ARLISS Team NH

1246759

Comments

  • sylvie369sylvie369 Posts: 1,618
    edited 2010-05-27 - 13:30:05
    I just hooked up a CO sensor with just power and ground (from a Homework board), and brought the heater line low (grounded it) for about a minute, with no smoke or anything of the sort. I've played with the sensor before - maybe if I get a chance a little later I'll hook it up the rest of the way and run that program. I don't have any idea why yours smoked. Can you double-check that the wires are attached correctly?
  • Mark in NHMark in NH Posts: 447
    edited 2010-05-27 - 16:10:12
    Sylvie,

    ·· I agree that we should recheck the wiring.

    1) Is this how you understand the pins on the CO sensor module (pluged into the breadboad)are supposed to be wired:

    ALM to P1

    HSW to P0

    +5V to (any available?) Vdd hole

    GND to (any available?) Vss hole

    2) Does·a sensor·pin have to be plugged into the same breadboard "row" that·its corrresponding·wire is plugged into?

    3) According to the sample program, after the CO sensor 'PURGES' for 60 seconds and it starts measuring CO,

    what's it supposed to do when it detects CO above a certain threshold? Show "ALARM" on the GUI? What if CO

    concentration doesn't reach that threshold?

    4) Where do we find what the pre-set CO threshhold for "sounding" the alarm is?

    Thanks!

    Lunch time is over!

    Mark
  • Mark in NHMark in NH Posts: 447
    edited 2010-05-27 - 16:27:31
    Even when no sensor is plugged into the breadboard, after it finishes PURGES it senses CO and the alarm goes off (alarm is text on screen).
    When wired (hopefully) correctly the alarm doesn't go off.

    Dylan and Mike
  • Mark in NHMark in NH Posts: 447
    edited 2010-05-27 - 16:39:05
    HEY TEAM! I think I found a guy who had the same problem we had, check it out!

    http://forums.parallax.com/forums/default.aspx?f=36&m=436101

    Read up on all of the comments, I think·this could solve our problem of wiring/programming the CO Sensor. This was a guy who seems to have had the same problem we are, and for those of you who don't know what our problem was, we believe the program/wiring on the CO Sensor was incorrect, and it is having some problems. Check it out team, thanks.

    -Mike

    P.S. Chris Savage helped this guy out, maybe (hopefully)·it can help us too.

    Post Edited (Mark in NH) : 5/27/2010 4:46:05 PM GMT
  • Mark in NHMark in NH Posts: 447
    edited 2010-05-27 - 18:40:15
    ROCKETEERS,

    Here's a link to a Parallax forum that helps explain how the CO sensor works and how to calibrate the sensor.
    Please print it out, read it, and put a copy of the documentation in your team binder.

    Dylan, this may help answer some of the questions you are going to pose to Dr. Allen, and it may help you pose more thoughtful
    questions when you ask them. In any event, read the link and the sensor documentation, then ask the questions. I'm eager to see
    the answers. Keep up the good work team!

    http://forums.parallax.com/forums/default.aspx?f=36&m=448929

    Aim high,

    Mr. Kibler
  • Tracy AllenTracy Allen Posts: 6,516
    edited 2010-05-27 - 19:08:23
    Right, "mV" is millivolts. The BASIC Stamp runs on 5 Volts, which is the same as 5000 mV. You are I think powering your atmospheric sampler from a 9 Volt battery, which is 9000 mV, but that voltage drops from about 9.3 volts when the battery is brand new, down, down, down as you use it until it is practically exhausted at 6.5 Volts. The BOE has a voltage regulator that maintains the voltage to the Stamp, USB datalogger, and sensors regulated at a constant 5.0 volts as the battery drops. The battery drops faster the more current your project needs to operate. We will need to think about that when you add a CO2 sensor, because it has a heater that requires quite a bit of current.

    @Justin, you pulled the figure 1.22 mV out of the data sheet for the LT1298 analog-to-digital converter, which can operate from the Stamp power supply. Do you understand the purpose of an analog-to-digital converter? Can you explain where the 1.22 mV comes from?

    @rocketeers, you are correct with your definitions of "interpolate" and "extrapolate". I'll expand on why I asked in another email. For now, suffice it to say that extrapolation is considered more "dangerous". Would anyone hazard a guess as to why extrapolation is more dangerous than interpolation?

