Author Topic: What is the difference between Sensitivity and Accuracy?  (Read 7674 times)

mrlegoman

  • Newbie
  • *
  • Posts: 9
  • Karma: +0/-0
What is the difference between Sensitivity and Accuracy?
« on: August 10, 2011, 06:18:21 PM »
We are in a dispute with our electric company over the electrical usage of our new home.  We installed foam insulation, high efficient windows and the latest and greatest efficient heat pump.  Our bills where lower then our old house as expected.  But apparently because our bills where lower then average for our neighborhood, the electric company said there was a problem with our meter and replaced it. Now our bills have doubled!  I need some way to verify the electric companies meter reading.  We do not have reliable internet access at our house, so we do not have a home network setup yet to use the 5000.  I would like to use the TED1000 to verify our meter reading.  I don't understand what they mean by Sensitivity up to 10w, but accuracy is 2%.  Does this mean that if we have an 8w trickle over the whole month, that this meter would not see it? Accuracy of what? Accuracy of total usage?

I guess my main questions is, is the TED 1000 accurate enough to determine and compare our actual electric usage to that of the main meter?

Thanks,
-Don

GAR

  • Full Member
  • ***
  • Posts: 131
  • Karma: +0/-0
Re: What is the difference between Sensitivity and Accuracy?
« Reply #1 on: August 10, 2011, 10:00:55 PM »
110810-1215 EDT

mrlegoman:

My personal opinion is that you will be better off with the 1000 than the 5000.

The 1000 series quantizes the measurement in 10 W increments vs 1 W for the 5000 series. I do not know where you found the 2% accuracy figure, but I may also have seen it somewhere. 2% means nothing unless you have some reference. What is the reference? It was not defined.

If you consider an ordinary analog moving coil with pointer instrument these were usually labeled as 1 or 2% accurate. Because of common knowledge and how these devices were made the implied reference was full scale. So if you had a meter with a 150 V full scale range, then 1% accuracy would mean that at any point between 0 and 150 V the error in the reading should not be more than +/-1.5 V from the reading. At a 75 V reading it means the actual voltage could be anywhere between 73.5 V and 76.5 V.

With digital voltmeters it has become a little trickier in the way things are specified. Partly because of the grossly different means of building these devices compared to moving coil meters. You may find a spec like +/-1% of reading +/-3 counts. One count would correspond the to the smallest increment of reading. A meter that read 2.03 would have a possible error of  +/-0.03 from the count limit plus +/-0.02 from the % of reading error. So the total error would be +/-0.05 .

To do a quick evaluation of your power company meter do the following.
1. Get a Kill-A-Watt EZ (Home Depot at about $30 is one source)
2. Get a suitable known resistive load in the 1500 W range. A small portable heater.
3. Use the Kill-A-Watt EZ to monitor the power and energy to the heater. I have found their accuracy to be fairly good, and it is simple with no complications for this test.
Keep in mind you are looking for a big error in something if your bill doubled.
4. Turn off all circuit breakers to everything in your house except one circuit to which you connect the test heater, and nothing else is on this circuit.
5. With all breakers off and before connecting the heater the KWH meter on your house should not change at all. If it is a rotating disk type the disk should be perfectly still. If change is occurring, then there is a load somewhere. Whatever that might be it must be turned off.
6. Connect the heater and set it at full power so any thermostat stays closed.
7. With the heater on, then the house meter should be changing, and the Kill-A-Watt should show maybe 1300 to 1500 watts of load.
8. Simultaneously reset the Kill-A-Watt EZ to clear the KWH and Time registers, and read the House Meter. To reset the Kill-A-Watt hold the RESET button down until the display changes to "rES". Do these two functions within a few seconds.
9. Allow the test to run for about 1 hour.
10. At the end of the time period read the house meter and the KWHs on the Kill-A-Watt. Should be in the range of 1.3 to 1.5 .
11. The change in reading of the power company meter is what you compare with the Kill-A-Watt reading.

My meter is of the spinning disk type and has the lowest reading dial marked in KWH from 0 to 9. So for small changes in energy you would interpolate between marks to make your readings. On my particular meter the disk makes one revolution in 26 seconds at a load of 1.83 kW. The disk is considerably geared down to the dials.

I do not know the quantizing level of power company digital displays. Suppose it was 1 KWH, then you would run the experiment slightly different. Watch for a transition of the of the power company meter reading, read the KWH on the Kill-A-Watt and record, at the next transition again record the Kill-A-Watt KWH. The difference of the two readings should be 1 KWH. If it is in the range of 0.5 KWH, then you would be correct that the power company meter was in error. If the difference is this big there should be no problem showing the power company the error.

If you saw a 10% difference between the the power company meter and the Kill-A-Watt. then may be their meter is close to correct.

In this current range I have found the Kill-A-Watt to be quite accurate. My Kill-A-Watt is on the order of 0.5% of reading accuracy at 6 amperes.

