Power measurement question. Kill-A-Watt and Watts Up? Pro ?

Hello all,
I have both a Kill-A-Watt and a Watts Up? Pro power meters. To check how close they're against each other, I connected them in series with the
Kill-A-Watt going into the plug, then the Watts Up? Pro and the load.
They should read about the same or the Kill-A-Watt should read the power consumption of the load plus the Watts Up?(no more than a watt or two). Comparing the actual meaasured values, the Watts Up? Pro consistently give a value about 10% higher than the Kill-A-Watt with inductive and non-sinusoidal loads and not quite as much difference with resistive loads. With the computer I'm using to write this message connected as the load, Kill-A-Watt is reading 174W and Watts Up? Pro is reading 190W.
Both devices agrees within a reasonable degree against a known good DMM for voltage. DMM: 120.3V W: 121V (does not resolve to 100mV) K: 120.4V
Current do not agree with eachother: DMM: Unable to measure, my DMM is not true RMS capable W: 2.37A K: 2.19A (w/ no load, device reads 0.02A, 0.0W)
PF: both devices reads 0.66
Here are the differences in construction:
Voltage measurement:
Watts Up? Pro: An isolating transformer drops the voltage used for both measuring the voltage and powering the device.
Kill-A-Watt: It is directly powered from the AC line through a series R- C circuit and a separate resistive divider is used for voltage measurement.
Current measurement:
Watts Up?: Current transformer.
Kill-A-Watt: Shunt
Which setup is likely to have an inherently better accuracy?
Product information:
Kill-A-Watt: http://www.p3international.com/products/special/P4400/P4400-CE.html
Watts Up? Pro https://www.doubleed.com/products.html
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Good question(s).
I definitely will pay attention for the answer because I was thinking of getting such a product. I went to Radio Shack but the local store never heard of it.
When this is resolved, I suppose I will go "on line" for my purchase.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
A shunt style would give better accuracy, but the CT may as well under resistive loads. The CT may be adding inductance or filtering out the AC line. It depends on what the designer did. And how they are calculating current. The effects of the CT can be compensated for.
I would go with the Shunt style.
Cheers

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

part of your answer can be found in the specs given for the Watts UP?
* 120 VAC, 60 Hz, 15 amps continuous
* True power, RMS volts, RMS current measured and displayed
* +/- 3%, +/- 2 counts of the displayed value for loads above 10 watts
* +/- 5%, +/- 3 counts of the displayed value for loads below 10 watts
Accuracy is of the displayed value, not the range. Some devices claim a smaller number for accuracy but it refers to the range. For instance, a specification of 0.2% of the range sounds good, but it is actually 3.6% of the display (.02 * 1800 = 3.6), which is a worse accuracy
consider that if one unit reads 3% high and the other reads 3.6% low the difference will be 6.6% and yet both units will be within spec.
as with any meter accuracy will depend on several factors including: calibration, stability within an environmental range, methodology of use.
i own one of these units but would not consider the reading to be sufficiently reliable to use if something serious (like my job) was at stake. they are fine for go/no-go applications like: is my AC circuit near max? is my UPS overloaded?
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
snipped-for-privacy@noaspamadelphia.net says...

<snip>
Yes and no. I consider it marketing type presentation for the naive type. Realistically speaking, high frequency load (say something that draws power in 25 15A spikes each half cycle, such as some copy machine/laser printer heater controller) rich in harmonics and high in crest factor would not give the meter same accuracy as measuring a plain resistive load.
Examples of highly harmonic loads:
Almost ALL IT equipments Most home electronics Residential electronic ballasts and CFLs
These loads frequently have a THD greater than 60%.

Which doesn't include current crest factor or frequency bandwidth or how it will behave with other peculiar current waveforms.

And Watts Up? claim maybe based on resistive load.

Watts Up? reads 10% higher than Kill-A-Watt.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
I purchase three kill-a-watt units last year and put them all into each other for comparisons, They all matched each other as close as the LSDigit would allow.
I also took one of the units and checked it against a lab standard traceable to the NBS standards and compared it for Voltage, Current, Power and Reactive Power and I can tell you this, you could use this unit interchangeably with our lab standard. No digit showing on the kill-a-watt unit to it finest resolution was out by even one count. Now our lab standard has a few more digits.
The all Vars (reactive power) and no real power (watts) comparison may be off a little on our lab standard and I did not compensate using known documented accuracy tables. The accuracy formulae is always divided by the PF which makes ???? accuracy but this lab standard is about as accurate as it gets in Canada without controlled environments etc.. etc..
I am really impressed with the Kill-a-Watt units. No tests on waveform distortion or harmonics were performed to date by me.
Waveform distortion form factor may be where the differences are found. OTOH the Watts Up may just not be calibrated properly or junk.
in message

meters. To check how

series with the

Pro and the load.

should read the power

than a watt or two).

Pro consistently

with inductive and

with resistive

connected as the load,

reading 190W.

against a known good DMM

capable
voltage used for both

through a series R-

for voltage

accuracy?
http://www.p3international.com/products/special/P4400/P4400-CE.html
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

It "might be fun" to see how much of a difference putting a reasonable sized motor run capacitor across the load would make. One would expect it to decrease the kVA a bit and have a small effect on the kW.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
says...