    The CO2 sensor would not be a good methane detector because the output does not change as a function of the concentration of methane. No response--No sensor! Those other gases in the box on the CO2 graph are what are called "interferences". Gas sensors often respond to more than one gas, and it is important to know what the interferences are. Your answers were very right, if you want to measure methane or CO, you should buy a sensor specifically for those gases. But always be aware of possible interferences.

    I see you are struggling to make your CO sensor work (smoke is not good!). Know that operation of the CO2 sensor is different. When it is delivered, I suggest that we take a close look at the circuit and how it works before you hook it up (no smoke!).

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Tracy Allen
    www.emesystems.com
  • Justin AbbottJustin Abbott Posts: 54
    edited 2010-05-27 - 20:38:00
    Dr. Allen

    Is the purpose of the digital-to-analog convertor to measure tiny changes in the input voltage? And does the 1.22 mV represent the amount of charge it can measure at a time?

    Thank you for your time,
    Justin
  • Andrew (ARLISS)Andrew (ARLISS) Posts: 213
    edited 2010-05-27 - 21:04:30
    Justin Abbott said...
    Dr. Allen

    Is the purpose of the digital-to-analog convertor to measure tiny changes in the input voltage? And does the 1.22 mV represent the amount of charge it can measure at a time?

    Thank you for your time,
    Justin

    As I understand it, simply, the purpose of a digital-to-analog converter is to take digital (binary) data and convert it into another form, readable/understandable by humans. For example, a CD player takes a digital music CD, takes the binary data on the disc and converts that to the audible sounds that we can hear and understand.
  • sylvie369sylvie369 Posts: 1,618
    edited 2010-05-27 - 23:11:07
    Rocketeers - "Digital to analog" is not the same as "analog to digital", and the two kinds of devices serve very different purposes.
  • Andrew (ARLISS)Andrew (ARLISS) Posts: 213
    edited 2010-05-27 - 23:17:47
    sylvie369 said...
    Rocketeers - "Digital to analog" is not the same as "analog to digital", and the two kinds of devices serve very different purposes.

    Oops, just noticed the original post said analog to digital, not digital to analog. I'll let Justin fully answer this question, since it was directed at him, but as I understand it, an analog-to-digital converter would be one that takes a continuous analog signal and converts it into a format readable by computers etc... In this case, the CO sensor and the BOE/BS2 Stamp.
  • Dylan LandryDylan Landry Posts: 235
    edited 2010-05-27 - 23:23:23
    Dr. Allen,

    We have just gotten our feet wet in programming. The team and I would like to know, how would we get the programming able to change the CO sensor's outputting Millivolts be changed in the stamp to PPM on the GUI? Actually thinking back to the graph, wouldn't you have to write a formula in the programming to pretty much take the outputting Millivolts, and convert them to PPM, a value able to be read by a human, just like the graph did? Is that why you asked us those questions? So we could interpret the Millivolts into PPM, and understand everything on the way?........Wow, so you knew that we would be at this exact point, about 4 days ago?!?

    Well this is, revolutionary... Well now I think we all know where we are going with this [noparse]:)[/noparse]. So yeah back to my question... How would we actually be able to write an algorithm to change the Millivolts into PPM?

    Dylan Landry
  • Justin AbbottJustin Abbott Posts: 54
    edited 2010-05-28 - 01:09:14
    Andrew,

    Thank you for that example it made it much easier to understand what a digital-to- analog convertor is.
  • Justin AbbottJustin Abbott Posts: 54
    edited 2010-05-28 - 01:11:12
    Oops i said digital- to- analog too my mistake.
  • Justin AbbottJustin Abbott Posts: 54
    edited 2010-05-28 - 01:16:47
    From what i understand now an analog-to digital convertor converts continous signals to a discret-time signal (which is a time series with a sequence of quantites). Where as a digital-to- analog convertor performs the reverse operation.
  • Mark in NHMark in NH Posts: 447
    edited 2010-05-28 - 01:29:08
    ROCKETEERS,

    ·· You're asking some really good questions·and doing an excellent job of answering·Dr.·Allen's "homework" questions

    thoroughly, accurately, and promptly. Please continue by answering this evening's questions as thoroughly. Remember

    that accuracy is more important than speed. With that said I would still like to see·your answers·tomorrow morning.

