James Eastham

Normal Distributions in MatLab

2012-02-05T13:55:00.000-08:00

%% Normal Distributions in Matlab
% James Eastham
% Member, IEEE
% Created on: 2/5/2012
%
% The following code creates multiple normal distributions

figure('Color',[1 1 1]);
x = 17:0.1:28; %Plot range

pop_average = 21.98; %population mean
sigma = .4; %population sigma
normal_dist = normpdf(x, pop_average, sigma);
plot(x, normal_dist,'LineWidth',3);hold on;

pop_average = 22.89;
sigma = .244;
normal_dist = normpdf(x, pop_average, sigma);
plot(x, normal_dist,'--r','LineWidth',3);hold on;

pop_average = 21.84;
sigma = .78;
normal_dist = normpdf(x, pop_average, sigma);
plot(x, normal_dist,'-.g','LineWidth',3);hold on;

pop_average = 22.24;
sigma = .91;
normal_dist = normpdf(x, pop_average, sigma);
plot(x, normal_dist,'-ko','LineWidth',3);hold on;

legend('build 1','build 2','build 3','overall')
grid on;
title('Design Build Performance');
xlabel('Gain(dB)');

Fourier Transform for Dummies: Matlab Implementation

2012-01-24T13:22:00.000-08:00

The following code implements a simple Fourier Transform using cos and sin functions. This project was inspired by a short pdf I stumbled onto during an internet search on this topic. Here is a link to the original: FT for Dummies (I will gladly credit the author if contacted, I just have no idea who wrote the original document as it is not cited)

The following Matlab code, implements a simple Fourier Transform manually in Matlab.

% Fourier Transform for Dummies
% Matlab Implementation
% James Eastham
% Member, IEEE
% 1/22/2012
clear all; %clear all variables
close all; %close all figures
fs = 100; % sample freq
Ts=1/fs; % sample period
n=0:1*fs; % time window (fs=samples per second), 1 second
tn=n*Ts; % sample index
sig_a=5*sin(2*pi*5*tn+pi/3)+2*sin(2*pi*30*tn+pi/2)+2.3*sin(2*pi*8*tn); %create complex signal
sig_a = sig_a + 2*randn(size(tn)); % add noise
figure('Color',[1 1 1]); % plot orignial signal in time domain
subplot(2,1,1);hold on;
plot(tn,sig_a,'-r');
title('Time Domain Signal w/ Noise');
grid on;
% The following loop multiplies the original signal
% by sin and cos at each freq betwen 1 and fs/2
% each sample is then summed, after the loop
% the total sum for each freq is computed
for i=1:fs/2;
detect_sin=sin(2*pi*i*tn);
detect_cos=cos(2*pi*i*tn);
mult_sig_sin=detect_sin.*sig_a;
mult_sig_cos=detect_cos.*sig_a;
total_freq_sin(i)=sum(mult_sig_sin);
total_freq_cos(i)=sum(mult_sig_cos);
end;
total_freq_sin=total_freq_sin/(fs/2);
total_freq_cos=total_freq_cos/(fs/2);
total_freq=sqrt(total_freq_sin.^2+total_freq_cos.^2);
subplot(2,1,2);
stem(total_freq);
xlabel('Frequency');
ylabel('Amplitude');
Title('Frequency Domain');
axis('tight');
grid on;

Running MatLab 2007 (R2007a) 7.4.0 in Windows 7

2011-12-02T09:44:00.000-08:00

If you are having a problem where Matlab errors out on startup running Windows 7, try this it solved my issue.

Switch your "theme" to "Windows Classic" (right click desktop, select preferences, find the classic theme in the Basic and High Contrast Themes area)

Open Matlab. Enjoy.

Don't ask why... I didn't have time to research it. Seems to be related to Java.

Here are the some of the errors I was getting:
java.lang.NullPointerException
javax.swing.border.EmptyBorder.(Unknown Source)
com.sun.java.swing.plaf.windows.WindowsTableHeaderUI$XPDefaultRenderer.getTableCellRendererComponent(Unknown Source)

Adding a NE5532 Op Amp Model to LTSpice

2011-11-30T07:22:00.000-08:00

To add the NE5532 Model, follow the steps in my LM741 post (see link below), using the NE5532 model shown here below.

Special thanks to Uwe Beis who provided the TI 5534 updated model:
(http://www.beis.de/Elektronik/Electronics.html)

Adding the LM741 model to LTSpice

***** NE5532 Source: Texas Instruments NE5534
* C2 added to simulate compensated frequency response (Uwe Beis)

* NE5532 OPERATIONAL AMPLIFIER "MACROMODEL" SUBCIRCUIT
* CREATED USING NE5534 model from Texas InstrumentsAT 12:41
* (REV N/A)      SUPPLY VOLTAGE: +/-15V
* CONNECTIONS:   NON-INVERTING INPUT
*                | INVERTING INPUT
*                | | POSITIVE POWER SUPPLY
*                | | | NEGATIVE POWER SUPPLY
*                | | | | OUTPUT
*                | | | | |
.SUBCKT NE5532   1 2 3 4 5
*
C1   11 12 7.703E-12
C2    6  7 23.500E-12
DC    5 53 DX
DE   54  5 DX
DLP  90 91 DX
DLN  92 90 DX
DP    4  3 DX
EGND 99  0 POLY(2) (3,0) (4,0) 0 .5 .5
FB    7 99 POLY(5) VB VC VE VLP VLN 0 2.893E6 -3E6 3E6 3E6 -3E6
GA    6  0 11 12 1.382E-3
GCM   0  6 10 99 13.82E-9
IEE  10  4 DC 133.0E-6
HLIM 90  0 VLIM 1K
Q1   11  2 13 QX
Q2   12  1 14 QX
R2    6  9 100.0E3
RC1   3 11 723.3
RC2   3 12 723.3
RE1  13 10 329
RE2  14 10 329
REE  10 99 1.504E6
RO1   8  5 50
RO2   7 99 25
RP    3  4 7.757E3
VB    9  0 DC 0
VC    3 53 DC 2.700
VE   54  4 DC 2.700
VLIM  7  8 DC 0
VLP  91  0 DC 38
VLN   0 92 DC 38
.MODEL DX D(IS=800.0E-18)
.MODEL QX NPN(IS=800.0E-18 BF=132)
.ENDS

LM741 Datasheet

2011-11-21T20:11:00.000-08:00

Many student projects use the LM741 op amp. Here is a link to the datasheet.

http://dl.dropbox.com/u/30278073/LM741_Datasheet.pdf

Thick Lines in LTSpice

2011-10-26T11:04:00.000-07:00

Ever wonder how to thicken up the lines in your schematic or simulations? Turns out this is really easy.

1. Open the control panel (tools menu bar)
2. To thicken the schematic lines, Select the Drafting Options tab and check "Draft Thick Lines"
3. To thicken simulation lines, Select the Waveforms tab and check "Plot data with thick lines"

Exporting Data from LTSpice

2011-10-05T21:17:00.000-07:00

LTSpice waveform data can easily be exported for use in other graphing software like Excel.

To do this:

Step1: Run simulation, select File -> export
Step2: Select the traces of interest
Step3: A text file is saved in the LTC->LTSpiceIV root directory, the title will be "Draft1" by default or enter a desired file name.
Step4: Open the text file in Excel, or cut/paste from the text file

Parameter Sweeps in LTSpice - The .step Command

2011-09-28T20:54:00.000-07:00

To step a parameter in LTSpice, simply place curly brackets around a variable name for the value you wish to step. We need to tell spice that we want to "step a parameter". To do this we us the "parm" syntax after the .step command. We need to tell the step command the parameter (variable) we want to step, the starting and ending value and the step increment.

Example: Let's say we want to step the value of a resistor from 10 to 100 ohms in 10 ohm steps. For the value of the resistor we use {R1}. Next, we use the ".step" spice directive followed by the step range and increment. We place the following directive on our schematic:

.step param R1 10 100 10

Here is an example stepping a voltage supply:

Technical Presenations - 11 Tips to Help!

2011-09-27T18:16:00.000-07:00

#1 Organization: Technical presentations should contain an agenda/outline and a summary. It's always good to tell the audience what you're going to tell them and then summarize what you told them at the end.

#2 Conclusions not Guesses: Provide conclusions when presenting data. When charts/graphs are shown let the audience know the conclusion - what the information tells us. Don't expect the audience to draw their own conclusions from data, there will be many many questions which may not be relevant to the discussion.

#3 Fonts & Colors: Use large fonts. A funny rule of thumb is Mean Age - 10. If your audience is an average of 50 years old, use a font size of 40!! Avoid using colors which do not mesh well on a projector. For example, using a black background with dark blue lines is almost impossible to see on a projector.

#4 Labels & Legends: All charts should be clearly labeled and should always contain a nice legend.

#5 Anticipate & Be Ready: Anticipate questions that might come up and be ready to answer - or better yet keep these from coming up.