It would only affect inductive loads. It would not affect non- sinusoidal distortion.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

it
Not so fast there!
It would tend to "soak up" the higher harmonics. An inductance (or even a small resistor) before the cap would convert some of the current peaks to voltage peaks and thus permit the cap to supply more of the harmonic currents.
Determining the power consumed by a VERY non-linear load isn't an easy problem. Putting an efficient low pass filter between the load and the power meter will reduce the complexity and increase the accuracy.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
John Gilmer wrote:

The question then becomes who benefits from accuracy -- the homeowner or the utility?
--
The e-mail address in our reply-to line is reversed in an attempt to
minimize spam. Our true address is of the form snipped-for-privacy@prodigy.net.
  Click to see the full signature.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

The KAW is a fine unit, but when I plug it into a Vector Maxx mod. square wave inverter and try to find the Hz, it just reads dC. Everything has limitations.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Perhaps your Maxx is making DC (?) A lot of appliances would work fine with 0 V, peak/2, peak, peak/2, 0, and so on.
Nick
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

But how does each one compare to the mechanical wattmeter that is likely spinning around outside your house?
In other words, if you are measing a complex load with complex impedance and harmonics to let's say, determine the precise amount of power consumption per hour. One of the electronic units may give you a figure of precise accuracy. but if that measurement doesn't match the recording characteristics of the simple meter that is used to actually used to bill you per kWh, then it doesn't make much sense unless you are doing some sort of precision lab experiment.
Beachcomber
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
snipped-for-privacy@ece.villanova.edu says...

Well, I contacted the manufacturer of Watts Up? Pro regarding the influence of high crest factor non sinusoidal loads on its accuracy. I will keep you guys posted with what they have to say.
Crest factor and bandwidth limits are both important specifications for a power meter and it is usually included in a higher grade power meter such as this: http://www.valhallascientific.com/data-sheets/pdfs/2100_DS.pdf
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Harry Chickpea wrote:

It works ok on a AC Delco 2500 watt MSW inverter.
--
Steve Spence
Dir., Green Trust, http://www.green-trust.org
  Click to see the full signature.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
SolarFlare wrote:

I remember when you did that. I've had no reason to doubt the accuracy of mine either except for the no-load - or very light load - situation.
OP reported:

Mine, for some reason, shows 0.04A and 0W with no load. More surprisingly, it showed the same thing plugged into a single strand of LED Xmas lights (which were rated at 4W).
--
derek

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
...
: OP reported: : >> K: 2.19A (w/ no load, device reads 0.02A, 0.0W) : : Mine, for some reason, shows 0.04A and 0W with no load. More surprisingly, : it showed the same thing plugged into a single strand of LED Xmas lights : (which were rated at 4W). : -- : derek
That's odd: Mine shows no current with no load. By "no load", do you mean nothing plugged into it? Or whatever's plugged in is just turned off? I can't see a 4W load directly at first, but ... if I leave it plugged in for a few hours, I'll get a fairly accurate watt-hour number. The PF goes wild though, and shows about 40%; don't think it's quite THAT low <g>! The VA was high, too, but ... guess nothing's perfect and sometimes confusing (as in, how VA can be off, and kWH still be close to calcs. ANYway, it's 400 & 4000 watts I'm concerened with!
Pop
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Even lab quality equipment can show PF as wild if the load isn't high enough to get accurate information.
I have a $6K wattmeter/phase angle meter that cannot show phase angles at light load also. It's an old box and technology/techniques have changed also.

load. More

strand of LED

"no load",

plugged in is

if I leave it

accurate watt-hour

40%; don't

but ...

in, how VA

it's 400 &

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Pop wrote:

I doubt it's that odd. If yours and mine are at opposite ends of the possible range shown by kill-a-watts, then OP's is right in the middle :-)

Nothing plugged in. My initial thought was that there was actually a phantom load in my new washer, so I unplugged it - and found it was still reading 0.04A.

Same here.

Being off-grid with a small energy budget, it's all the 5-10W loads that really bug me. I never pay attention to the A or VA values, anyway. It's the watts that I worry about.
--
derek

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

An update on Kill-A-Watt,
I ripped it apart and started probing around. The shunt's output is rather low.
The shunt is placed across the neutral and looks like a 12 gauge wire looped into a U-shape, but I'm not sure what its made of. It gives a 47mV voltage drop with a 12A 1.5kW space heater connected, which tells me the shunt is 3.917 miliohms. The signal from shunt is routed on the board for 3" or so to an LM2902N op-amp. With around 560mW of dissipation, the shunt gets hot to touch and I'm not sure how much the heating affects the resistance of the shunt.
The current resolution on the Kill-A-Watt is 0.01A and this translates to current signal input resolution of 39V, which might make the device suspecticle to noise considering the signal path is not shielded at all. With a one kilowatt resistive load, it jumps around few tens of watts.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here. All logos and trade names are the property of their respective owners.