    Since our team members are·from different schools we don't the luxury of meeting together as a group/ team as often

    as we might like. We also don't have the luxury of "extra" time.·Consequently·you'll·have to budget your·time·and make

    the most of the limited time·you do have. As you see, and as I said at our first team/parent meeting, much of our work

    will be done here on the forum, like real rocket scientists.

    Let's·heed Dr. Allen's advice and NOT connect the CO2 sensor to the BOE until understand completely how it's

    supposed to be wired and programmed (since we also don't have the luxury of a budget and extra parts...!)

    Keep up the good work,

    Mr. Kibler
  • Dylan LandryDylan Landry Posts: 235
    edited 2010-05-28 - 02:34:07
    Dr. Allen's homework assignment,

    The reason why *extrapolate* is more dangerous to use then interpolate....

    If you extrapolate, you are adding data that you knew in the first place ( In most simplistic way)....
    If you interpolate, you are adding data generally, most likely data you did not know....

    * After 20 minutes of thinking*

    If you are on the forum and someone says,"Blah, blah, blah.. I have a problem..Blah, blah, blah... I am trying to Interpolate data." You interpret that as in he can't find the data to add it to whatever he needs to add it to.

    If you are on the forum and someone says,"Blah, blah, blah.. I have a problem..Blah, blah, blah... I am trying to extrapolate data." You interpret that as in he is having problems not in finding the data, but formating data he knows already into the graph, chart, table etc... People could reply saying, " What are you inserting your data into? Word, excel, power point? What are you inserting, html, java, gif., .jpg, .jpeg? Bottom line, you could make people interpret something you said the wrong way, taking up valuable time and effort. Also maybe attracting the wrong crowd of professors. You could attract advertising majors (I'm guessing they are experienced in having problems inserting images) instead of maybe a biology major ( Forwhat ever you are trying to *interpolate*)...

    Dylan Landry
  • Tracy AllenTracy Allen Posts: 6,516
    edited 2010-05-28 - 06:04:58
    Following up on the analog to digital conversion question. You all had good answers. Justin said, "Is the purpose of the digital-to-analog convertor to measure tiny changes in the input voltage? And does the 1.22 mV represent the amount of charge it can measure at a time?". That is correct. But why 1.22 mV? The LT1298 is an analog to digital converter that has 12 bits of resolution. 12 bits means 2 to the 12th power = 2*2*2*2*2*2*2*2*2*2*2*2 = ????????. Then take that number and divide it into 5000 (which is the power supply for the BASIC Stamp). What do you get?

    Dylan asked, "The team and I would like to know, how would we get the programming able to change the CO sensor's outputting Millivolts be changed in the stamp to PPM on the GUI? Actually thinking back to the graph, wouldn't you have to write a formula in the programming to pretty much take the outputting Millivolts, and convert them to PPM, a value able to be read by a human, just like the graph did? Is that why you asked us those questions?"
    Yes!
    "Well this is, revolutionary... Well now I think we all know where we are going with this [noparse]:)[/noparse]. So yeah back to my question... How would we actually be able to write an algorithm to change the Millivolts into PPM?"
    Good question. How would you approach it?

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Tracy Allen
    www.emesystems.com
  • Tracy AllenTracy Allen Posts: 6,516
    edited 2010-05-28 - 07:21:43
    interpolate and extrapolate do have a colloquial meaning that is similar to their scientific meaning. Extrapolate especially, is taken to mean that you have taken a line of reasoning that goes way beyond the available facts. You might be right (or not).