#6 "I don't know" is better than making stuff up. It's best to say, "I'm not sure" than to try and fumble your way through and guess at information.

#7 Stand Up!! Stand and speak so everyone can hear. Look at your audience, speak clearly, and make sure you are speaking loud enough so the room can hear.

#8 Technical Accuracy: Technical accuracy is king. If calculations, data, information is incorrect early, your entire presentation will be difficult. Unfortunately (or Fortunately in some cases), it's human nature to find faults, so make sure conclusions are correct, calculations are correct, and the data supports conclusions.

#9 Pace, don't run.. walk: You are taking your audience on a journey. No need to run. Keep a steady slow pace, speaking clearly, and slow enough for folks to follow. Don't directly read the slides, if you need notes it's best to print out notes to read. Reading the exact bullets on your slides is a no no. This approach is boring for everyone. Slides should simply highlight the points you are making verbally.

#10 Number of Slides: Plan on 3 minutes per slide. If you have a large amount of data, plan on 5 minutes per slide to explain the information and answer any questions.

#11 Nervous? Show up early and give your presentation to an empty room. If you get really nervous speaking in front of groups, it really helps to show up early, walk around the room, and present the presentation to an empty room. Practice is very important. If you can have someone sit and review this really helps.

Super Crunchers Book Review & Thoughts

2011-09-01T12:36:00.000-07:00

I. Ayre, Supercrunchers: Why Thinking-by-Numbers Is the New Way to Be Smart. New York: Bantam, 2007.

I found this to be a fascinating read. I knew about much of the data collection that was going on but did not really grasp the magnitude of the random trials that were going on all around me. I work in a very data driven environment (high tech) and was shocked to learn that other professions are JUST NOW starting to use regression and other data analysis techniques. The fact that physicians are just starting to use data mining, regression, correlation, etc. to help diagnosis patients was sort of baffling. For some reason, I thought when my general doctor left the room and returned after 15 minutes she was actually looking in some master database for my symptoms. I never asked why she made certain conclusions, but after reading this book I'm going to start asking the origins of her diagnosis. I found the chapters with medical field examples to be the most interesting. As I mentioned above, I sort of already knew about the things airlines and credit card companies were doing.

During the course of this book, I often thought about the ethical issues surrounding the methods presented, especially the random trials. To deny patients of potentially helpful drugs for the sake of conducting trails presents some ethical dilemmas. One could argue that this 'holding back' of critical care is for the greater good, helping future patients with the statistically correct forms of treatment. A similar dilemma presents itself when conducting random trails using impoverished sections of society. In one case crucial financial help may be denied for the sake of random trials, with no information provided as to why the recipient was denied assistance. In other cases, impoverished families are given all sorts of benefits. Although I agree that efficient use of monies is of utmost importance, it's a bit depressing to think aid is being withheld.

Another potential ethical issue was presented regarding companies displaying, or burying in this case, information regarding warranties on their web sites. Companies are using random trails to determine if they should be up front and honest regarding warranty information, or if they are better off (in terms of sales) to bury this information on their website, essentially hiding it from the consumer. The author presented several cases where corruption and cheating was actually exposed using supercrunching. In one example, through extensive data analysis, a study determined that some basketball games were subtlety rigged during the last few minutes to reduce point spreading. It's encouraging to think about how data mining and analysis techniques can help reduce corruption, cheating, and selective targeting.

Dr. Ayres concluded his book with a chapter about the 2SD rule. This is the basic rule that 95% of a population is within two standard deviations of the mean. He provides some nice examples of how the rule works for every normal distribution. He provides real-world examples related to basketball game spreading, adult heights, and political races.

As a data driven engineering type, I would have loved to see some charts and figures supporting some of the examples and cases presented. The book did spark enough interest for me to go and research some of the cited cases myself out of curiosity. Charts and figures could have been used to better illustrate some of the correlations and regressions presented.

I found the book to be an easy read but sort of inconstant in terms of explaining some of the more technical aspects. In the beginning of the book the author did a good job explaining data mining and analysis at a level even 'non-techies' could understand. The author went into a fair amount of detail explaining the methods so that readers from most backgrounds could understand the foundation of the text. However, when regression was introduced it wasn't explained very well and did not show any graphical examples. The author writes "A regression is a statistical procedure that takes raw historical data and estimates how various causal factors influence a single variable of interest." Beyond this short explanation, regression as a procedure or method is not described. Given the amount of non-technical explanations explained earlier in the book, I expected a better foundation to be provided to the reader. Personally, I know what regression is and how to apply this method, but others reading this book could benefit from a bit more explanation and background. Ayres also fell a bit short when explaining standard deviation. He explained the term but did not provide an example of how standard deviation is calculated. He mentioned, in more than one section, how easy the calculation is in Excel but never showed the manual calculation.

Overall, I left this book with somewhat mixed feelings. I was very excited about the future potential of supercrunching in the medical field. In contrast, I'm a little scared about where we might be headed in terms of commercial driven supercrunching. Although these methods can help both the advocate as well as the big corporations, the privatization of huge data concentrators means that the folks with the most money (companies) have access to the best and biggest data sets. My worry is that the advocate is going to be 'out gunned' as compared to the folks looking for ways to make more and more money. With examples like Enron and the recent housing "bubble" (crash), we've seen what happens in a greed driven environment. I'm finishing up this review on an international flight from Houston to Costa Rica. My supercrucher mixed feelings continue as I'm stuck on the last row of the plane with a seat that doesn't recline, and a shortage of food when the cart finally makes it to me. The flight was overbooked, not one empty seat and hardly any extra room for carry-on bags. I'm typing these last few sentences with one hand because space is so confined these days that one has no hope of actually using two hands to type on the computer. All for the sake of higher profilts.

I'd highly recommend this book to just about anyone, it's a great read, very interesting.

Calculating Focal Adjustments Using Regression and Multiple Goal Linear Modeling

2011-07-21T09:19:00.000-07:00

Last year I took a course on linear programming and spread sheet modeling (ETM540). This course we really good and provides some very powerful tools for many areas. I used LP for calculating focal adjustments and performance merit increases (raises). I've posted a paper to help other engineering managers with the difficult task of figuring out how much of a raise to give team members while staying within budget and recognizing top performance. This approach removes the subjectivity which can sometimes influence traditional approaches.

The paper can be viewed here.

Hata Model - DTV Propagation

2011-03-04T22:02:00.000-08:00

The Hata model is optimized for urban areas and is also known as the Okumura-Hata model. This model has two additional varieties for transmission in suburban and open area. I implemented this model in Matlab so that the user can define the desired model variety by changing one variable. The Hata model applies for frequencies from 150MHz to 2000MHz, transmitter heights up to 200 meters, and link distances up to 20km. I selected this model to see if it was more/less accurate than others. I also implemented the Cost231 (officially named COST-Walfisch-Ikegami-Model) model which is a slight variant of the Hata model used in wireless communications.

Here is the matlab implementation:

% Hata path loss calculated below

hata = 69.55 + 26.16 * log10(f) - 13.82 * log10(ht_m)-hata_ch+(44.9-6.55*log10(ht_m))*log10(d_km);

MOSFET Applet

2010-11-22T21:01:00.001-08:00

Here is a cool MOSFET applet online:

http://www-g.eng.cam.ac.uk/mmg/teaching/linearcircuits/mosfet.html

Modeling LM741 Slew Rate in LTSpice

2010-10-18T21:17:00.000-07:00

Modeling the slew rate of the LM741 can be accomplished using a simple voltage follower as shown above.

The slewing can be observed with a transient analysis.

Installing Solver in Excel 2007

2010-10-10T10:26:00.000-07:00

Solver is a very powerful add-in for solving linear programming problems. To install Solver, complete the following steps:

Step 1. Click the office button, and click the excel options button

Step 2: Click Add-Ins, and then click "GO" to manage add-ins

Step 3: Now, just click to add the Solver add-in and click OK

Step 4: After install you can start using the Solver! Solver is now located on the Data -> Analysis toolbar.

Modeling a basic inverting op amp in LTSpice

2010-10-06T21:45:00.001-07:00

Here is a basic op amp schematic in LTSpice using the "universal op amp" component.

Modeling a VCVS in LTSpice

2010-09-29T22:46:00.000-07:00

The ideal model for a VCVS (voltage controlled voltage source) is shown below:

Modeling a VCVS in LTSPice is pretty simple using the "voltage dependent voltage source" component in LTSpice. This component is called "e" in the symbol library. The symbol as inserted into LTSpice is shown below:

A VCVS amplifer with the following specifications is shown below:
Rin = 1k ohm
A = 50
Rout = 100k ohm
Rsrouce = 1k ohm
Rload = 100 ohm

Notice we use a standard voltage source (Vs, "voltage" component in LTSpice") connected to our VCVS. The source resistance, input resistance, output resistance, and load resistance are all represented in the schematic.

