Cal Poly Computer Engineering Senior Project Joe on the Go

Cal Poly Computer Engineering Senior Project
Joe on the Go
Grayson Meurrens
Nico Ledwith
Advisor: Bruce DeBruhl
Spring, 2017
1 Introduction
1.1 Project Goals
The ultimate goal for this project is to
design an automatic, large capacity coffee
maker. The system should be able to detect
when a cup or mug is in position to receive
coffee, then dispense a cup’s worth (~ 8oz) of
coffee. When the coffee in the urn is getting low,
our system should be able to allow a person to
put new coffee grounds in the top, then push a
button to start the coffee brewing process.
More specifically, we had to accomplish
the following tasks in order to meet our goals:
solving the issue of transporting water from a
reservoir to a coffee urn through a heating
component so that it can brew coffee grounds;
figuring out how to measure how much coffee
has been dispensed by the urn to keep each pour
at a consistent 8 fluid ounces; detecting when a
cup or mug is in proper position; and keeping
track of the amount of coffee left in the urn at all
times so that the system can know when more
coffee needs to be brewed.
1.2 Project Objectives
In order to meet our goal of having an
automated coffee maker, we had to integrate
many hardware components. Many of these had
to be controlled by a central microprocessor.
This system consisted of a large container to
hold water, a heating component, and a coffee
urn to keep large amounts of coffee warm. The
heating component can provide the
transportation to the coffee urn through some
interesting physics, so we as designers just need
to position it correctly. Two liquid flow meters
were used to measure the flow into and out of
the coffee urn, giving us enough information to
determine when new coffee needed to be
brewed. A sonar sensor placed under the coffee
dispenser provided distance information to
detect when a cup or mug is in proper position.
Finally, a water valve controlled coffee flow out
of the coffee urn to provide consistent 8 ounce
pours.
1.3 Project Outcomes and
Deliverables
Outcomes and deliverables of this project
consist of:
● An all-in-one, automatic, large capacity
coffee maker, that only requires human
interaction when the coffee grounds
need to be replaced.
● A technical manual describing the
end-to-end process of how water
becomes coffee in our system.
2 Background
2.1 Julian’s in Kennedy Library
During the 2015-2016 academic year,
Julian’s coffee stopped allowing customers to
pour themselves coffee. The main issue was that
people were stealing coffee. All a customer
needed was a cup, and they could just walk up to
the coffee urn, pour themselves coffee and leave.
Additionally, some workers would just hand out
free or reduced price cups, which the customers
would use to pour themselves coffee. Now,
Julian’s requires the worker to pour the coffee
themselves, then give it to the paying customer.
On a related note, for customers just
buying a single cup of coffee, the line at Julian’s
can be a huge deterrent to providing Julian’s
with business. We thought that there must be a
1
better way to get people coffee without waiting
in line. We concede that if a customer wants a
latte, or espresso, or anything ​other​ than plain
coffee, then waiting in line is appropriate. But
for a cup of brewed coffee, an expedited version
would surely be an easy solution. Thus the idea
for an automated, large capacity coffee maker
emerged. A fully imagined product would take
credit card payments, venmo, and ​even​ cash,
then dispense a specified amount of coffee to the
customer. Customers could pay this product,
walk own cup or mug, and walk away with a full
cup of coffee. Think of the automatic filtered
water dispensers across campus, then add coffee
and payment. The fully imagined product is
quite a bit of work, so for this senior project, we
set our sights simply on getting a working
version of all the physical components and left
the payment system for the future.
