use your debugging skills to resolve it.
If you have a question regarding the project requirements, please let me know (either during lecture or
send me an email). If the instructions are unclear, I can update the documentation and add an entry in
the errata. justin.goins@oregonstate.edu (mailto:justin.goins@oregonstate.edu)
Grading Rubric
Implementation & Functionality
Your code will be tested with a wide variety of configuration settings. Keep this in mind while designing
your algorithms and test your code so that it works for all valid combinations of input data. You should
expect the radius and standard gravitational parameters to be modified across the range of possible
values. Partial credit will be awarded for student code that functions for some test cases.
(24 pts) Implementation Details
Code was written using the required template file. The labels are utilized as written and nothing was
changed below the lines that read "Do not change anything below this point".
Code correctly utilizes the SelectedPlanet information in order to determine which planet information
is needed to compute the corresponding solutions
Code dynamically fetches and utilizes the values of data that are provided in program memory
Quotients are always rounded to the nearest integer
Square roots are rounded down (if the answer is not already an integer)
All values are correctly interpreted in little-endian format

(60 pts) Mathematical Calculations
Correct values are computed and stored into the Quotient location (in data memory)
Correct values are computed and stored into the Velocity location (in data memory)
Correct values are computed and stored into the Product location (in data memory)
Correct values are computed and stored into the Period location (in data memory)
(16 pts) Handling Special Cases
if r <= 1000, Velocity is stored as -1
if velocity rounds to 0, a value of -2 is stored into the Velocity data location
if GM is <= 1000, the Period is stored as -1
if a period of less than 25 seconds is computed, a value of -2 is stored into the data location for
Period
Written Report
Feel free to use the same template format that you have previously used for lab reports.
However, your report must address each of these points:
(14 pts) Report provides reasonable explanation of the student's design. There should be enough
information that a knowledgeable programmer would be able to draw a reasonably accurate block
diagram of the program.
(6 pts) In your code, what is the maximum number of bits that each stage of your program can
handle? For example, what is the largest number (in bits) that your code can divide? What is the
largest number (in bits) that you can multiply? When you determine the square root, what is the
largest number (in bits) that your code can handle?
(6 pts) Describe the algorithm that you used to determine the square root of a number.
(7 pts) What were the primary challenges that you encountered while working on the project?
(7 pts) What features of the program took longest to implement?
(5 pts) Is there anything you would design differently if you were to re-implement this project?
(6 pts) Draw a pie chart and give estimates of the time that you invested into this final project. In
particular, be sure to label these three categories:
time spent doing design and research work
time spent programming productively
time spent debugging bugs or unexpected obstacles
(5 pts) Report is written in a professional manner and does not include spelling mistakes or
grammatical errors.
Errata
This section will be updated when changes or clarifications are made. Each entry here should have a
date and brief description of the change so that you can look over the errata and easily see if any
updates have been made since your last review.

March 1st - Posted the original documentation
March 9th - Added clarification that your code should not require more than 160 million clock cycles. In
other words, I primarily care that your code doesn't get stuck inside an endless loop.
March 10th - Corrected typo in project description. The example math implementation said that velocity
should be stored as a 24 bit value, but this did not match the 2 byte allocation in the project template.
The text has been updated to reflect the corrected instructions. Velocity should be stored in a 2 byte
value (as shown in the project template).
- Changed the wording again to make it clear that the exact execution time does not matter as long as
the code does not enter an infinite loop.
March 11th - Added a grading rubric with an explanation of the point distribution.
March 12th - Added a few sentences in the Special Cases section to clarify the representation of -1, and
-2 in little-endian format.