Creating Random Numbers in Matlab

2010-08-09T14:41:00.000-07:00

Generating a matrix of random numbers:
The following code generates a 1 by 1000 matrix of 1000 random numbers between 0 and 1.

rand(1,1000);

Generating a Number between a and b:
The following code generates a uniform whole random number between 1 (a) and 100 (b).

a = 1;
b = 100;
x = round(a + (b-a) * rand);

a and b can be adjusted as desired. For a normal Gaussian random number, the randn function can be used in place of rand above.

Here is the Matlab code for the plot above.

%% Generating random numbers in Matlab
% James Eastham
% Member, IEEE
% Revision: R1-0-0
clear all;
close all;
%% Generating 1000 random numbers using rand and randn
% This section generates a matrix of 1000
% random numbers between 0 and 1 using the
% rand and randn functions. The rand function
% creates a uniform distribution while the
% randn function generates a normal (Gaussian)
% distribution
x = rand(1,1000); %generates a 1 by 1000 matrix
figure('Color',[1 1 1]);
subplot(2,1,1);
hist(x);
title('1000 Random Numbers: Uniform');
x = randn(1,1000); %generates a 1 by 1000 matrix
subplot(2,1,2);
hist(x);
title('1000 Random Numbers: Gaussian/Random');
%% Generating a random number between 1 and 100
% The following code demonstrates a method
% for generating a random number between
% 1 (a) and 100 (b). The round function is
% used to round the number to the closest whole
% number.
a = 1;
b = 100;
x = round(a + (b-a) * rand)

Effective Team Leadership: A Summary of "Strategies of Effective New Product Team Leaders"

2010-05-30T21:11:00.000-07:00

A. Jassawalla and H. Sashittal, "Strategies of Effective New Product Team Leaders," California Management Review, vol. 42, no. 2, pp. 34-51, Winter 2000.

ARTICLE OVERVIEW
Published in the Winter of 2000 volume of California Management Review, A. Jassawalla and H. Sashittal’s article titled “Strategies of Effective New Product Team Leaders” presents a process for team leader transformation aimed at improving new product development (NPD).
Traditional NPD literature provides team leaders with guidance on what they should do to be effective. The literature and main-stream training provides NPD team leaders with a toolbox of “do this and you will reap the rewards” type strategies. Many of these tools include “soft” skills such as effective communication, human side interaction, and conflict resolution. Jaswalla and Sashittal believe that simply knowing the best practices of NPD team leadership isn’t enough to guarantee successful new product development. Team leaders and Senior Managers alike, must commit to developing critical interpersonal skills which companies tend to overlook.
The authors present a new process of team leadership which includes effective ways to mitigate typical environmental forces encountered in NPD. The article provides helpful strategies for interpreting these forces, developing action plans, and examples of actions team leaders take to help transform their teams.

KEY LEARNING POINTS
Jaswalla and Sashittal present the key finding of their study; five key milestones as well as environmental forces with corresponding objectives and strategies employed by effective team leaders. These objectives and strategies are necessary for team transformation and rapid new product development.
Often NPD decisions are driven by a controlling R&D department. In this environment, NPD is hampered due to a sequential development process and disconnected team members. Because R&D often plays a dominating secretive role, critical input from other functional team members may not be sought after or volunteered. Because of these factors, other functional groups may not really feel committed to the development process. Jaswalla and Sashittal found that the most effective team leaders actively developed relationships with key department heads in other functional areas. By actively engaging the other key functional groups, disseminating information to the entire team, and holding each member accountable for input into the team as well as their own output, effective team leaders ensure commitment from the entire NPD team. In general, I agree with the authors’ analysis. However depending on the organizational structure of the company, obtaining commitment and holding members accountable might be very difficult. I’ve seen cases where members are committed to the team but their functional area managers may not be as open to accepting responsibility for NPD delays.
Jaswalla and Sashittal found that in order to have rapid decision making in new product development, information must be shared openly with the team. Effective leaders eliminate information that is hard to understand or explain. These leaders boost communication through increased meetings, the collation of information, and in some cases physically co-locating key members. Personally, I think establishing the proper level of communication is difficult. The solutions presented by the authors may not work for some organizations and teams. For example, increasing the frequency of meetings may actually hamper R&D as engineers may be stuck in meetings all day.
The most effective leaders studied viewed themselves as facilitators, striving to minimize bureaucratic forces which can hinder team performance. They serve behind the scenes to seek out members with the best set of skills and talents. They focus on human interaction to help bring together the right mix of team members. The study showed that by focusing on human interaction, team leaders were able to foster more effective communication and information sharing within the team.
Finally, the most effective leaders have a deep interest in change and risk taking. They focus the team on learning, supporting risky ideas and innovative actions. Effective team leaders allow members to have autonomy and the freedom to make their own mistakes. In my opinion, trying new things and taking risks is good but team leaders should call “time out” if they see a potential risk being taken which could lead to major NPD slippage. Risks should be taken, but the impact of potential failures must be well understood by the team.

FOLLOW ON RESEARCH
S. Zhang et al. (2005) provide useful strategies for developing virtual team leaders. Three often over looked factors were analyzed: media richness of communication technology, goal-frustrating events and leader-follower gender. The authors propose that media richness of communication provides a conduit by which team leaders inspire team performance. S. Zhang et al. provide team leaders with motivational strategies for dealing with goal setbacks. This paper proposes that female leaders add a certain level of charisma and social-emotional skills which can improve virtual team performance [1].
Thamhain (2007) reiterates many of the attributes of effective team leaders discussed in this article as well as some other important factors. The author stresses the importance of personal and professional needs of team members. If the goal of the team helps fulfill personal and professional needs of its members, commitment tends to follow and the overall performance of the team benefits [2].
Assmann and Gallenkamp (2009) explored the topic of communication team leader trustworthiness in virtual teams. Building trust is an important task of the team leader, which was discussed in this brief. However, building trust across different cultures in a virtual team setting presents many challenges. Assmann and Gallenkamp highlighted the need for team leaders to understand the influence culture, self-disclosure, and communication media has on perceptions of trustworthiness of the team leadership of virtual teams [3].
IMPLICATIONS
This article certainly points to the root cause of many delays in new product development. In order to be successful, teams need committed functional members who feel they are a major part of the development team. Team leaders must work to improve communication and remove “hidden agendas” and other barriers.
Team leaders are possibly the most important members of new product development teams. The insights presented in this article are essential for effective team leadership. Companies invest heavily in training team leaders on traditional “tools of the trade”. However, arguably more important is training and development in the “soft” skills presented in this article. Many companies drastically overlook the importance of these skills and thereby miss an opportunity to create more effective teams and speed new product development.

REFERENCES

[1] S. Zhang et al., "Leadership Styles in Virtual Team Context: Limitations, Solutions, and Propositions," presented at Proceedings of the 38th International Conference on System Sciences, Hawaii, 2005.

[2] H. Thamhain, "Leadership Lessons from Managing Technology-Intensive Teams," presented at PICMET 2007 Proceedings, Portland, Oregon, 2007.

[3] J. Assmann, J. Gallenkamp, "Communication and Leadership Trustworthiness in Virtual Teams: An Empirical Comparison of the US and China," presented at Proceedings of the 42nd International Conference on System Sciences, Hawaii, 2009.

Simple FM Modulator in LTSpice

2010-05-30T20:53:00.000-07:00

Here is a simple FM modulator driven by a staircase voltage.

http://tinyurl.com/25rdzg7

This FM modulator can be used as a signal source for simulating an FM amplifier or other circuit.

NOTE: This modulator was obtained from the LTSpice Yahoo! group. I did not develop this schematic. This schematic was posted by the user "Petzen".

Portland Area TV Channels

2010-05-21T19:23:00.000-07:00

Here is a up to date list of Portland area channels.

http://simonslick.com/TV/

RF Link Budget Calculator

2010-05-07T17:34:00.000-07:00

Here is a quick link budget calculator.

http://spreadsheets.google.com/pub?key=tzm8TYYcW1Qp6VO3Fcuxayw&output=html

Enjoy!

Creating a Terrain Profile for DTV RF Propagation Analysis

2010-04-30T18:22:00.001-07:00

ATTN: This tutorial only works for FireFox, do not use IE, the profile website doesn't work with IE for some reason.

Step 1: Determine the coordinates of the transmitter (in decimal format by clicking on the coordinates in Wikipedia, uses geohack)

http://en.wikipedia.org/wiki/KOIN-TV

Step 2: Determine the coordinates of the receiver location (your house), also in decimal format. This step can be skipped as you can click directly on the map in step #3 to locate the receiver location.

http://mapki.com/getLonLat.php

Step 3: Create a profile using the coordinates from above. Enter the coordinates as 45.51611,-122.732778

http://www.heywhatsthat.com/profiler-0904.html

Step 4: Adjust the height above terrain for the transmitter and receiver by clicking on the height hyperlink next to the coordinates.

Step 5: Note if you have line of sight (LOS) and the distance of the path. The path distance will be used later when we calculate path loss and link budgets.