2.2 Bubble Pump
In the tube leading to the heating
element is a check valve, shown below. This
valve is a simple ball in a socket that allows to
water to move one direction only: through the
heating element in this case. Water pressure in
the reservoir will push water through the valve
and into the heating element. Another tube leads
out of the heating element and into the coffee
grounds sitting above the carafe. But water
pressure alone is not enough to pump the heated
water into the coffee grounds. Here is where the
bubble pump comes into play. When the water is
heated in the heating element, the cool water
turns into water vapor and hot water. This
mixture of vapor and hot water is moved up the
tube in two ways. First, the mixture is less dense
than the cool water, so the cool water naturally
pushed the hot water up the tube (the one way
valve prevents any backflow). Second, the water
vapor also only have one way to go, so they flow
up the tube as well and help move the hot water
out of the heating element. [1]
Through research, we discovered that a
traditional drip coffee maker utilizes a
phenomenon called a gas lift, or “bubble pump,”
to pump hot water up a tube in the machine into
the coffee grounds. Cold water sits in the
reservoir of a coffee maker, with a tube going
out leading to the heating element, a U shaped
metal tube connected to 120V to heat up the
water. A process flow diagram is shown below,
providing a visual representation of this pump
system.
Figure 2: Check Ball Valve [2]
Figure 1: Process Diagram for Bubble Pump
2
3 Engineering Specs
3.1 Use Cases
To help us in the design process, we
created what we thought would be the typical
use case for the device. Through this use case
we were able to envision scenarios in which the
system would be used, thus giving us better
insights when making design decisions. The
following paragraph describes the primary use
case for the device.
The primary use case for this project is a
student or faculty member of a university,
wanting to buy a cup of coffee without waiting
in a line. The “customer”, as we will refer to
them, will perform the following actions when
using our system: set cup or mug under the
dispenser and into range of the distance sensor;
wait while coffee is dispensed, remove cup or
mug from under the dispenser. While this is a
simple set of tasks for the customer, our coffee
system must function in a way that the consumer
would expect in order to meet the customer’s
expectations. These expectations entail certain
specifications about how long the coffee takes to
dispense, the temperature of the dispensed
coffee, the distance between the cup and the
dispenser, and the amount of coffee being
dispensed. We have developed engineering
specifications in order for this use case to meet
the customer expectations, tabulated in the next
section.
3.2 Specifications
By thinking about customer needs and
expectations, we developed a list of engineering
specifications, with discrete values and margins
of error. Having concrete values and parameters
to design for helped us when making many
design decisions. The following table lists out
our engineering specifications. It includes a
column for risk analysis as well.
Parameter
Specification
Unit
s
Tolerance
Risk
Water temp
when
brewing
180
˚F
Min
L
Amount of
water
poured in
urn
300
fl
oz
± 15 fl
oz
M
Range cup
is detected
at
6
cm
± 1 cm
M
Amount of
coffee
dispensed
7.75
fl
oz
± 0.25 fl
oz
M
Time it
takes to
dispense
coffee
6
Sec
ond
s
±1s
M
Coffee
temp when
dispensed
180
˚F
Min
L
Table 1: Engineering Specifications
3
4 Design Development
The following section will describe the
design development of our system. This section
will focus on the various hardware components
we decided to implement into our system; it
starts with a block diagram showing the
connections between all components, then goes
into detail about each individual one.
Throughout, we give a description of why the
each component is necessary in our design and
provide images of the component.
4.1 Block Diagram
Figure 4: Heating Element / Pump
4.2.2 Relay
For us to control high voltages we have
a 4 way mechanical switch relay. This is used to
control many parts of the project. The voltages
that it controls are the 120V required for the
pump and the 12V DC that the valve uses.
Figure 3: Hardware Block Diagram
4.2 Hardware
4.2.1 Pump
In order to heat up water and push it to
the coffee urn we implemented the bubble pump
described above. This allows us to have a
stationary water source that we could control as
well as heat. This is connected to the relay in
order to do so.
4
Figure 6: Flow Meter
4.2.4 Solenoid/Valve
Figure 5: Hardware Relay
4.2.3 Flow Meters
We needed a way to measure the
volume of water and coffee we are outputting at
certain points in the design. We opted to use two
flow meters. These meters work by having an
internal wheel that toggles a trigger output every
full revolution that it makes. Using this we
calculate the average frequency over time and
implement some algorithm to finally calculate
ounces dispensed. We have one meter that is
placed right before the pump so we can tell how
much water we are putting into the urn and one
that is right after the spout so we can read how
much coffee is being poured out.
