Sunteți pe pagina 1din 12

Lab 1

9/29/15
Jon Weber
Alex Rambke
Brad Zehr
Abstract
The experiments done in this laboratory show that there is electrical
noise being emitted and picked up by systems along with ways or possibly
reducing the noise. Ways of reducing noise experimented with are twisting
the wires and twisting the wires with foil wrapped around them to act as a
shield that grounds the unwanted noise. The results are show that the 95%
of the confidence bounds dont match what would be expected with a
reduction of noise. This shows that more testing would need to be done to
ensure that the results are conclusive since depending on the technique it
could go wither way.
Introduction
With the ever-advancing technology that is being produced their needs
to be a high degree of precision and accuracy. Todays technology needs the
signals coming into the instrumentation to be precise and accurate in order
to get a proper reading. Precision is having the signal be consistent with little
to no divergence. That is compared to accuracy, which is how closely the
signal is to the original one that was to be received.
One of the first steps to help achieve this is by first honing in and
making the signal more precise. The reason why the electrical signals arent
very precise right now is due to electrical noise. Electrical noise is the
unwanted electrical signal that distorts the original signal that was desired
[1]. The distortion can make it difficult for the signal being sent to the
instrument to be read accurately. This noise is caused by many things that
are either in the system or out side of it. These noises come from other
electrical systems that give off an electrical magnetic field. With all of the
different electrical fields being given off causes the signal to change with
every one that ends up interfering with the original signal. Once the precision

ME 303 Lab 1

Weber 1

of the signal is achieved it is much easier to correct the accuracy since the
signal isnt being scattered over a range or as big of a range.
The main objective is to reduce the noise that is coming into the
system since there are vast amounts of interference being emitted that it is
too difficult and unfeasible to reduce the sources. There are several ways of
reducing the noise entering the system like shielding the system from the
unwanted distortion. Also twisting the wires helps to target the noise given
off from the current and twisting the wires reduce the loop size and changes
the direction of current in the successive loops. With the loops changing one
right after another and reducing the loop size causes the signals to cancel
each other out. Shielding the system basically catches the noise that want to
enter the system and grounds it keeping the noise from interfering with the
signals.
The instrument used in this laboratory to measure consisted of a digital
multimeter and a DAQ system. A digital multimeter is a devise that has two
leads that are used to send a signal to the multimeter that measures it as
voltage, resistance, or current. The DAQ system consists of both software
and hardware. The hardware is a breakout box that allows one to attach a
variety of sensors that give off a signal that sends the readings to the
computer. In the computer there is software that takes the signals that are
received and converts that into what the sensor was reading. This can range
from pressure, current, voltage, temperature, or even sound.
Once the data is collected normal distribution needs to be determined
and that is done with Gaussian distribution equation:

The Gaussian distribution is used to see where the data tends to lie.
The equation consists on the variable x that is the data array in this case is
the voltage. Then there is u which is the variable for the true mean and

ME 303 Lab 1

Weber 2

which is the standard deviation both of which are from the data collected. and Sx are two ways
that standard deviation can be seen as can be solved by using the equation:

The standard deviation equation is used to show the amount deviation


of the data. The variables in this equation are xi,

x , and n. The xi is the data at

that point. x is the mean of the data. n is the size of the sample that was taken. One equation
that uses standard deviation in it to determine if there are outliers is the Chauvenet's criterion
equation:

If the result of the equation ends up higher then that of the

max

then

the data point has the possibility of being an outlier compared to the rest of
the data. This equation has a lot of interpretation when it comes to whether
or not the data point is an actual outlier or not. It is all up to the person
analyzing the data to determine if the data does seem as though it is and
outlier by looking at the data as a whole and making sure that it is for sure
not consistent with the rest of the data.
There is also the confidence bounds which is used to say how certain
that a percentage of your data lies between these two bounds. That equation
is as follows:

The percentage is determined when the


students t value table. That value of

is selected from the

is selected by using the percent in

which you want you confidence bounds to consist of and the degree of
freedom of the data set, which is n-1.