Portland Oregon DTV Channel Frequencies (HDTV)

2010-04-23T20:31:00.000-07:00

Determining receiver sensitivity using kTB, NF, and S/N requirement

2010-04-16T19:07:00.001-07:00

Receiver sensitivity is the minimum input signal required to produce a required signal to noise ratio. In other words, it answers the common question in virtually any communication system "Is my signal good enough?"

Many system specifics are based on receiver sensitivity. For example, in a cellular system it can help you determine "good" coverage from a cell tower antenna. In digital TV receiver sensitivity can be used to determine if you need an antenna on your house to have a 'good' signal.

In test and measurement, receiver sensitivity is used to determine what is required to make a 'good' (accurate & repeatable) measurement. If you know the receiver sensitivity for a spectrum analyzer for example, you can make judgments about the quality of measurements and how to best setup the measurement.

Calculating rx_sens (receiver sensitivity) is pretty straight forward.

Step 1: Calculate kTB (noise floor) for the bandwidth of interest
where:
k = Boltzmann constant (1.38*10E-23)
T = temperature in Kelvin
B = bandwidth of interest in Hertz

Step 2: Convert to dBm
30+10*log(kTB)

For example: For a bandwidth of 30KHz
example: For a signal with a bandwidth of be 30Khz
Noise Floor = kTB = 1.38E-23 * 300 * 30000 = -129dBm

Step 3: Estimate receiver senstivtiy
Rx_sens = Noise Floor + NF of the receiver + S/N required

Example: For 10dB noise figure receiver, with a 20dB S/N requirement:
RX_Sens = -129dBm + 10dB + 20dB = -99dB

Now we know what level we need at the receiver. Stay tuned for the next post regarding link budgets. We'll see how to use this number in our link calculations to determine if the transmitted power level is 'good enough' to meet our RX_Sens requirement.

%% Receiver Sensitivity
% James Eastham
% Member, IEEE
% Revision: R1
clear all;
close all;
b = [1 10 100 1e3 10e3 100e3 500e3 1e6]; %bandwidth every decadeHz
k = 1.38E-23; %Boltzmann's constant
T = 300; %Temp
n_floor = k*T*b;
n_floor_dBm = 30 + (10*log10(n_floor));
figure('Color',[1 1 1]);
bar(n_floor_dBm,b);
ylabel('bandwidth(Hz)');
xlabel('Noise Floor (dBm)');
title('Receiver Sensitivity vs. Bandwidth');

Railways of the World the Card Game

2010-04-10T16:35:00.000-07:00

My first serious attempt at game design is going to print! RotWTCG is scheduled to hit stores June 2010.

Here is the boardgamegeek entry:
http://www.boardgamegeek.com/boardgame/68228/railways-of-the-world-the-card-game

Interested in ordering a copy?
http://www.funagain.com/control/product?product_id=021660

RF Connector Types & Proper Connector Care

2010-04-02T17:56:00.000-07:00

Chess-set photo found on leapsecond.com!

There are many different types of coaxial RF connectors. The standard types can be reviewed on wikipedia at: http://en.wikipedia.org/wiki/RF_connector#Standard_types

The types which students will most encounter include the BNC, F connector, N connector, TNC connector, SMA, and UHF connector. All of these connectors can be reviewed online. Each connector has a male and female version. The N-Type is a very common threaded RF connector seen on most RF equipment.

There are a few important "Do's and Dont's" associated with connector care. I found this list on the microwaves101 site and found it usesful (http://www.microwaves101.com/encyclopedia/connectorcare.cfm). Here are the ones I find most important for students:

#1: Don't use pliers on a stuck connector - ever. A properly sized wrench is the tool you should be using.
#2: Don't make critical measurements with dirty connectors. Make sure connectors are clean (http://www.microwaves101.com/encyclopedia/connectorcare.cfm#cleaning)
#3: Don't use part of a calibration kit for adapters / connectors.
#4: Use a torque wrench (or gently tighten by hand for most non-critical classroom measurements).
#5: Don't overtighten, or tighten too much!!! These connectors in some cases cost more than your tuition. Be extra careful with them, pretend they are made of eggshells and filled with explosives.
#6: Don't 'spin' the connectors joining each other, only turn the threaded sleeve. If you 'spin' the junction you can wear our the mating surfaces.

Here is a photo of a damaged SMA from work. The cable this was connected to cost about $1,000. That's not the bad part, the damaged cable also caused about 4 hours of downtime for a well paid engineer! BE CAREFUL WITH CONNECTORS!!!!

DSP Tutorial #11: Loading, Plotting WCDMA Signal

2010-02-21T15:50:00.001-08:00

The raw WCDMA signal can be located on the menu to the right, or by using the link below:

DSP: WCDMA File (wcdma2.csv)

The sample frequency used to sample this signal was fs=32496000

The following code loads the signal and produces a time series plot.

%% Loading WCDMA 3G Signal
close all;
clear all;
fs=32496000; %sample frequency
x=load('wcdma2.csv');
figure('Color',[1 1 1]);
plot(x);
axis('tight');
title('Time Domain Plot of WDMA Modulated Signal');

DSP: Tutorial Post #10: Windowing

2010-02-21T11:04:00.000-08:00

- DFT can only be performed over a finite-time sample interval

- We can think of the DFT input singal as the product of an input signal existing for all time and a rectangualr window whose magnitude is 1 over the sample interval

- Anytime we take the DFT of a finite-extent input sequence we are, by default, multiplying that sequence by a window of all ones and effectively multiplying the input values outside the window by zeros

- Remember: The continuous Fourier transform of the rectangular window IS the sinc function

Windowing premultiplies input data supplied to the FFT with a value that smoothly decreases to zero at each end of data. The purpose is to reduce "leakage" aberrations in the FFT output that are introduced by sudden changes in the data at the start and end of data. Windowing reduces DFT leakage by minimizing the magnitude of the sinc function’s sin(x) / x sidelobes. This is accomplished by forcing the amplitude of the input time sequence at both the beginning and the end of the sample interval to go smoothly toward a single common amplitude value. (taken from Understanding Digital Signal Processing – Lyons).

Matlab supports the following windows:
barthannwin,bartlett, blackman,blackmanharris,bohmanwin,chebwin,flattopwin,gausswin, hamming,hann,kaiser, nuttallwin,parzenwin,rectwin,triang,tukeywin

In this example we will compare the sidelobes created by the rectangle window as compared to the blackman window.

%% Example of Windowing
close all;
clear all;
% Example of windowing
freq = 12; % continues time signal frequency in Hz
fs=800; % sample frequency 800Hz (how many samples per second)
Ts=1/fs; % sample period (time between consecutive digital captures of the continues time sigal)
n=0:1*fs; % sample index (fs=samples per second), 2 second (how long to watch the signal)
tn=n*Ts; % our x axis, descrete measurements
x=cos(2*pi*freq*tn); % create signal
w=blackman(length(x)) % create a blackman window the same length as x(t)
c = x.*w'; % multiply our signal and window, the ".*" completes element by
% element multiplication. The "'" transposes our window matrix
figure('Color',[1 1 1]);
plot(tn,c,tn,w,'--r');
title('x(t) windowed with Blackman');
axis([-.5 1.5 -1.5 2.5]);
% Freq domain plot
[Y,f]=ComputeSpectrum(x,fs,2^14);
[X,f]=ComputeSpectrum(c,fs,2^14);
figure('Color',[1 1 1]);
plot(f,20*log(Y),f,20*log(X),'--g');box off;hold on;
xlabel('Frequency');
ylabel('Amplitude');
title('Spectrum w/ Window');
axis('tight');
legend('Rectangular Window','Blackman Window');

DSP Tutorial #6: Looping in Matlab

2010-02-14T14:02:00.000-08:00

Looping is helpful if we need to repeat certain specific commands. In this example we will use the for loop to create a sinusoid of various frequencies as a function of the loop variable. The following code creates six (6) plots, each with a different 'frequency' as determined by the conditional if statement based on our loop variable i.

close all;
clear all;
for i=1:6;
if i==1,freq=10;,end;
if i==2,freq=25;,end;
if i==3,freq=50;,end;
if i==4,freq=75;,end;
if i==5,freq=100;,end;
if i==6,freq=125;,end;
fs = 8000;
Ts=1/fs; % sample period
n=0:1*fs; %sample index (fs=samples per second), 1 second
tn=n*Ts;
x=cos(2*pi*freq*tn);
% Time domain plot of x(t)=cos(2*pi*5*tn)
figure('Color',[1 1 1]);
h=plot(tn,x);box off;
xlabel('Time(s)');
ylabel('Amplitude');
title(['x(t)=cos(2*pi*freq*tn) in the time domain (freq=',num2str(freq),'Hz)']);
end;

DSP: Lowpass Elliptical Filtering

2010-02-12T20:56:00.001-08:00

The following function creates a lowpass elliptic filter. By implementing this filter in a function we can easily call the lowpass filter in our code, passing our signal "x", sample frequency "fs", and desired cutoff frequency "fc".


function [y,N,B,A]= lowpass(x,fs,fc);
% Description: This function uses an elliptical filter
% to filter out high frequencies.
%
% [y,N,B,A] = lowpass(x,fs,fc)
%
% Variable Definition:
% x: Input signal
% fs: Input signal sampling frequency in Hz
% fc: cutoff frequency
%
% y: Filtered signal (output)
% N: Filter order
% B: Feedforward (FIR) coefficients
% A: Feedback (IIR) coefficients
Rp = 0.5;   % Ripple in dB (passband)
Rs = 20;    % Attenuation
mf = fs/2;  % Maximum frequency
Wp = fc/mf; % Cuttoff frequency
Ws = 1.2 * Wp; %Stopband frequency
% First, determine the minimum filter order to meet specifics
[N,Wn] = ellipord(Ws,Wp,Rp,Rs);
% Next, Design the filter (find the coefficients)
[B,A] = ellip(N,Rp,Rs,Wn,'low');
% Now, apply the filter (filter the signal)
y = filtfilt(B,A,x);
% Plot the filter
figure('Color',[1 1 1]);
freqz(B,A,2^14,fs);

DSP: Noise Averaging WCDMA Example

2010-02-10T13:27:00.000-08:00

The following example show noise averaging affects for a WCDMA signal. Coherent averages were taken of a WCDMA signal with additive noise k=.1.