In order to be able to control the coffee
flowing out of the urn, we needed some
hardware that could either control the built-in
spout or act as one itself. Our first idea was to go
with a push/pull solenoid. But once we tested it,
the solenoid proved to be too weak to actuate the
spout and was ruled out. We then opted to go
with a plastic valve. However, this proved to
also be inefficient since it require a minimum
pressure threshold that we would not be able to
satisfy once the coffee in the urn reached a
certain level. Finally, we moved on to our
current brass valve that does not require a
minimum pressure to work. This requires 12V
DC, so it is also connected to the relay switch.
5
Figure 8: Distance Meter
4.3 Raspberry Pi
Figure 7: Solenoid, Plastic Valve, Brass Valve
(clockwise from top left)
4.2.5 Distance Sensor
We needed a way to detect if the user
placed a cup in range of the valve, so the system
could tell if it was ready to pour. To do so, we
use an ultrasound transducer that only requires
5V to operate. This tells us in real time how far
an object is away from it.
To act as the control unit for our project
we chose to implement a Raspberry Pi 3 board.
This gives us a nice balance of powerful and
ease of use. At the early stages of development
we ran into a connectivity problem with the Pi.
We were unable to ssh into it through WiFi. Our
solution was to connect to the Pi through
ethernet everytime we wanted to edit our
program. The Pi controls the relay and reads
from the distance sensor and flow meters. It does
so through a Python program.
Figure 10: Raspberry Pi (in case)
6
4.4 Software
Item
Cost
The whole system runs on a Python
program that implements a state machine that
changes states based on the sensors that the
Raspberry Pi reads from. The five states that we
use are Idle, Pouring, Ready to Serve, Paid, and
Serving. By rotating from state to state we are
able to maintain a consistent and efficient
system. We also have multiple small programs
for testing purposes. For each of the hardware
components we have a small script to test it
works how we expect.
Coffee Urn
$34
Bubble Pump from Coffee
Maker
$9
Water Source
$3
Tubing
$5.78
Raspberry Pi 3 + Kit
$70
Relay Switch
$7
Flow Meter (2)
$19
5 Final Concept
Distance Sensor
$6.38
Brass Valve
$24
Total
$178.16
5.1 Software State Machine
Table 2: Cost Estimate
6 System Integration and
Testing
Figure 11: Software Finite State Machine
Diagram
Note: We decided to simulate payment instead of
implementing a whole transaction system. The payment
was simulated through a hardware button press.
5.2 Final Product Cost Estimate
The following table lists each component in our
system along with its cost. This table provides a
total bill of cost for the whole system. Cost of
components was a consideration when
designing, as we had a $200 limit.
After we decided which components to
implement into our design, we needed to test
them. The following section provides details
about testing methods for each hardware
components. We discuss those that succeeded as
well as those that failed.
6.1 Hardware Tests
6.1.1 Pump
❖ Control Test
➢ To test that the pump actuates
with the relay switch, we made
a program that toggles the relay
with a key press. If the pump
7
turned off/on with the key press,
then it passes.
❖ Volume Loss Test
➢ To test that the pump does not
lose any of the water that flows
throughout, we measured the
amount of water entering the
pump in fl oz and then the
amount exiting in fl oz and
compared. If the amount after
was within 5% of the amount
before, then it passes.
6.1.2 Flow meter
❖ Volume Test
➢ To test how accurate the flow
sensors were, we developed a
program that keeps a running
average of the average
frequency of the flow meter.
With that average we were then
able to calculate the total
amount of volume that passed
through. We ran this test with a
known fixed amount of water to
see how accurate that program
was. If the test showed the the
measured amount was within 1
fl oz of the actual amount, it
passed.
6.1.3 Relay
❖ Control Test
➢ Same test as the control test for
the relay.
6.1.4 Distance Sensor
❖ Precision Test
➢ In order to test if the distance
sensor could detect objects from
a relatively close distance we
developed a program that
continuously outputs the reading
from the sensor. We then moved
a cup closer and closer to the
sensor and if the sensor was
able to pick up the object from 6
cm or less, it passed.