S x

is the standard deviation of the mean

which is calculated using the equation:

ME 303 Lab 1

Weber 3

This equation uses SE x

is the same as S x

from the previous equation. N in this

equation is the size of the data that you are finding the standard deviation of. All of these
equations will help one to determine and explain how data of different experiments compare and
what they show.
Procedure
The equipment used in this experiment consisted of a DC power
supply, National Instruments 68-pin connector block, digital multimeter,
computer with excel and labview, several wires, and foil. The digital
multimeter had accuracy to the thousandths place, as that was the amount
of digits past the decimal place it gave a reading for. The National
Instruments and labview showed a digital step size when the data was
zoomed in on. This can be seen in figure 1.1.

Figure 1.1 graph showing the digital step sized resulting from
the labview and National Instruments equipment.
For the first part of the lab the DC power source, computer, and digital
multimeter were used. For this part the power supply was plugged in and
was checked to make sure that the voltage was turned all the way down. The
power supply was then turned on and was turned to one volt, which was
displayed on the power supply. Once that was done an excel file was opened
ME 303 Lab 1

Weber 4

and was used to document the voltage over time. The voltage was read
using the digital multimeter by placing the two leads into the power supply
positive and negative nodes. The multimeter had the setting on 2V. The
voltage was to be read every ten seconds for 50 times. The data started at
time zero and went to 500 seconds. Once that was done the power supply
was turned down to zero volts and shut off.
The second part used the National Instruments 68-pin connector block
in which two wires were connected. These wires were connected to pin 67
and 68. Pin 67 labeled as AI GRN and pin 68 as AI 0 (AI 0+). The wire
connected to 67 was then connected to the negative node on the power
supply and the wire in 68 was connected to the positive node. The wires
were not twisted or wrapped making this the standard settings for the lab.
The connector block was then looked up in the computer to determine what
port it was plugged into to then make sure it matched with the labview
program. This was done since some of the equipment in the laboratory was
under a different port, which wasnt registered with labview causing no
readings to be taken. The labview program was then read over and changed
settings to RSE. Next the pins being used were input into labview along with
setting the acquisition rate to 1000 and the number of data points to 10,000.
After checking that the power supply was turned on. The digital multimeter
was then used to ensure that the voltage from the power supply was the
same as that of part one. Once that was done and stabilized the button to
record data was clicked on. After a few seconds the program ended and the
data was saved. The power supply was then turned to zero volts and
powered off.
For the third part the same process as part two was taken except for
the fact that changes were made to the wire to try and reduce electrical
noise. First the wires we taken off of the power supply nodes and were
twisted around each other ensuring that no metal from either wires were
touching. The wires were then attached back to the power supply making
sure that the wires were still going to the right nodes. The process for part
ME 303 Lab 1

Weber 5

two was then repeated for the wired twisted. After that the wires were taken
off of the node and foil was wrapped around the wires getting as close to the
ends as possible without having the foil touch the metal. The foil was then
attached to ground wire. The wires were now twisted around each other and
wrapped in foil. The wires were then attached back to the power supply
nodes and the process was then repeated from part two. These two tests
were to compare how these techniques did at reducing noise.
Result/Discussion
The data from part one did not change throughout all 50 data point.
This made the data not have a normal distribution. The difference from the
digital multimeter to the National Instruments DAQ system was apparent
once the data was collected and graphed. The data from the DAQ system
was accurate to about the 5th or 6th significant figure (as seen in figure 1.1)
compared to the digital multimeter that went to 3 significant figures.

Table 1.1 results of the standard deviation and range of data


Part 1

Part 2

Part 3

Part 3

Multimeter

Standard

Wires

Wires

settings

twisted

twisted and

0.000215857

0.000216613

shielded
Standard

6.72762E-16

0.000211167

deviation
(V)

ME 303 Lab 1

Weber 6

0.00148

0.00171

0.002073

1.021

1.023790678

1.023894026

1.023955045

1.021

1.024006536

1.024110639

1.024166212

+2

1.021

1.024222393

1.024327253

1.024377379

Range of
data
(V)
Average
(V)

Table 1.1 shows how the precision of the multimeter is very good by
looking at the standard deviation and how low it is. Also looking that how low
the range is shows that the multimeter is just one number, which is as
precision. Unfortunately that is not completely true due to the fact that the
step size and amount of significant figures is relatively low compared to that
of the DAQ system. This shows that it may look precise but its only half of the
story. The DAQ system is adherently less precise but much more accurate in
its reading do to how small the digital step size is. The original voltage is
likely to have been set to not exactly one volt. This is due to the in accuracy
of the power supply reading that displayed one volt but as seen in the data
above was slightly about one volt. This causes the accuracy of the other two
devices hard to see exactly since it was consistent through out the whole
experiment.