The first plot shows a pure WCDMA modulated signal with additive noise.

Next, the signal is plotted with three (3) noise averages:

Finally, ten (10) averages are simulated:

The plot below shows a zoomed in area of the shoulder.

2010-02-06T18:50:00.000-08:00

Audio Processing Using Basic DSP Techniques

Prepared by: James Eastham, Member IEEE

In this tutorial we will use an audio signal as a practical example of common DSP techniques. We will apply the concepts of sampling, time-domain analysis, spectral analysis, and digital filtering. We will load an external sound file in wav format into Matlab for this analysis. Basic Matlab plots will be used to display our results. The Spectrogram will also be introduced.

Before we start, let’s review audio frequencies. Taken from Wikipedia: “An audio frequency is characterized as a periodic vibration whose frequency is audible to the average human. While the range of frequencies that any individual can hear is largely related to environmental factors, the generally accepted standard range of audible frequencies is 20 to 20,000 hertz. Frequencies below 20 Hz can usually be felt rather than heard, assuming the amplitude of the vibration is high enough. Frequencies above 20,000 Hz can sometimes be sensed by young people, but high frequencies are the first to be affected by hearing loss due to age and/or prolonged exposure to very loud noises.

We will use the Wikipedia definition as a guide to help us filter the audio signal.

To load the wav file (Pule Code Modulated file) we use the wavread function in Matlab. We need to know the sample frequency that was used to produce the file to play it at the correct speed. The following code will open the file and play it using the soundsc function.

clear all; %clears all variables
close all; %closes all plots
fs=8000; %sample frequency
Ts=1/fs; %sample period
n=0:3.5*fs-1; %sample index (fs=samples per second), 3.5 seconds
tn=n*Ts;
y = wavread('fp3.wav'); %open the message
y=y(1:28000); % This simply trims the file to the desired length

Plotting the Audio File in the Time domain:
The following code will plot the audio file in the time domain.

figure('Color',[1 1 1]);
plot(tn,y);
axis('tight');
title('Speech file fp3.wav');

Plotting the Audio File in the Frequency Domain:
The following code will plot the audio file in the frequency domain. This section uses the ComputeSpectrum function we created in a previous blog post. Notice the spectral content, most of the content is below 500Hz. This is good, this is what we expect from an audio signal.

%******** Amplitude Spectrum of *.wav file *************
figure('Color',[1 1 1]);
[X,f] = ComputeSpectrum(y,fs,2^14);
plot(f,20*log(X));
title('Amplitude Spectrum of *.WAV');
xlabel('Frequency (Hz)');
ylabel('|X(f)|');
axis([0 4000 0 100]);

Filtering the Audio File:

Next we will use digital filtering to filter out undesired signals from our audio file. Matlab provides many different filters, in this example we will use the Butterworth filter. To read more about the Butterworth filter see: http://en.wikipedia.org/wiki/Butterworth_filter

The following code implements a tenth order (ten pole) low pass Butterworth filter with a cutoff frequency of 3400Hz. Cutoff frequencies are specified between 0.0 and 1.0 where 1 corresponds to half the sampling frequency: fs / 2 (4000 in our example). For example, for a cutoff frequency of 3400Hz, we specify this as 3400/(fs/2).

fNorm = 3400/(fs/2);
[b,a]=butter(10,fNorm,'low');
filterLow_y=filtfilt(b,a,y);
freqz(b,a,128,fs);
figure('Color',[1 1 1]);
plot(tn,filterLow_y);
axis('tight');
title('Speech file fp3.wav Low Pass Filtered at 200Hz');
soundsc(filterLow_y);
%******** Amplitude Spectrum of filtered *.wav file *************
figure('Color',[1 1 1]);
[X,f] = ComputeSpectrum(filterLow_y,fs,2^14);
plot(f,20*log(X));
title('Amplitude Spectrum of *.WAV');
xlabel('Frequency (Hz)');
ylabel('|X(f)|');
axis([0 4000 0 100]);

The figure below shows the frequency spectrum of the filtered signal; notice how all frequencies above 3400Hz are basically gone now.

Matlab provides the calculated coefficients if needed for later use or reporting. These are contains in the variables [a,b].

Filtering the Audio File (continued):

Matlab includes function the following Butterworth filters:

* 'low' : Low-pass filters, which remove frequencies greater than some specified value.
* 'high' : High-pass filters, which remove frequencies lower than some specified value.
* 'stop' : Stop-band filters, which remove frequencies in a given range of values.

To implement a ‘high’ pass filter, simply specify the cut off frequency as we did above, and use the ‘high’ keyword in the function call.

To implement the ‘stop’ band, a range must be defined. The following code will define this band. This code basically ‘chops’ out all frequencies from 750Hz to 950Hz.

% bandstop example
fNorm = [750/(fs/2), 950/(fs/2)];
[b, a] = butter(10, fNorm, 'stop');

The Spectrogram:

The spectrogram plot allows use to not only see which frequencies are present in our signal, but also when. Basic spectral analysis using the FFT does not allow us to see what spectral content is present at a given time. The spectrogram is a useful tool in spectral analysis. To display the spectrogram, the following code is provided. The 512 below is the number of samples that are used for the DFT, this becomes a grouping factor per column. Think of this as the granularity of the analysis, 512 is typically adequate for analysis. The spectrogram presents interpolated colors of magnitude vs. time. In the plot below, red for example means there is more frequency content at that frequency than say the light blue at a given time. Our audio file contains noise and voice harmonics therefore many frequencies are seen at any given time, however the magnitude of the voice range below 500Hz is see as the darkest (darkest red / black).

figure('Color',[1 1 1]);
specgram(y,512,fs);

Be Careful!!

2010-02-06T13:02:00.001-08:00

DSP: Obtaining Raw Stock Data for Analysis

2010-02-06T12:50:00.000-08:00

Raw stock data can be obtained by using Google Finance. Search for the stock of interest, and use the "Historical Prices" link on the left. You will see an option to download the data to a spreadsheet. Once you have the spreadsheet downloaded, you can delete the unnecessary columns (only need to keep the close price) and save the file as text for use with Matlab.

DSP: Using a basic digital filter: Stock market example

2010-01-31T21:40:00.000-08:00

Digital filters are very powerful tools in DSP. In this example we will use a digital filter to help with stock market analysis. In the figure above, you can see the raw data plotted with a 5 day moving average and a 20 day moving average. These two moving average lines can be used for buy/sell predictions.

To calculate the 20 day moving average, the window size and filter coefficients must be established. The following Matlab code sets the window size and calculates the filter results. The five day filter is implemented using the same code, with a windowSize = 5. The matrix x in the example contains the raw stock data. For more info on the filt function: http://www.mathworks.com/access/helpdesk/help/techdoc/ref/filter.html

windowSize = 20; % Twenty day average
% The following filter applies the 1/windowSize coefficients
twentyDay = filter(ones(1,windowSize)/windowSize,1,x);

Use of Windowing Techniques on 3G CDMA Signal

2010-01-28T21:21:00.000-08:00

Here is an example of how windowing (blackman in this example) can help reduce sidebands. This technique is useful in making spectral measurements such as Adjacent Channel Leakage Ratio. By minimizing the sidebands generated by an arbitrary waveform generator, for example, the true device performance can be more accurately measured. For more information on windowing, see the following article from National Instruments: http://zone.ni.com/devzone/cda/tut/p/id/4844

Coherent Noise Averaging

2010-01-26T20:38:00.000-08:00

% James Eastham
% Member, IEEE
% Revision: R1
% Contact: james.eastham@ieee.org
%
% Description: This script shows the affect of coherent averaging to
% reduce random noise.