    On Wednesday morning, Mike posted a very nice piece of work, taking the original graph from the data sheet...
    attachment.php?attachmentid=70582
    blowing it up and extending it to answer the questions about the EMF(mV) that the sensor would produce at about 4% (40000 ppm) CO2 as in exhaled air. Here I have taken that graph and added notes.
    attachment.php?attachmentid=70689
    -- blue box around the original graph from the sensor data sheet.
    -- blue arrows to a couple of data points that the manufacturer must have taken with known concentrations of CO2 , to find what EMF(mV) the sensor would produce.
    -- a light blue area to the straight line or smooth curve that the manufacture drew between the data points. That is the manufacturer's best educated guess of what the sensor does in between the measured points. That is interpolation.
    -- a funny purple curve between two data points. Not a straight line. That could happen, but probably not. To rule it out, you have to make an educated guess, knowing something about the kind of chemical reaction that goes on inside the sensor.
    -- a green line that extends the response down past the edge of the original chart and down past the last experimental data point. This is extrapolation.. It is a good guess. But maybe it goes like the red extrapolation instead. What do you think? That would be a common thing to happen in a chemical reaction. It reaches a point called "saturation". That would be my educated guess, but I don't know where exactly it would happen. Data. We need more data!
    -- At the upper left end of the chart, there is another green line, sort of continuing the known line up and to the left. But maybe the line is really more like the red curve. This is another educated guess, that the sensor has reached the limit of sensitivity and it curves over sharply below 400 ppm.

    Atmospheric CO2 and the exhaled air fall at the extreme left and right ends of the graph, where it is necessary to extrapolate and to make assumptions about the shape of the curve. This is going to call for some experiments when you do receive the sensor from Parallax!

    The number on the bottom axis of this graph are all scrunched up, 100, 1000, 10000. Do any or all of you know what that kind of scale is called?

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Tracy Allen
    www.emesystems.com

    Post Edited (Tracy Allen) : 5/28/2010 7:51:35 AM GMT
    666 x 428 - 50K
  • sylvie369sylvie369 Posts: 1,618
    edited 2010-05-28 - 10:00:20
    Suppose you went around a grade school and asked kids their ages, and how many hours per day they spend on a computer. Imagine that you found that the 5 year olds averaged 2 hours per day, the 6 year olds averaged 4 hours per day, the 7 year olds averaged 6 hours per day, and on up to 12 year olds averaging 16 hours per day (see first image). It would be reasonable for you to guess that if you ran into a 9 year old, he or she would spend 10 hours per day on the computer.

    Is that an interpolation or an extrapolation?

    Now suppose that you ran into an 18 year old? What do your data say about how much time per he or she would spend on the computer? (see second image)

    Is that an interpolation or an extrapolation? Is it a reasonable guess?







    Post Edited (sylvie369) : 5/28/2010 10:06:57 AM GMT
    715 x 316 - 43K
    844 x 355 - 54K
  • Mike NHMike NH Posts: 34
    edited 2010-05-28 - 12:19:50
    sylvie369,

    1.) "It would be reasonable for you to guess that if you ran into a 9 year old, he or she would spend 10 hours per day on the computer.

    Is that an interpolation or an extrapolation?"

    It is an interpolation becuase you had data points before and after 9 year olds. To get the data point for 9 years old, you would follow the data path for the rest of the data points and FILL in the data point for 9 years old.

    2.) Now suppose that you ran into an 18 year old? What do your data say about how much time per he or she would spend on the computer? (see second image)

    An 18 year old would spend 28 hours (per day? there's only 24 hours per day) on the computer.

    3.) Is that an interpolation or an extrapolation? Is it a reasonable guess?

    We think it to be an extrapolation because the graph is being EXTENDED to data for 18 year olds. It is a reasonable guess because the line shows a straight, positive trend.

    Submitted by: Justin and Dylan
  • sylvie369sylvie369 Posts: 1,618
    edited 2010-05-28 - 13:20:17
    You're correct on which is interpolation and which is extrapolation.
    But think a bit more about that second "reasonable guess". I'm 49 years hold, and by extrapolation of our data, I should be spending 90 hours per day on the computer.

    Tracy's graph suggests that sometimes when you go beyond the bounds of the data that you have, some other factors come into·effect, and the straight line you see within your data isn't valid out there. Remember his original question:

    Tracy wrote:
    "For now, suffice it to say that extrapolation is considered more "dangerous". Would anyone hazard a guess as to why extrapolation is more dangerous than interpolation?"

    How would you answer that now?
  • Mark in NHMark in NH Posts: 447
    edited 2010-05-28 - 14:52:57
    ROCKETEERS,

    Let me chime in from western Massachusetts where I'll be this weekend for Christopher's graduation. I just wrote a long post to the forum, pushed the wrong button, and shut the computer down without sending it. So let me be succinct:

    1) If you interpolate data ("inter" means "between") do you already know the actual upper and lower limits of your data?

