Hand Hygiene

I’ve reprogrammed the door minders in the computer lab with the software I worked on last night. Doorminder 098 is in the north lab, 096 in the South Lab. The trigger in the north lab is 138, the trigger in the South Lab is 154. Doorminder 098 can hear the trigger in the South Lab, sometimes. Put all the software on /vinci.1/home/wescrub/DoorminderTrigger/compLabExp-Nov-11-2010/software. I pressed the trigger once immediately after installing each doorminder.

Here are the summarized results from yesterday’s test:

dhcpw80ffc324:ComputerLabTestOctNov10 Geb$ ./summarize.py -i 096.clean.unique.txt
Summary of the file:

Trigger Number of Minimum Maximum
Observed Observations Time Time
154 12 129 63576
138 3 11737 38910
96 1157 3 69012

dhcpw80ffc324:ComputerLabTestOctNov10 Geb$ ./summarize.py -i 098.clean.unique.txt
Summary of the file:

Trigger Number of Minimum Maximum
Observed Observations Time Time
154 12 75 63520
98 30331 0 68978
138 12 16 39020

Updated with the results from 11/11-11/12:
Summary of the file:

Trigger Number of Minimum Maximum
Observed Observations Time Time
154 15 7 96018
138 5 29762 94640
96 1529 2 103945
grok-user@south:~/Downloads/analySoftwareDoorminderTrigger$ ./summarize.py -i 098Unique.txt
Summary of the file:

Trigger Number of Minimum Maximum
Observed Observations Time Time
154 9 11792 86166
98 1116 2 104000
138 16 5 94606

To program the triggers:
Go to Downloads/downloadflash
The command is: python clearAndReprog.py

To program the doorminders:
Go to Downloads/Doorminder2.1
The command is: make telosb install.# (# = the mote ID number)

To download data from doorminders and triggers:
Go to Downloads/HandHyg8/SourceCode/moteFlash

The command is: make -f Makefile.TFlash telosb install.#
Then go to the Downloads/HandHyg8 directory:

The commands are:

./readcli -t -s -u > #.dat

python appl/convlog.py #.dat | tee #.txt

To clear flash:
Go to Downloads/downloadflash
The command is: python clearAndReprog.py
Note: For the doorminders, after the flash is cleared, the script will fail when it tries to reprogram the mote. Reprogramming must be done manually.

Today we set up an experiment for the tilt sensor. It was attached to a hospital bed (Hill-Rom total care model p1900), and the following summarizes what has happened in the experiment so far.

The experiment is being left for the weekend. Phil is going to change the angle during that time, and record the changes he makes. The plan is to download the data on Tuesday.

After programming the Arduino board for a full cycle, the temperatures near the melting point weren’t accurate. Upp’d op amp supply voltage to 10 volts to increase range and recalibrated the voltage vs temp.
Voltage Temp
2.37 200
2.19 180
2.9 261
2.71 240
2.27 190
1.99 162
1.78 137
1.62 119
1.5 107
1.2 74
1 54

After graphing:

Plugged new temperature equation into program and heating process occurred as desired.

time to solder…

HAWK-NODE is our purpose built, Epic platform based mote. The Basic concept is to create a user interface aligned along a single edge of the mote making it easier to build devices around. It is essentially a single board version of the IRENE design of Prabal Dutta. The first prototype was constructed based on design iteration 0.3. The TX and RX serial lines were reversed on the board, but after correcting these the mote has worked well. The blink program installed with no problem, and the radio performance appears to be comparable to the telosb.

Board Layout

The user interface consists of (across the top, R to L): Auxiliary user button interface, User button, RGB feedback LED’s, on-off-reset switch, mini-b USB connection, USB/Charge status light, program/serial switch. The board outside dimensions are 63 x 35mm.

Schematic

After several repeated failed attempts of trying to duplicate the desired reflow curve, resulting temperatures were inconsistent. After finding the temperature control in the toaster oven is a bimetallic thermostat, we decided to create our own using an Arduino board and the thermocouple.
Since we already determined the voltage out of the thermocouple is a linear relationship based on temperature, we are going to use this signal ranging between 0-10 mV for our thermocouple and amplify it using a non-inverting amplifer as shown below.