%% Affect of Additive Noise
close all; % close all plots
clear all; % clear all variables
fs = 500; % sample freq
Ts=1/fs; % sample period
n=0:1*fs; % time window (fs=samples per second), 1 second
tn=n*Ts; % sample index
x=cos(2*pi*50*tn); % 50Hz sinusoid
k=.5; %noise amplitude
n=randn(size(x)); % random noise
s=x+(k*n); %signal + random noise with amplitude k
% Time domain plot of x(t)=cos(2*pi*50*tn)
figure('Color',[1 1 1]);
plot(tn,s,'g','LineWidth',2);box off;hold on;
title('Sinusoid w/ Additive White Noise');
xlabel('Time(s)');
ylabel('Amplitude');
% Frequency Domain Plot
figure('Color',[1 1 1]);
[X,f]=ComputeSpectrum(s,fs,2^12);
orig_X=X;
h=plot(f,20*log(X));
box off;
grid on;
set(h,'Linewidth',1);
xlabel('Frequency(Hz)');
ylabel('Amplitude(dB)');
title(['Frequency Spectrum w/ Additive Noise k=.5']);
axis([0 250 -50 120]);

%% Example of Noise Averaging, 3 Coherent Averages
fs = 500; % sample freq
Ts=1/fs; % sample period
n=0:1*fs; % time window (fs=samples per second), 1 second
tn=n*Ts; % sample index
x=cos(2*pi*50*tn);
k=.2; %Noise amplitude
n1=randn(size(x));
n2=randn(size(x));
n3=randn(size(x));
s1=x+(k*n1);
s2=x+(k*n2);
s3=x+(k*n3);
average_s=((s1+s2+s3)/3);
% Time domain plot of x(t)=cos(2*pi*50*tn)
figure('Color',[1 1 1]);
h=plot(tn,average_s,'g','LineWidth',2);box off;hold on;
title('Sinusoid w/ Additive White Noise: 3 Averages');
xlabel('Time(s)');
ylabel('Amplitude');
% Frequency Domain Plot
figure('Color',[1 1 1]);
[X,f]=ComputeSpectrum(average_s,fs,2^12);
plot(f,20*log(X),f,20*log(orig_X),'--r');
box off;
grid on;
set(h,'Linewidth',1);
xlabel('Frequency(Hz)');
ylabel('Amplitude(dB)');
legend('3 Averages','Original');
title(['Frequency Spectrum w/ Noise Averaging: 3 Averages']);
axis([0 250 -50 120]);

%% Example of Noise Averaging, 10 Coherent Averages
fs = 500; % sample freq
Ts=1/fs; % sample period
n=0:1*fs; % time window (fs=samples per second), 1 second
tn=n*Ts; % sample index
x=cos(2*pi*50*tn);
k=.2; %Noise amplitude
n1=randn(size(x));
n2=randn(size(x));
n3=randn(size(x));
n4=randn(size(x));
n5=randn(size(x));
n6=randn(size(x));
n7=randn(size(x));
n8=randn(size(x));
n9=randn(size(x));
n10=randn(size(x));
s1=x+(k*n1);
s2=x+(k*n2);
s3=x+(k*n3);
s4=x+(k*n1);
s5=x+(k*n2);
s6=x+(k*n3);
s7=x+(k*n1);
s8=x+(k*n2);
s9=x+(k*n3);
s10=x+(k*n1);
average_s=((s1+s2+s3+s4+s5+s6+s7+s8+s9+s10)/10);
% Time domain plot of x(t)=cos(2*pi*50*tn)
figure('Color',[1 1 1]);
h=plot(tn,average_s,'g','LineWidth',2);box off;hold on;plot(tn,x,'--r');
title('Sinusoid w/ Additive White Noise: 10 Averages');
xlabel('Time(s)');
ylabel('Amplitude');
legend('composite signal','averaged');
% Frequency Domain Plot
figure('Color',[1 1 1]);
[X,f]=ComputeSpectrum(average_s,fs,2^12);
plot(f,20*log(X),f,20*log(orig_X),'--r');
box off;
grid on;
set(h,'Color',[0 0.8 1]);
set(h,'Linewidth',1);
xlabel('Frequency(Hz)');
ylabel('Amplitude(dB)');
title(['Frequency Spectrum w/ Noise Averaging: 10 Averages']);
legend('10 Averages','Original');
axis([0 250 -50 120]);

DSP Tutorial #5: Frequency Analysis

2010-01-22T20:56:00.000-08:00

This example computes the spectrum of a complex signal using the ComputeSpectrum function.

Couple important things to notice:
- Our range of frequencies on the x-axis is Fs/2
- As Fs increases, the resolution in the frequency domain is much better
- As our FFT bins (2^N) increases, the more bins we have (we can still resolve the freq content provided we meet the sample theorem, we have more bins - basically a finer plot)
- As our windows length increases, the better resolution we have

close all;
clear all;
fs=500;
Ts=1/fs;
n=0:1*fs;
tn=n*Ts;
x=5+cos(2*pi*200*tn)+10+cos(2*pi*100*tn);
% Time Domain Plot
figure('Color',[1 1 1]);
plot(tn,x);box off;
title(['Time Domain Signal']);
[X,f]=ComputeSpectrum(x,fs,2^14);
figure('Color',[1 1 1]);
plot(f,20*log(X));
title(['Freq Domain']);
axis('tight');

The Compute Spectrum Function

2010-01-22T18:22:00.000-08:00

This function uses the FFT algorithm to compute the spectrum of an input with a number of points:

function [X,f]=ComputeSpectrum(x,fs,N);
% ComputeSpectrum Compute the Spectrum
%
% x input signal
% fs Sampling frequency of the input signal
% N number of points to evaluate spectrum (integer number of 2)
%
% X Absolute value of the spectrum (capital X for spectrum)
% f Frequency axis of the spectrum
%
% This function uses the FFT algorithm to compute the spectrum of an
% input signal with a number of points (N) N-point FFT where N must be
% a power of 2 (e.g. 2^10)
%
%Example: Compute the spectrum of sinusoid at 3Hz, fs=36Hz
%
% x=cos(2*pi*3*[0:12*3]*1/(12*3)); 12 times max freq
% [X,f]=ComputeSpectrum(x,12*3,2^12);
%
%Revision:0.0.1 jeastham
%
X=abs(fft(x,N));
f=(fs/N)*[0:N-1]; %make the axis freq
X=X(1:end/2);
f=f(1:end/2);

DSP Tutorial #4: Plotting Complex Signals

2010-01-18T09:38:00.001-08:00

In this example we will plot a more complex time domain signal in Matlab:



clear all;   %clear all variables
fs = 50;     % sample freq
Ts=1/fs;     % sample period
n=0:1*fs;    % time window (fs=samples per second), 1 second
tn=n*Ts;     % sample index
sig_a=cos(2*pi*5*tn);
sig_b=1.5+cos(2*pi*20*tn);
x=sig_a+sig_b;
% Time domain plot of x(t)=cos(2*pi*5*tn)
figure('Color',[1 1 1]);
h=plot(tn,x,'g','LineWidth',3);box off;hold on;
plot(tn,sig_a,'--r','LineWidth',2);
plot(tn,sig_b,'--k','LineWidth',2);
title('Example of Composite Signal: 2 Signals Added');
legend('Composite','Signal 1','Signal 2');
xlabel('Time(s)');
ylabel('Amplitude');
figure('Color',[1 1 1]);
plot(tn,sig_a);
title('Stem Plot Example: Fs = 50 (50 Points)');
axis('tight');



Sum of Sinusoids Triangle Approximation
2010-01-18T09:15:00.000-08:00

Triangle Approximation: Sum of Sinusoids Example

The following Matlab code provides an example of the fact that any signal can constructed using a sum of sinusoids. 



clear all;   %clear all variables
fs = 50;     % sample freq
Ts=1/fs;     % sample period
n=0:1*fs;    % time window (fs=samples per second), 1 second
tn=n*Ts;     % sample index
wave_1=sin(2*pi*5*tn);
wave_2=sin(-1/9*sin(6*pi*5*tn));
wave_3=sin(1/25*sin(10*pi*5*tn));
wave_4=sin(-1/49*sin(14*pi*5*tn));
wave_5=sin(1/81*sin(18*pi*5*tn));
wave_6=sin(-1/121*sin(22*pi*5*tn));
wave_7=sin(1/169*sin(26*pi*5*tn));
x = (8/(pi^2))*(wave_1+wave_2+wave_3+wave_4+wave_5+wave_6+wave_7);
% Time domain plot of x(t)=cos(2*pi*5*tn)
figure('Color',[1 1 1]);
h=plot(tn,wave_1,'--r',tn,wave_2,'--k',tn,x,'g','LineWidth',2);box off;hold on;
xlabel('Time(s)');
ylabel('Amplitude');
title('Square Wave Approximation');
axis([0 1 -1.5 1.5]);


2010-01-14T15:21:00.000-08:00
Plotting a Discrete Time-Domain Signal in Matlab







  In order to properly construct a sinusoid in Matlab we need to do a few things. We need to give Matlab the equation of our sinusoid. We also need to tell Matlab how often to sample the sinusoid and for how long we want to sample our sinusoid. Essentially, we want to tell Matlab “ok, watch this 5Hz sinusoid for 1 second. In that one second, take 50 discrete samples (actually 51), and plot the data. I should see a point every 0.02 seconds, for a total of 51 total points (Matlab starts at zero).” What we want to do may seem pretty simple but this process often confuses students. This overview will help in understanding the proper construction of sinusoids. I say “proper” because we will use the proper DSP terms when we tell Matlab what to do. For example, we will use terms like “sample frequency” and “sample period” to describe our signal.
   