    2) If the data you're trying to find is located between these upper and lower limits (and especially if the data trendline is straight), is it likely that the value you're trying to figure out will fall somewhere on this data trend line?

    3) If you extrapolate data ("extra" means "outside of") can you be sure which way the data trend line goes (staright? curved?) above or below the minimum and maximum known data values? (HINT: *Refer to the multi-colored lines that Dr. Allen added to the graph that you posted to the forum.)

    4) So then why is it more dangerous to extrapolate data than interpolate data?

    Dylan, you posed an excellent question to Dr. Allen and he gave you an excellent response:

    5) What's next so far as building the ASP-2 and programming the CO2 sensor when the CO2 sensor and BS2e stamp arrive?

    Dylan, you've been experimenting with the BOE, CO sensor, and some basic programming at school. Andrew, you're the Team Captain and you have experience with designing, building and programming. I'd like to see the two of you guide the conversation-- and guide your team-- in answering these questions clearly and logically. Don't be ambiguous with your questions and answers. Be clear and succinct. Be specific. Form a step-by-step plan of action so we can move forward with building and programming right after the holiday. Remember that our time together is limited and most of our work and communication needs to be done here on the forum.

    I'll be checking in throughout the weekend from western Massachusetts to monitor your progress. Please be sure to thank your mentors--Sylvie and Dr. Allen-- for the gift they are giving you: their time, commitment, and experience. I look forward to hearing back from you.

    ANDREW: Please e-mail Sean and Thomas and ask them to be more involved here on the forum. Please 'cc' a copy of the e-mail to me. Thanks.

    Aim high,

    Mr. Kibler
  • Dylan LandryDylan Landry Posts: 235
    edited 2010-05-28 - 17:04:50
    Dear Mr. Kibler, Sylvie and Dr. Allen,

    Thank you very much for your devoted time and effort. I am not just saying this because Mr. Kibler said so, but because·Mike and I are·really thanking you. We really appreciate the help you have given us. Now (not to be rude) but on to the questions Mr. Kibler posted, then on to the question Dr. Allen posted.

    P.S. I am going to my dad's house from today to sunday late afternoon. I will not be able to post untill sunday. Also, if i don't answer all the questions in this post, I will try to this late afternoon if possible. Just giving you a heads up.

    1.) "If you interpolate data ("inter" means "between") do you already know the actual upper and lower limits of your data?"

    Answer: Well, to interpolate, you do need to know the two extremes on both sides of the section of data you are trying to interpret. Do you mean the limits as in if the x-axis was labled hours per day the limit would be 24? Please tell if it does...

    2.)"If the data you're trying to find is located between these upper and lower limits (and especially if the data trendline is straight), is it likely that the value you're trying to figure out will fall somewhere on this data trend line?"

    Answer: Of coarse.. Nothing in the universe is completely random, so everything must follow even a minute trend.. There is not one graph in existance that has a topic with completely randomly ploted points that do not follow some type of positive or negitive trend.

    3.) "If you extrapolate data ("extra" means "outside of") can you be sure which way the data trend line goes (staright? curved?) above or below the minimum and maximum known data values? (HINT: *Refer to the multi-colored lines that Dr. Allen added to the graph that you posted to the forum.)"

    Answer: No. Not completely. If you have a graph of how much water is on earth (not including ice) and use extrapolation (Is that a word?) to predict it in 10 years due to trends that happened a couple years ago, you could be wrong. What could mess it up could be mabey an event that melted or imported water into the earth. That could mess up your data completely.

    4.) "4) So then why is it more dangerous to extrapolate data than interpolate data?"

    Answer: (This will be my last question answered before class ends.) If you interpolate on a graph you have a higher chance of being correct in your predictions. If you extrapolate, it is like predicting into the future. It like shooting a basketball it to the air in a pitch black court, and seeing if it goes in. If that makes any sense...

    I will try to post to the next questions online tonight. I might not be able to for the entire weekend for my schedule is extremely busy..

    Dylan Landry
  • Justin AbbottJustin Abbott Posts: 54
    edited 2010-05-28 - 18:57:01
    Mr. Kibler,

    "If you interpolate data ("inter" means "between") do you already know the actual upper and lower limits of your data?"

    A) Most likely yes because to interpolate data you would need some refrence from thee top and lower limits of your graph. Without them, you may be extrapolating your data.