6.1.5 Solenoid - Failed
❖ Push test
➢ To see if the solenoid was
sufficient to actuate the spout of
the urn, we applied the correct
voltage (20V DC) to the
solenoid and positioned it right
above the spout. Since the
solenoid was not strong enough
to push it, the test failed and we
had to move onto another
option.
6.1.6 Valves
❖ Flow Test
➢ Since the solenoid failed, we
had to figure out a new way to
control the flow of coffee out of
the urn. We opted to go with a
valve. To test the valve we used
the control test explained above
and attached it to the urn. Once
the voltage was applied, the
valve passed if coffee was able
to flow out. The first valve we
had was a plastic one that
required a minimum amount of
water pressure. This valve failed
since the pressure threshold was
not being met. The next valve
was brass and did not have a
minimum pressure requirement
so it passed.
8
6.2 System and Unit Tests
For the overall system we wrote several test
cases to check the functionality of the project.
Unit tests were also performed on each of the
above hardware components to ensure they
functioned as we expected. Each of these tests
consisted of a short Python script to control and /
or measure the component. Manual physical
testing and validation was performed in
conjunction with each script. Once each
component was tested, we combined them all
into our system and tested it in a similar fashion.
Using our main Python script, we ran our system
and manually tested and verified that certain
behaviors were ones we expected. This included
turning the pump on when the coffee level
became too low and dispensing the correct
amount of coffee each time via the brass valve.
7 Division of Labor
➢ Bubble Pump
➢ Flow Meter 1
➢ Relay Switch
❖ Mechanical Setup
➢ Coffee Urn
➢ Tubing
➢ Wiring
➢ Mounting Raspberry Pi
❖ Corresponding Software Design and
Testing
7.2 Teammate 2 - Grayson
Meurrens
❖ Frontend Hardware
➢ Valve
➢ Flow Meter 2
➢ Distance Sensor
❖ Raspberry Pi Setup
➢ Installing OS
➢ Connectivity Issues/Solution
❖ Corresponding Software Design and
Testing
7.1 Teammate 1 - Nico Ledwith
❖ Backend Hardware
9
8 Appendix
8.1 Python Code
"""
Joe on the Go
Nico Ledwith and Grayson Meurrens
Cal Poly Computer Engineering Senior Project
"""
#enable GPIO on Raspberry Pi
import​ ​RPi​.​GPIO ​as​ GPIO
#activate Broadcom-chip pin numbers
GPIO​.​setmode​(​GPIO​.​BCM)
#Set state constants
IDLE ​=​ 1
BREW ​=​ 2
READY ​=​ 3
PAID ​=​ 4
SERVING ​=​ 5
CUP_SIZE ​=​ ​8​ ​#cup size in fl oz
URN_SIZE ​=​ ​300​ ​#urn size in fl oz
#Setup data
pinSetup​()
state ​=​ 1
while​ ​True:
​if​ state ​==​ IDLE:
​#press c to start
​if​ key ​==​ ​'c':
state ​=​ BREW
​elif​ state ​==​ BREW:
​#once urn is full, state is ready
​togglePump​()
​if​ urnLevel​()​ ​>=​ URN_SIZE:
togglePump​()
state ​=​ READY
​elif​ state ​==​ READY:
​#check for payment
​if​ hasPaid​():
toggleDist​()
state ​=​ PAID
​elif​ state ​==​ PAID:
​#if customer paid, then wait for cup to be in position
distance ​=​ checkDist​()
​if​ distance ​<​ ​6​ ​and​ distance ​>​ ​4:
toggleValve​()
state ​=​ SERVING
10
​elif​ state ​==​ SERVING:
​#pour unless cup moves or is full
distance ​=​ checkDist​()
level ​=​ amtPoured​()
​if​ distance ​>​ ​6​ ​or​ distance ​<​ ​4:
toggleValve​()
state ​=​ PAID
​elif​ level ​>=​ CUP_SIZE:
toggleValve​()
​#check if urn needs to be refilled
​if​ urnLevel​()​ ​<​ 2 * CUP_SIZE:
state ​=​ BREW
​else:
state ​=​ READY
​else:
​print​(​"Invalid State\n")
"""list of functions
pinSetup​()
togglePump​()
urnLevel​()
hasPaid​()
toggleDist​()
checkDist​()
toggleValve​()
amtPoured​()
"""
8.2 Configuration
Via ​https://www.raspberrypi.org/forums/viewtopic.php?f=66&t=18207​:
- On the Raspberry Pi we entered the bash command:
- sudo apt-get update
- sudo apt-get install libnss-mdns
- Then we connected our Mac to the Pi via a Cat5 Ethernet cable
- The Pi was available to ​ssh​ into via ​ssh -X pi@raspberrypi.local
8.3 Sources Cited
[1] ​http://petrowiki.org/Gas_lift
[2] ​https://cdn.instructables.com/FJM/8GU0/HAQ3A9GZ/FJM8GU0HAQ3A9GZ.MEDIUM.jpg?width=614
11
8.4 Brief Technical Guide
Brief Technical Document Describing End to End Process of ​Joe on the Go
Grayson Meurrens
Nico Ledwith
Cal Poly, SLO
CPE Senior Project
Spring 2017
This document serves as a technical guide to describe the function of each component in the ​Joe
on the Go ​system. These descriptions will be facilitated through a description of the various
states our system goes through in order to turn water into coffee.
