Power Meter with accuracy above 700W

[SKYMTL]

So the time has come to upgrade my test system to something that can can stress 700W+ PSUs.

The problem is that I am a bit worried about the accuracy of my UPM power meter above 500W after reading Jonny's post about how the Kill-A-Watt became inaccurate at higher loads.

I am looking for something that is rated a bit higher and is a bit more of a "Professional" tool. So far I have found a few options:

Brultech2020 (http://www.powermeterstore.com/index.php?cPath=112&products_id=4124)

WattsUp Pro with PF logger (http://www.powermeterstore.com/index.php?cPath=112&products_id=1206)

Fluke Clamp meter (http://www.powermeterstore.com/index.php?cPath=&products_id=3542&p_tab=accessories) which I could use with my Fluke 177. The only problem with this approach is (if I read right) the Fluke 177 can only read up to 10A accurately if I am reading this correctly:
Fluke 187 Specs (http://us.fluke.com/usen/products/specifications.htm?cs_id=26864(FlukeProducts)&category=HMA(FlukeProducts))

Or a true clamp meter (http://www.extech.com/instrument/products/310_399/380926.html)

So, what do you guys think? What should I do?

[jonnyGURU]

The clamp meter is a good idea. I had the Fluke model and now I have a UniTrend that I'm very happy with.

Extech is a good brand. It's what all of the AMD techs use. :) I'd suggest poking around the Extech site a little more too. They actually have an industrial version of what a Kill-A-Watt does on there.

[SKYMTL]

Yeah, I actually found a pretty good price on a kit here:

Test Kit (http://www.professionalequipment.com/extech-electrical-test-kit-w-ir-thermometer-case-tk430-ir/extech-electrical-testers/)

It has a multimeter, clamp-meter (up to 400A) and the IR theremometer. How does that look?

As for the power meter, do you mean somthing like this:
http://www.extech.com/instrument/products/310_399/380801.html

For some reason, I can't find it for sale on that same site.....

[Spectre]

I considered the Watts-Up as I have recently upgraded as well (aka F***ing POS power supply killed 2 Kill-A-Watts). I never could find out if it was more robust than the Kill-A-Watt. I ended up going with a Brand Electronics Power Meter with an Extech slip on current transducer. It seems to be working very well, as I happened to have a test report for a couple units from the factory and my results were a bit lower but well within reason of what the chroma spit out.

[SKYMTL]

I considered the Watts-Up as I have recently upgraded as well (aka F***ing POS power supply killed 2 Kill-A-Watts). I never could find out if it was more robust than the Kill-A-Watt. I ended up going with a Brand Electronics Power Meter with an Extech slip on current transducer. It seems to be working very well, as I happened to have a test report for a couple units from the factory and my results were a bit lower but well within reason of what the chroma spit out.

What I like about the WattsUp is that it has a Power Factor reading too. If I go for the Extech Power Analyzer it will put me back close to $600.

How much did the Brand Electronics and the Extech transducer put you back?

[Spectre]

What I like about the WattsUp is that it has a Power Factor reading too. If I go for the Extech Power Analyzer it will put me back close to $600.

How much did the Brand Electronics and the Extech transducer put you back?

I got the unit refurbed after some strings were pulled but the exact model I have NEW will run $349 for the Model 21-1850/CI (uses serial port for logging), the Model 21-1850/USB (USB port for logging) with 32mb is $400.00, and the current transducer is $75. They cap at 1850VA but so far I have put 1500VA through it accurately, and it does log PF, peak wattage, rmsI, VA, and cost!

They also do custome jobs of up to 4000VA AFAIK, and they make some units that will read multiple sources at once.

[SKYMTL]

Yikes....I don't think I will need to put that much current through the thing.

What was the peak your Kill-A-Watt accurately tested up to?

[Spectre]

Yikes....I don't think I will need to put that much current through the thing.

What was the peak your Kill-A-Watt accurately tested up to?

Well it up and died at anything over 1500VA (actually gave me over unity once...then stopped working). From what it looks like here it starts getting less accurate (more than the billed 2%) after 1200VA depending on the unit.....which I have no idea why that was. With a lot of 1kw units coming through (and getting tired of returning dead Kill-A-Watts) I am going to be over 1200VA a lot so I started hunting for something new.