    "If the data you're trying to find is located between these upper and lower limits (and especially if the data trendline is straight), is it likely that the value you're trying to figure out will fall somewhere on this data trend line?"

    A) Yes because the line would have already been drawn and if it is a straight line, the data point will most likely·have to fall·on the straight line or along it.

    "If you extrapolate data ("extra" means "outside of") can you be sure which way the data trend line goes (staright? curved?) above or below the minimum and maximum known data values? (HINT: *Refer to the multi-colored lines that Dr. Allen added to the graph that you posted to the forum.)"

    A) Not all the time.·As shown in the graph, the line could·continue to go straight, but could also curve. That's why extrapolating the data is dangerous to do, because you could be making a false assumtion.

    "So then why is it more dangerous to extrapolate data than interpolate data?"

    A) As I said before, when you extrapolate data you don't really know how the data points will line up or where they will fall on the graph. This could lead to false assumtions. Interpolating the data is safer because you have good refrence data around the point you are trying to fill in, which makes where that point will fall more accurate.

    ·
  • Mark in NHMark in NH Posts: 447
    edited 2010-05-28 - 20:31:55
    Justin,

    I believe you've hit the proverbial nail on the head!

    ROCKETEERS,

    You seem to have a handle on the basic theory behind how EMF (mV) is converted into numbers (ppm)... analog signal converted into a digital read-out.

    So how then will you use interpolation and extrapolation, and the actual data on the sensitivity graph that Dr. Allen asked you to download, to program the CO2 sensor and the ASP-2? That's a question you should ask Dr. Allen and Sylvie. I have to say I'm not 100% sure (though I have a good idea) but I think their answers will be much clearer, and more accurate. They're the experts; I'm just the chauffeur on this adventure!

    Good work team! Enjoy the weekend and please keep in touch on the forum.

    From Massachusetts,

    Mr. Kibler
  • Mark in NHMark in NH Posts: 447
    edited 2010-05-28 - 20:37:57
    PS ROCKETEERS: Dr. Allen posed some questions to·you in his "multi-colored graph line" post:

    1) "What kind of scale is it when the units go from 10 to 100 to 1000 (and to 10,000, which we added?)"

    Dylan and Justin, you already answered·this when·you made the revised sensitivity·graph in class.

    ANDREW, SEAN AND THOMAS:

    ·· See if you can reply back correctly before Dylan, Justin or Mike...!

    Mr. Kibler
  • Dylan LandryDylan Landry Posts: 235
    edited 2010-05-29 - 01:11:10
    Dear Mr. Kilber,

    Thank you so much for the book! How did you get it to me? I have read so far just chapter 1.. But the book does ask college sized questions for college sized answers, they make for a better feeling (If this is the correct term to describe it) one you figure them out. I do now have basic understanding of what DEBUG means (Basically sends text to the screen) and the END code.. But to fully check my understanding i would need to be using the PBASIC Stamp editor. I do understand you can get the BS2 2.5 editor for free, but to do things like," DEBUG CR,"7*11" I remembered I need the stamp with the micro-controller ( As it describes it, a simple small computer(Which amazingly, all it does is add 1's and 0's)) to do the algorithms. So Dr. Allen, I will be able to answer your question towards me on how we would go about programming it maybe even with the full program tongue.gif!!!
    I will return it to you Monday. The only setback would be that I would not be able to go far into the book for I am in loss of a stamp. I checked radio shack and they barely even had circuit boards and no BOEs or stamps. They are i'm guessing only for the small components that you buy, LEDs, capacitors, comparators, resistors etc...

    Dylan Landry

    P.S. : As I said before I will be at my dad's house this weekend and might not be able to reply to some posts or questions, like the ones before. It is late and I just got back home at 9:00. The rest of the days will be like this one for the rest of the week. But I will make sure to read the book as we are driving around. Thanks again Mr. Kibler! But really, how did you get that to me?!?
  • edited 2010-05-29 - 02:13:59
    Everyone,
    Sorry I have not been replying on the post lately.
    What kind of scale is it when the units go from 10 to 100 to 1000 (and to 10,000, which we added?)"
    Logarithmic scale i believe·