If this short time simple experiment does not show something like a 50% error, then depending upon the error seen you have to make a judgement on the next step. The next step may be a more controlled experiment.

.

SherlockOhms

  • Full Member
  • ***
  • Posts: 168
  • Karma: +0/-0
  • Sherlock
Re: What is the difference between Sensitivity and Accuracy?
« Reply #2 on: August 15, 2011, 08:27:52 PM »
Accuracy and resolution are different things....

resolution of 10watts means that ... lets say you have a base line of 100watts right now...and you turn on a 15watt light bulb... you will show 115 on the KW.  at same time that will be added the cumulative. Now if you turn on an 8watt light bulb the screen will NOT show that extra load...however it's still added to your cumulative. 

Accuracy to 2% means for total monthly cycle your error will not be more then %2 of your utility meter. It can be as much as a daily averge off...because we count from midnight.... electric company can come in any time ot theday or later next day or earlier... 

mrlegoman

  • Newbie
  • *
  • Posts: 9
  • Karma: +0/-0
Re: What is the difference between Sensitivity and Accuracy?
« Reply #3 on: August 17, 2011, 06:07:47 PM »
Thanks for the info guys.  I guess that was my concern was with the unit being sensitive to 10watts, whether or not it was accumulating loads under that mark. Which obviously would effect the accuracy over a monthly period.  But 2% over the month is more then acceptable.

I do have a Kill-o-watt and will try out the heater test this weekend.  I went ahead and bought a TED 1002 too. 

Our electric company has now twice come out and replaced our readers, then come back and said there was an malfunction in their unit.  Get this, they said that while the display showed one number, "the meter had recorded additional usage that it had stored up and then displayed all at one".  But the unit was already replaced.  So there was no way to verify their claim.  So an additional 1400 Kwh was added onto our last bill which normally averages only 1000kwh.  I argued that our energy usage history did not show where an additional 1400 kwh could be probable.  They called yesterday and said they could compromise by taking 1000 kwh off our bill.  I'd like to use the TED to keep a monthly record of my usage to in the event that this happens again I can present evidence to the SCC and FERC.


GAR

  • Full Member
  • ***
  • Posts: 131
  • Karma: +0/-0
Re: What is the difference between Sensitivity and Accuracy?
« Reply #4 on: August 17, 2011, 10:03:47 PM »
110817-1158 EDT

mrlegoman:

I do not know precisely what is inside the 1000 series MTU, but from the outside it appears that power measurements are quantized at the 10 W level, whereas the 5000 is at the 1 W level. This will be determined by the A to D converter used and possibly on how power is measured.

Assume it is the 10 W level, then as a result of this you might see up to 10 * 24 * 31 / 1000 KWH = about 7.5 KWH error per month. Probably not this bad because of random variations. 7.5 out of 1400 is only 0.5 % error. There are probably analog errors that contribute more error than the quantizing level.

There is no reason you can not use the 5000 series without a network. The Gateway can be connected directly to your computer via an Ethernet cable, straight thru or crossover.

On the power company electronic meter what is the resolution on the display? Does it read to 0.1 KWH or even 0.01 KWH or only integers quantized to 1?

Anytime the power company changes your meter make sure to record the value on the old one before removal, and its serial number. Do the same on the new unit just after installation.

Get the inline filter and install it on a new breaker with no other load on the new circuit except the MTU and the RDU or Gateway. Bring this new circuit to an outlet next to the panel and an extension cord from there to the RDU. Or create a new circuit to an outlet near where the RDU will be located. This will provide the cleanest signal between the MTU and RDU and the fewest dropout errors (loss of communication).

For the computer you need one with battery backup. With the 1000 series the Footprints program is located on the collection computer. Thus, loss of power causes the program to stop, but data up to this time is resident on the hard drive. When power returns, if this is after backup power to the computer is lost, then the Footprints program will not restart. Thus, new data won't be recorded. If this is a possibility, then you need to determine what should be done. There are possible solutions.

A possible source of error is when the MTU looses power and the computer is still running. Does the Gateway program continue to process the accumulation of energy with the last KW reading? I have not run this kind of test.

You can export the data from the Footprints data base to another file for safe keeping. You can also compare your TED data with the power company meter daily, and within a few days determine if there is a noticeable difference.

Do check your TED data with a known load. I just did with a 100 W bulb. With Kill-A-Watt I read 123.1 V and 103.6 W. On the TED 1000 base load was 1.500 KW after about 4 seconds the RDU read 1.600 KW. It took 2 seconds to get to about 1.6, then overshot a little, and stabilized at 1.600 KW. Starting at 13:23:20 the exported results at each second were:
1.50 KW
1.51
1.51
1.50
1.57
1.61
1.60
1.60
1.61
1.59
1.50
Notice this data is quantized to 10 W. The bulb was probably actually turned on at between :22 and :23 . A complete transition seems to take 2 seconds. And there is probably a delay from the action of turn on or off to data being received at the RDU.

.