Table 1.2 95% confidence bounds

ME 303 Lab 1

Part 1

Part 2

Part 3

Part 3

Multimeter

Standard

Wires

Wires

settings

twisted

twisted and
Weber 7

shielded
Upper

1.021

1.024213759

1.024318588

1.024368932

1.021

1.023367598

1.023469463

1.023541157

bound
(V)
Lower
bound (V)
Looking at both table 1.1 and 1.2 and comparing

+2 and the

upper bound show that the standard deviation of part 2 and part 3 of the
experiment are all on the outside of that bound. This is interesting since it
was assumed to lie within those bounds for at least part 3 to go along with
the reduction of noise. Also looking only at the bounds it is seen that the
bounds grow as the noise reduction techniques are added. This in
unexpected since it is assumed that the data would be more precise causing
the bounds to be tighter as the noise reduction was added.
Table 1.3 Outlier elimination for test run 3 with the wires twisted
and wrapped in the foil shield

In table 1.3 it can be seen that 1.0252 in the group interval could be a
possible outlier first by observing the data and seeing how far out those two
data points are from the main set of data. Then the Chauvenet's criterion equation
ME 303 Lab 1

Weber 8

was used. The

max

for a data set of 10,000 is 4.718 as seen in the table. After

putting in the numbers for Chauvenet's criterion the result for i is about 5.896 which
is larger then max . That shows that it is a possible outlier and after looking at the data and that
graphs it was determined to be an outlier. This was the same case for a data point in the run with
the wires twisted. For the data in the standard settings run the outliers were not bigger then
max

making them remain in the data. The Chauvenet's criterion helped reduce the number of

outliers but with the amount of data taken it had little effect on most of the results except for the
min/max and range of the data.
Figure 1.2 Histogram from part 2 data at standard settings with a
bin size of .0002V

Run 1 @ standard settings

Voltage

Predicted Voltage

Figure 1.3 Histogram from part 3 data were wires were twisted
around each other with a bin size of .0002V
ME 303 Lab 1

Weber 9

Run 2 wires twisted

Voltage

Predicted Voltage

1.0234
Voltage (V)

Figure 1.4 Histogram from part 3 were the wires were twisted and
wrapped in foil to shield it with a bin size of .0002V

Run 3 Wires twisted and sheilded

Voltage

Predicted Voltage

Figures 1.2, 1.3, and 1.4 show close to the same thing except for the
fact that figure 1.4 shows a little bit higher of a frequency in the middle
section. This shows that the data was a bit more clumped together. This can
be seen when even looking at the standard deviation of the part 3 with the
wires twisted and wrapped in foil. The standard deviation of that part is lower
ME 303 Lab 1

Weber 10

then that of the part 2 and other ways of reducing noise in part 3. This goes
along with what was predicted. Reasons to why there wasnt as an abrupt
change can be due to many factors. Some of these factors could be that we
didnt run multiple trials and that we didnt try many different noise
reduction techniques. Another reason could be that the laboratory this
experiment was conducted in had an abundant amount of electrical noise.
Possibly doing the experiment in a more regulated and confined space may
help to show a greater difference.
That being said the data still shows a small change but doesnt
conclusively prove that it necessarily reduces noise much. Running the tests
multiple times across different devices would help to ensure and give more
conclusive results.
Summary/Conclusion
The results of the experiment seemed to have give mix results. For the
twisted wires with the shield it seemed to have reduced the noise but barely.
Then there was the noise reduction technique of just twisting the wires and
that result came in unexpected being worse then that of the control of
standard settings. Adding these noise reductions do show an impact but with
not a lot of significance. This lab shows that more experimentation may be
needed to help to get conclusive results.

ME 303 Lab 1

Weber 11

References
[1] Vijayaraghavan, G., Malcolm Barnes, and Mark Brown. "Electrical Noise
and Mitigation - Part 1: Noise Definition, Categories and Measurement | EE
Times." EETimes. Web. 29 Sept. 2015.
"How to Calculate Standard Error." How to Calculate Standard Error. Web. 29
Sept. 2015.
"Twisted-pair Wire's Effect on Common Mode Noise." - Electrical Engineering
Stack Exchange. Web. 29 Sept. 2015.

ME 303 Lab 1

Weber 12

S-ar putea să vă placă și