  To establish our signal we will need to:


   
  Set      the sample frequency FS
Set      the sample period Ts
Set      the time window (number of samples)
Set      the sample index Tn (time index)
Establish      the signal x(t) and plot



    1. Set the Sample Frequency (Fs):

This is our sample frequency. This is basically how many “points” we will use to construct each signal. Remember it’s zero based, so it’s really plus one. In the example below, Fs was set to 50, so we have 51 points in our plot, we ‘sample’ the signal 51 times. This is a 5Hz signal, so we are sampling it 10 times per cycle, which satisfies Nyquist-Shannon’s sampling theorem. 
  

Matlab code:

Fs = 50;







  2. Set the Sample Period (Ts):

Next, we will set the sample period, Ts. The sample period is calculated as 1/Fs, or 1/50 in this example = 0.02. This is essentially the ‘granularity’ of our sampling. In other words we will have a discrete point every 0.02 units (which will be time).  If we zoom into our signal and only look at the first 0.2 seconds, we can see that we have 11 datapoints, one data point every 0.02s (remember it’s zero based).
  


Matlab code:

Ts = 1/Fs;
  


  3. Set the Time Window (n):

Next we will establish out how many points are in one second of our signal. We will setup a matrix from zero to one second. In our example Fs = 50, so n = 0:1*Fs. When we set n in Matlab, all we are really doing is setting up a matrix of numbers from 0 to 50. That is: 0,1,2,3,4,5,6,…..,50. We will use this matrix to set the sample index Tn in the next step. 
   
  n is now equal to the matrix below:
   0     1     2     3     4     5     6     7     8     9    10    11    12    13    14    15    16    17    18    19    20    21    22    23    24    25    26    27    28  29    30    31    32    33    34    35    36    37    38    39    40    41    42    43    44    45    46    47    48    49    50
   
  

Matlab code:

n =0:1*Fs;
   
  

4. Establish the Sample Index (Tn):

Now we will tell Matlab how to chop up or signal. We set the Sample Index as Tn = Ts * n. In our example, Tn will equal .02. We want to see a discrete point every 0.02 seconds. We will take n, or matrix from 0 to 500 and multiply it by Ts which was 0.02. Now we have a matrix with 51 elements from 0 to 1 spaced every 0.02 seconds. We are now ready to plot our sinusoid.
   
  Tn is now equal to the matrix below:
  0    0.0200    0.0400    0.0600    0.0800    0.1000    0.1200    0.1400    0.1600    0.1800    0.2000    0.2200    0.2400    0.2600    0.2800    0.3000    0.3200  0.3400    0.3600    0.3800    0.4000    0.4200    0.4400    0.4600    0.4800    0.5000    0.5200    0.5400    0.5600    0.5800    0.6000    0.6200    0.6400    0.6600    0.6800    0.7000    0.7200    0.7400    0.7600    0.7800    0.8000    0.8200    0.8400    0.8600    0.8800    0.9000    0.9200    0.9400    0.9600    0.9800    1.0000
   
  

Matlab code:
  Tn =Ts*n;
   
   
  

5. Plot our 5Hz Sinusoid:

Plotting our sinusoid is pretty easy at this point. Remember, we want a 5Hz cosine signal constructed of 51 points in one second; we want to “watch” the signal for 1 second and plot the results. We can plot the X axis correctly with time also using tn.
  

clear all;   %clear all variables
  fs = 50;     % sample freq
  Ts=1/fs;     % sample period
  n=0:1*fs;    % time window (fs=samples per second), 1 second
  tn=n*Ts;     % sample index
  sig_a=cos(2*pi*5*tn);

figure('Color',[1 1 1]);
  stem(tn,sig_a);
  title('Stem Plot Example: Fs = 50 (50 Points)');
  axis('tight');







DSP Tutorial Matlab Script
2010-01-08T20:40:00.000-08:00


The script below will get you started plotting basic time domain signals. The following tutorials are available on additional DSP topics:
DSP Tutorial #1: Basic Plots
DSP Tutorial #2: Signal Creation
DSP Tutorial #3: Sum of Sinusoids
DSP Tutorial #4: Complex Signals
DSP Tutorial #5: Frequency Analysis
DSP Tutorial #6: Looping
DSP Tutorial #7: Audio Processing
DSP Tutorial #8: Spectrum Function
DSP Tutorial #9: Basic digital filter
DSP Tutorial #10: Windowing





This is a simple Matlab script which describes the basic plotting of time-domain signals.

%% DSP Tutorial Script
% James Eastham
% Member, IEEE
% Created on: 01/08/2010
% Revision: R1

%% Plotting a Basic Time Domain Signals
% Sinewave with a peak amplitude of 1 and a fequency of freq
clear all;
close all;
freq = 5; %5Hz
n=0:.01:.2; %number of discrete values on X axis
x = cos (2*pi*freq*n); %our cosine signal
figure('Color',[1 1 1]);
plot(n,x);
title('Example of Tim-Domain Signal');
xlabel('Time(s)');
ylabel('Amplitude');

%% Plotting a Basic Time Domain Signal Continued
figure('Color',[1 1 1]);
stem(n,x,'--rs','LineWidth',2,'MarkerEdgeColor','k','MarkerFaceColor','g','Markersize',10);
hold on;
plot(n,x);
title('Example of Time-Domain Signal Plot');
xlabel('Time(s)');
ylabel('Amplitude');


%% Plotting with the Subplot Feature
% subplot(m,n,p) breaks the figure window into an
% m-by-n matrix
figure('Color',[1 1 1]);
subplot(2,1,1);
plot(n,x);
title('Example of the Time-Domain Signal');
xlabel('Time(s)');
ylabel('Amplitude');
ns=1:length(n);
subplot(2,1,2);
stem(ns,x,'.r');
title('Example of the Descrete Points');
xlabel('Descrete Sample');
ylabel('Amplitude');
axis([1 21 -1.2 1.2]);
grid on;


Modeling Temperature in LTSpice - Part I
2009-11-19T20:07:00.001-08:00
We can model temperature behavior in LTSpice using the built in step function.

To model a list of temps, insert the spice directive ".step temp list      "
example: .step temp list -25 -10 10 20 25 50 75

To model a sweep, insert the spice directive ".step temp   "
example: .step temp -55 125 25

The following example models a simple transistor amplifier from -55C to 125C in 25C steps.



The transient analysis below shows the source current over temperature.




Modeling Diode Reverse Breakdown Voltages in LTSpice
2009-11-07T14:16:00.000-08:00
The included models in LTSpice do not include reverse breakdown. In order to model the reverse breakdown "vrev" can be specified in the diode model. For this example, create a new schematic and call the diode "mydiode". Insert the .model directive to add the reverse breakdown. Below, I used all the standard parameters for the 1N4148 and added vrev=50. The vrev specifies the reverse breakdown at -50V. Many other diode properties can be defined in the model including the non-linear characteristics of the breakdown.

Forward and reverse breakdown and current limitations can be modeled by using a more sophisticated model. Using the model below, forward and reverse current limiting can be observed.

.model mydiode D(vrev=40 Ibv=20nA BV=40 Ibv=1e-10 Is=1e-7 Rs=10 Tt=2n Ilimit=5mA Vpk=40 Revilimit=5mA)




Adding a LM741 Op Amp Model to LTSpice
2009-10-16T15:13:00.000-07:00
     
  Simulating operational amplifiers in LTSpice using non-ideal characteristics is desirable for students to understand op amp AC and DC limitations. Many student projects require the use of the classic LM741 op amp. The following steps can be followed to get the LM741 model into LTSpice for simulation.
   
  Step 1: Obtain the spice model for the LM741 (see complete model below)
  Step 2: Copy this model into the LTSpice sub directory. The file must be given a *.sub extension and not a *.txt extension. Make sure to save the file as “all file types” with the .sub extension. Save the file as LM741.sub. The directory is typically: C:\Program Files\LTC\LTspiceIV\lib\sub
        Step 3: Start up LTSpice, insert the op amp2 component. Right click on the symbol and change the value to LM741.
   
   
   
   
   
   
   
   
   
   
  































Step 4: Add the spice directive to the schematic using the .op. Add “.lib LM741.sub” to the schematic.


























   
   
   
  





Step 5: You are now ready to run simulations with the LM741.


 

   
  



















Spice model below:


*//////////////////////////////////////////////////////////////////////

* (C) National Semiconductor, Inc.