12
1. Initial state
a. Room temperature water sits in the reservoir, a five gallon plastic water dispenser
with a vinyl tube attached to the bottom.
b. The vinyl tube in a) is attached to a flow meter which is attached to a heating
element, which in turn is attached to the top of the coffee urn with a second vinyl
tube. The heating element is also electrically connected to a physical power relay.
c. The coffee spigot of the coffee urn is constantly in the “pour” position, but coffee
flow is controlled by a brass valve connected to the end of the spigot.
d. The valve in c) is connected to the same relay as the heating component, as it
requires 12V to operate.
e. Connected to the end of the valve is a flow meter, to measure coffee flow.
f. A distance meter sits below the flow meter to measure the distance of a cup
receiving coffee.
g. A Raspberry Pi controls: two flow meters, the physical relay, the brass valve, the
distance meter, and a button used to simulate a payment being made.
2. Brewing
a. The Raspberry Pi sends a HIGH signal to the pin connected to the physical relay,
thus providing 120V to the heating element.
b. The heating element begins to heat up to around 200 ºF
c. This heat, combined with a small check ball valve, moves heated water through
the flow meter and out of the element, pumping it into the urn.
d. The Pi begins reading signals sent from the flow meter. Through this equation,
F = 69 * Q(L/min) , where F is the frequency of signals received by the Pi, Q is
the flow rate, and 69 is a derived constant, the number of fluid ounces is
determined.
e. Once the number of fluid ounces reaches the set capacity of the urn, the Pi sends a
LOW signal to the physical relay, thus turning the heating element off and
stopping the flow of water.
f. The amount of water pumped from the reservoir is saved for future use.
3. Ready
a. In this state, hot water is either dripping through coffee grounds or sitting in the
urn.
b. The urn is connected to 120V, providing it enough power to keep the brewed
coffee at around 180 ºF.
c. The Pi is constantly polling the distance meter so see if a cup is within range as
well as the payment button.
13
d. If the distance meter reads that a cup is within 6 ± 1 cm for 1.5 seconds, and the
payment (simulated by a button press) is received, the Pi will send a signal to the
relay to actuate the valve and open it.
4. Serving
a. After the Pi opens the valve, it begins reading the signals from the flow meter
b. Using the aforementioned equation, the Pi determines how much coffee is being
dispensed from the urn.
c. Once this amount reaches 7.5 oz, the Pi sends a signal to the relay to turn the
valve off. 7.5 was chosen because the flow meter has about a 0.25 oz error in
measurement.
d. At this point, the Pi has a decision to make:
i.
If the amount of coffee dispensed causes the amount of coffee in the urn to
fall below 16 oz, the Pi goes into the Brewing state.
1.
ii.
Note: at this point, more coffee grounds will need to be manually placed in the
coffee urn.
Otherwise, the Pi goes back into Ready.
14
Download PDF
Similar pages