    1) If you interpolate data ("inter" means "between") do you already know the actual upper and lower limits of your data?
    Yes you do know the limits of your data.
    2) If the data you're trying to find is located between these upper and lower limits (and especially if the data trend line is straight), is it likely that the value you're trying to figure out will fall somewhere on this data trend line?
    Yes because is located inside your data. Also like you said if the data trend line is straight then definitely.
    3) If you extrapolate data ("extra" means "outside of") can you be sure which way the data trend line goes (straight? curved?) above or below the minimum and maximum known data values? (HINT: *Refer to the multi-colored lines that Dr. Allen added to the graph that you posted to the forum.)
    When you extrapolate data it is almost like you are guessing, so you can be sure somewhat, but not as sure if you were interpolating the data.
    4) So then why is it more dangerous to extrapolate data than interpolate data?
    Because when you extrapolate data it is sort of like a guess and when you interpolate data you are seeing the data within your data. Almost like your "reading between the lines".
    Sean
    ·
  • sylvie369sylvie369 Posts: 1,618
    edited 2010-05-29 - 02:46:22
    All of that is correct. Can you take it the next level? Why is it "almost a guess" when you extrapolate, but not when you interpolate?

    In his post with the graphs, Tracy wrote

    "a green line that extends the response down past the edge of the original chart and down past the last experimental data point. This is extrapolation.. It is a good guess. But maybe it goes like the red extrapolation instead. What do you think? That would be a common thing to happen in a chemical reaction. It reaches a point called "saturation". "

    Why is that different from interpolating? Why wouldn't you worry about "saturation" when you're trying to interpolate from the data he presented?

    There's a really key idea here, one that most adults are completely unaware of. Call me a geek, but I'm pretty excited at the idea that you guys are going to learn about this as part of this project. I sure would like to see my students make the kind of efforts I'm seeing you guys making here.
  • Tracy AllenTracy Allen Posts: 6,516
    edited 2010-05-29 - 20:54:53
    @Dylan, don't worry about taking the weekend off! Enjoy it and come back refreshed. I think you will enjoy programming the BASIC Stamp over the summer. I'll only be checking in occasionally too.

    @ Sylvie369, Wow! How do you find those extra hours in every day! I wish I knew how to extrapolate like that!


    I want to convey a sense that one should read data sheets with a critical eye. Never assume anything beyond the specifications that the manufacturer has provided, and think critically even about the explicit claims and conditions. It is something you gain with experience, often from having been burnt!

    In the case of this chemical sensor, interpolation is safe, because by nature this kind of electro-chemical reaction has a EMF output that is proportional to the logarithm of the concentration. (Right Sean, "logarithmic scale") Each time the concentration doubles, from 400 to 800, 800 to 1600, 1600 to 3200 and so on, the EMF(mV) output changes by about 15 mV. The x axis is logarithmic scale, the y axis is linear scale. That kind of graph (linear against log) is called, "semi-logarithmic". From the underlying chemistry, we expect that the curve will be smooth and "monotonic" (no funny dips or jumps).

    But experience also says that this kind of catalytic reaction will reach a point where further increases in concentration have little or no effect on the output. Where that happens is not shown in the data sheet, so you are left to test it experimentally yourself or to dig out that information somewhere. Similarly, the data sheet does not show what happens below 400ppm of CO2, to answer the question, "what is the limit of detectability?". And it does not say anything about variability from one sensor to another or what the effects of temperature or atmospheric pressure (altitude!) might be. When you buy an expensive sensor ($$$$) you expect all of those things to be already "well characterized". But with a cheap ($) sensor, you are sort of on your own, and that is okay, because you are not here just to plug and play, you are here to learn.

    @Mike. All that being said, I liked your philosophical answer:
    .... is it likely that the value you're trying to figure out will fall somewhere on this data trend line?"
    Answer: Of coarse.. Nothing in the universe is completely random, so everything must follow even a minute trend.. There is not one graph in existance that has a topic with completely randomly ploted points that do not follow some type of positive or negitive trend.
    Sometimes your assumptions surprise you, and sometimes nature surprises you. Simple systems can exhibit chaotic behavior that can hardly be distinguished from completely random. When you look closely enough, what appears to be smooth in space or time, turn out to be a fractal on the edges.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Tracy Allen
    www.emesystems.com

    Post Edited (Tracy Allen) : 5/29/2010 9:00:32 PM GMT
Sign In or Register to comment.