* Models developed and under copyright by:

* National Semiconductor, Inc.



*/////////////////////////////////////////////////////////////////////

* Legal Notice: This material is intended for free software support.

* The file may be copied, and distributed; however, reselling the

* material is illegal



*////////////////////////////////////////////////////////////////////

* For ordering or technical information on these models, contact:

* National Semiconductor's Customer Response Center

* 7:00 A.M.--7:00 P.M. U.S. Central Time

* (800) 272-9959

* For Applications support, contact the Internet address:

* amps-apps@galaxy.nsc.com



*//////////////////////////////////////////////////////////

*LM741 OPERATIONAL AMPLIFIER MACRO-MODEL

*//////////////////////////////////////////////////////////

*

* connections: non-inverting input

* | inverting input

* | | positive power supply

* | | | negative power supply

* | | | | output

* | | | | |

* | | | | |

.SUBCKT LM741 1 2 99 50 28

*

*Features:

*Improved performance over industry standards

*Plug-in replacement for LM709,LM201,MC1439,748

*Input and output overload protection

*

****************INPUT STAGE**************

*

IOS 2 1 20N

*^Input offset current

R1 1 3 250K

R2 3 2 250K

I1 4 50 100U

R3 5 99 517

R4 6 99 517

Q1 5 2 4 QX

Q2 6 7 4 QX

*Fp2=2.55 MHz

C4 5 6 60.3614P

*

***********COMMON MODE EFFECT***********

*

I2 99 50 1.6MA

*^Quiescent supply current

EOS 7 1 POLY(1) 16 49 1E-3 1

*Input offset voltage.^

R8 99 49 40K

R9 49 50 40K

*

*********OUTPUT VOLTAGE LIMITING********

V2 99 8 1.63

D1 9 8 DX

D2 10 9 DX

V3 10 50 1.63

*

**************SECOND STAGE**************

*

EH 99 98 99 49 1

G1 98 9 5 6 2.1E-3

*Fp1=5 Hz

R5 98 9 95.493MEG

C3 98 9 333.33P

*

***************POLE STAGE***************

*

*Fp=30 MHz

G3 98 15 9 49 1E-6

R12 98 15 1MEG

C5 98 15 5.3052E-15

*

*********COMMON-MODE ZERO STAGE*********

*

*Fpcm=300 Hz

G4 98 16 3 49 3.1623E-8

L2 98 17 530.5M

R13 17 16 1K

*

**************OUTPUT STAGE**************

*

F6 50 99 POLY(1) V6 450U 1

E1 99 23 99 15 1

R16 24 23 25

D5 26 24 DX

V6 26 22 0.65V

R17 23 25 25

D6 25 27 DX

V7 22 27 0.65V

V5 22 21 0.18V

D4 21 15 DX

V4 20 22 0.18V

D3 15 20 DX

L3 22 28 100P

RL3 22 28 100K

*

***************MODELS USED**************

*

.MODEL DX D(IS=1E-15)

.MODEL QX NPN(BF=625)

*

.ENDS

*$[/url] 

   
  


Getting Started with LTSpice
2009-08-29T16:23:00.000-07:00

LTSpice is a great free circuit simulator. The software and introduction videos can be found at: www.cmosedu.com


New Mac Mini
2009-06-14T10:15:00.000-07:00
I finally took the plunge and picked up a Mac. I'm recording a brief diary of pros/cons from my journey: 
Mac Impressions Diary




68HCS12 Heartbeat
2009-04-13T13:59:00.000-07:00
It's often useful to have a flashing LED in embedded systems to verify the system is 'running' and healthy. I coded up a 500ms heartbeat using the Motorolla 68HCS12 in C. See below.

#include "vectors12.h"
#include "ioregs12.h"
#include "bits.h"  // custom header containing bit values in hex

#define INTERRUPT void __attribute__ ((interrupt))
#define sei() __asm__ __volatile__ ( "sei" )
#define cli() __asm__ __volatile__ ( "cli" )
#define wai() __asm__ __volatile__ ( "wai" )

volatile unsigned int hbcount;

INTERRUPT rti_interrupt(void);

int main(void)
{
 // Initialization
 sei(); // diable interrupts
 UserRTI = (unsigned int) &rti_interrupt;
 RTICTL = 0x40; // set RTICTL for 1ms interrupt
 CRGINT = BIT8; // enable RTI interrupts
 DDRT |= BIT8; // or equal
 cli();  // clear interrupt flag

 while (1)
 { // idle loop
   wai();
 };
 return 0;
}

INTERRUPT rti_interrupt(void)
{
    CRGFLG = 0x80; // acknoledge RTI & clear interrupt flag
    hbcount++;
    if (hbcount >= 489) // Reset hbcount every 500ms
    {
       hbcount = 0;
       PTT = PTIT ^ BIT8; // exclusive or PTT
    }
}


PDP-8 Minicomputer
2009-01-13T20:37:00.000-08:00

It's 40 year old technology, but it still pretty cool and complex. I'm working to realize the PDP-8 minicomputer from the early 70s using an FPGA. Stay tuned for progress updates!


Parallel Adder / Multiplier using VHDL
2009-01-02T14:34:00.000-08:00
I designed a state machine controller using VHDL. This state machine implemented a parallel adder and multiplier with overflow capability. This project used the Nexys2 board from Digilent. I used Xilinx ISE for the coding and Modelsim for the simulation. For a full report see "Adder / Multiplier in VHDL" under papers.
  


Current Project: Temperature Controlled Voltage Source
2008-12-14T09:28:00.000-08:00

I designed a temperature controlled voltage source. This design uses a discrete BJT operational amplifier design with a AD222100STZ temperature sensor. For a full report, see "Temp Voltage Source" under the papers section.


Modeling Temperature Behavior in LTSpice
2008-10-18T10:06:00.000-07:00

Modeling temperature behavior in LTSpice.

I'm working on a temperature controlled amplifier using a thermocouple and operational amplifier. The simple circuit is shown above. The operational amplifier simply amplifies the thermocouple input voltage. The thermocouple input voltage is obviously a function of the temperature (measured between the reference junction and the measuring junction). For modeling purposes, the thermocouple is modeled as a voltage supply.

To simulate the changes in voltage as a function of temperature, setup a sweep in LTSpice using the .step command as shown. Next, establish a parameter using the .PARAM command, creating a new variable as a function of the sweep. This variable can then be used in the circuit. In this example, the "thermocouple" supply voltage changes based on the 'THERM' variable. The voltage graph shows output voltage as a function of temperature.




Active Filters
2008-10-04T11:44:00.000-07:00

Filters are used in a wide variety of electronic
applications. Band-pass filters are used in the field of
telecommunications for speech processing. High-frequency
band-pass filters are used for channel selection in many telephone
applications. Digital signal processing systems use low-pass
anti-aliasing filters as well as low-pass noise filters in conditioning
stages. Power supplies use band-rejection filters to suppress line
frequency signals and high frequency components. Additionally,
filters can be specifically designed to provide phase shifts to each
frequency component while not filtering any frequencies. Active
filter schemes use operational amplifiers, or other active
components, in conjunction with resistors and capacitors to
provide filtering without the need for large inductors as required
in passive filtering. This paper compares the difference between
passive and active filters, and presents all the schemes highlighted
(i.e. low-pass, high-pass, band-pass, band-stop). Popular active
filtering techniques are presented including the Butterworth
filter, Chebyshev filter, and the Bessel filter. The low-pass
Butterworth filter is analyzed in detail, simulation and
experimental results are presented. See "Active Filters" under "papers" on the right for the full document.


Direct Sequence Spread Spectrum
2008-10-04T11:31:00.001-07:00


DSSS is a pretty cool technology. This paper provides a pretty high level overview with supporting Matlab simulations. See "DS-SS Tutorial" under "papers" on the right for the full document.


BJT Op Amp Project
2008-03-27T09:44:00.000-07:00















I designed/simulated/built/tested a 3-stage BJT op amp. Results were pretty good. See BJT Op amp under papers section for full report.


Outer Rim
2008-03-02T17:43:00.001-08:00























Outer Rim is a game Brent Edington and I designed. We've play tested it a couple of times and plan to clean up the rule set soon.


Current Project: Napoleon Orders of Battle
2008-02-16T07:53:00.000-08:00























NOoB is a game I'm working on in my "spare" time.  It's a tactical wargame based on Charles Grant's 1974 book - Napoleonic Wargaming. Dan has agreed to help co-develop NOoB which should help this game see the light of day.


Current Project: Railway Builder
2008-02-09T09:17:00.000-08:00















Railway Builder is a card game I designed. It's being evaluated by Playroom Entertainment (publisher). Fingers crossed!


Current Project: Discrete Multistage BJT Amplifier
2008-01-28T23:48:00.000-08:00
Gain (Av) = 100
Zin > 250KΩ
Zo<100Ω
Vcc = 15V
Rl = 1Ω
Vs = 10mV @ 1KHz