After several tests with a 5 volt power supply determined the maximum output of our amplifier is approximately 3 volts. The relationship between temperature and voltage becomes non-linear above that region. To stay under this voltage we used R1 = 12 ohms, and R2 = 2.7 K ohms, resulting in a gain of 226.
We first need to determine the fluke and thermocouple are accurate. By running 3 trials we gathered the following info:
mV T f(V) (B-D)^2
1.5 60 60.7 0.454276
1.7 65.6 65.6 0.000196
2.8 92.8 92.8 0.000256
3.5 109 110.1 1.153476
4.6 136 137.2 1.547536
5.8 167 166.9 0.013456
6.3 177 179.2 4.990756
6.7 188 189.1 1.240996
8 222 221.2 0.602176
8.6 237 236.0 0.913936
9.2 252 250.9 1.290496
0.9 46.4 45.9 0.298116
2.3 81 80.4 0.320356
3.6 111 112.5 2.383936
4.6 136.4 137.2 0.712336
5.5 158.5 159.5 0.948676
6.1 175 174.3 0.498436
7 197 196.5 0.226576
7.9 219 218.8 0.060516
8.5 233 233.6 0.329476
9.2 251 250.9 0.018496
0.5 38 36.0 4.104676
1.3 57 55.7 1.602756
2.8 92.4 92.8 0.147456
4.1 125.7 124.9 0.649636
5.6 162 161.9 0.003136
6.2 177 176.8 0.055696
6.7 190 189.1 0.784996
7.1 198 199.0 0.988036
7.5 208 208.9 0.763876
9.1 250 248.4 2.579236
29.68398
variance 0.957548
s.d. 0.978544

f(v) is our calculated temperature based on our graphs computed trendline. We then computed the variance and standard deviation to test the accuracy.
We are then able to determine whether this is accurate by boiling water and testing fluke-thermocouple reading and compare it to our expected results. The fluke read 99.9 degrees Celsius and 3.1 mV out, which matches our expectations.
By now knowing our thermostat is accurate, we can then program the Arduino board to take this voltage as an input and convert it over to a temperature. After connecting our amplifer to the output of our thermocouple, we need to once again calibrate our device. After collecting data:
Volts Temp
0.957 43.1
1.7 116
2.1 160
2.25 178
2.42 196
2.6 217
1.63 109

After programming the Arduino board with the temperature conversion, we are able to determine the temperature of the oven based on the input to the Arduino board from the amplifier. We can then write the Arduino software and control our oven’s temperature using a relay which controls the power to the heating elements of the toaster oven.
The programming is as follow:
void loop() {
temp = 106.4*voltage – 62;//converting for amplifier
sensorValue = analogRead(sensorPin);
voltage = (sensorValue*5/1024.00);
if(element==0 && temp < 160){
element=1;
digitalWrite(ledPin, HIGH);
// read the value from the sensor:
sensorValue = analogRead(sensorPin);
delay(10000);
}
/*if(element==0 && temp160){
sensorValue = analogRead(sensorPin);
element = 0;
digitalWrite(ledPin, LOW);
}
delay(1000);
Serial.print(“\nSensor value is “);
Serial.print(sensorValue);
Serial.print(” Voltage is “);
Serial.println(voltage, 2);//print to 2 decimal places
Serial.print(” Temp is “);
Serial.println(temp, 2);

Resulting in the following temperature graph (the drop in temperatureat 500 seconds is opening to door to see how it reacts and corrects itself):

We got the first decent dataset from the test of the trigger/door minder combination we’ve been running in the computer lab for a few weeks.  The doorminder recorded all the trigger and door pass events for 6.44 days (including the 8 clock rollovers in the data files).  There were 7193 unique events (events separated by at least 2 seconds from any one device).  7086 of these were from people passing through the door, 107 from using the hand hygiene dispenser.  The data file looks like this:

7968 records gathered
id time xmit/voltage version rssi
101 0 101 3 0
101 13 101 3 0
101 14 101 3 0
101 15 101 3 0
101 16 101 3 0
101 17 101 3 0
101 18 101 3 0
101 21 101 3 0
101 21 101 3 0
101 44 101 3 0
101 67 101 3 0
101 81 159 3 254
101 81 159 3 254
101 81 159 3 1
101 83 101 3 0
101 92 101 3 0
101 122 101 3 0
101 124 101 3 0
101 126 101 3 0
101 140 101 3 0
101 153 101 3 0
101 161 101 3 0
101 167 101 3 0
101 170 101 3 0
101 177 101 3 0
101 182 113 3 227
101 182 113 3 224
101 182 113 3 225
101 187 113 3 221
101 187 113 3 211
101 187 113 3 216
101 190 101 3 0
101 212 101 3 0
101 218 101 3 0
101 219 101 3 0
101 221 101 3 0

After establishing the relationship between the resistance of the RS meter and the temperature using the thermocouple, we can begin the heating process to try to duplicate the ideal conditions.

Test 1: full heating to max temperature

Using the temperature equation stated previously and the resistance measured by the RS meter, the temperature gradually increased to 281 degrees and then slightly declined and leveled off. I conducted another test using a second meter of the same brand just to make sure I was getting similar results.

In an attempt to duplicate the ideal heating curve, on the next attempt I tried to reach the soaking temperature and remain there for approximately 2 minutes. I changed the temperature of the oven to 200 and started it up. The oven temperature gradually increased to 243 degrees, which is still too high. Not remaining at the soaking temperature for 1-2 minutes can cause thermal shock, resulting in warpage and bad solder joints because of the temperature variations in the board.

Since it heated to quickly, I decreased the starting temperature to 145 and started the oven up again. After 140 seconds, because the temperature started to drop off, I increased the temperature of the oven to 175, resulting in too big of a temperature increase. Then after a total of 240 seconds, I increased the temperature to max to verify the reflow stage. After the oven reached it’s peak temperature I unplugged the unit and at 500 seconds I opened the door to the oven. Unplugging the unit resulted in the proper gradual decline but opening the door resulted in too quick of a decline possibly resulting in thermal shock.

On test 5, I started off with baking at 145 degrees and in an attempt to keep the temperature constant for soaking increased the temperature to 155 after it started falling after reaching its peak. At 250 seconds I increased the temperature to its max and the final result somewhat resembled the desired heating curve.

Davie tested a bunch of glues yesterday to attach the microswitch to the bottle tops.  Today we evaluated their bonds.  Here are the results:

Loctite Quick Set Epoxy made a reasonably strong bond, but pulled away with strong force with fingernail.  Too much epoxy, got into switch and switch no longer depresses.

Gorilla super glue.  No significant bond.  Glue got into the switch and ruined it.

Loctite 454.  Good bond.  Switch still works. 

Duco cement.  Failed quickly.  Switch is OK, but started a bit rough and got smoother after a little play.

Elmers Super Fast Epoxy.  Good bond.  Switch still works, though seems like it might stick at extreme high levels.

Testing our ability to manufacturer our own circuit boards using a toaster oven and solder paste: The steps would involve putting solder paste on the board, using either a stencil or a syringe, placing the surface mount components on the board, and then placing the circuit board in the toaster oven to heat the solder.

The temperature depends on the melting point of the solder but the ideal conditions for the process are:

We don’t have a thermocouple for the RS(Radioshack) meter to give us the temperature as a function of time, but we do have a thermocouple for the Fluke meter. It doesn’t give us accurate temperature measurements with the RS meter. If we know the relationship between the fluke temperature and resistance, and the relationship between the fluke resistance and RS resistance, we can determine the temperature of the oven using the resistance measured by the thermocouple connected to the RS meter. This will allow us to connect the RS meter to the computer and measure the temperature as a function of time.

We need to first verify the relationship between the temperature and the resistance of the thermocouple is linear. Alternating between ohms and temperature on the fluke indicated it is linear(graph below).

Unfortunately the resistances measured by the two different meters are significantly different with the same temperature. They both measure around 4.4 ohms at room temperature, but at the melting point of solder the fluke measures 14.6 ohms compared to 42.8 by the RS meter. Fortunately, the relationship between these two are linear, as shown in the graph.

Using the relationships between temperature and resistance between the 2 meters, and the equations given to us by the trendline, we are able to determine
Temp=(RS ohms)*6.29 – 2.28