[SKYMTL]

So even if I was able to draw 800W from the wall, a Kill A Watt should still be more or less OK. Correct?

[Spectre]

So even if I was able to draw 800W from the wall, a Kill A Watt should still be more or less OK. Correct?

Yeah it should be and mine were within the advertised 2% range. It was above that range of 900-1200VA that it starts getting wonky at times.

[SKYMTL]

Someone with a load tester should do a roundup of sub- $150 power meters. There are quite a few:

Seasonic Power Angel
Watts UP
UPM
Kill a Watt

I wonder which is the most accurate.

[Spectre]

Someone with a load tester should do a roundup of sub- $150 power meters. There are quite a few:

Seasonic Power Angel
Watts UP
UPM
Kill a Watt

I wonder which is the most accurate.

Well the Seasonic and the Kill-A-Watt are based on the same underlying hardware AFAIK (UPM is which one again? Becuase i think there is a europena model also based on teh Kill-A-Watts hardware)....so I figure they will be the same. The Watts-UP is the one i would be interested in knowing about.

[SKYMTL]

UPM is this one:

LINK (http://www.upm-marketing.com/products/ProductDisplay.cfm?CFID=1051324&CFTOKEN=23535517&pt=%22(%20'A%0A&rtRn=484%263O%5E_%3A29VU%5B%5ERP%3D%5BY6%226N%22W% 5EL%20%0A&item=%22)0%3FA%0A)

Specs say it is good to 1800W and it doesn't look like it is based on the Kill-a-Watt hardware.

I wonder if Seasonic will release a new load accessory for their Power Angel. It would be nice to have a 100W-500W variable load accessory for the Power Angel.

[jonnyGURU]

As for the power meter, do you mean somthing like this:
http://www.extech.com/instrument/products/310_399/380801.html

For some reason, I can't find it for sale on that same site.....

That's the one I'm talking about.

Someone with a load tester should do a roundup of sub- $150 power meters. There are quite a few:

Seasonic Power Angel
Watts UP
UPM
Kill a Watt

I wonder which is the most accurate.

I'd love to do that, but I'd have to buy the meters. :( $$$

[SKYMTL]

How about you buy them, test them and I offer to buy the best performing one from you? :D

[alexk]

I'd love to do that, but I'd have to buy the meters. :( $$$

You can try to contact the companies (or the distributors who sell these things) and ask for a free "evaluation unit" for an "upcoming review", or something like this. You probably already know how to do that properly :D

[eightballrj]

I wish I could send you the Watts up Pro that I have sitting in my office right now... BUT... Inventory Item. I have a Uei clamp on also. It has served its purpose of the last little bit. Don't know about accuracy because I have never compared it to anything else.

[Kvar]

So the time has come to upgrade my test system to something that can can stress 700W+ PSUs.

The problem is that I am a bit worried about the accuracy of my UPM power meter above 500W after reading Jonny's post about how the Kill-A-Watt became inaccurate at higher loads.

I have considerable amount of experience with AC power measurement and currently own two power meters. Used to own three, but one was bought back by the manufacturer due to unresolved issues.

The Kill-A-Watt is without doubt, amazing value for the cost. To get something with a comparable functionality from an instrument manufacturer you'll be paying a mid three to low four figures for their low end stuff. Four to five figures for the top of the line stuff.

What is it about Kill-A-Watt you find not good enough and most importantly, what is the worst accuracy that meets your purpose? Accuracy cost a lot of money in general in the metrology world.

You won't find one that meets the published specs for a bargain. High accuracy instruments are individually calibrated by hand by skilled metrologists and the factory specs are only guaranteed to be valid for usually a year, then it has to be sent back for a recertification and a calibration.

Even if the calibration is expired by a few years you'll get a good ballpark figure but looks like that's not what you are after, so doing away without a current calibration certificate is apparently not an option for you. The usability depends on the amount of error that is acceptable to you. You can pull "I reckon it's accurate within 5%" out of your butt. You won't be able to say "I'm 95% confident that my measurements fall within xx% of the measured value"

You might find one at a significant discount online second hand, but expect the calibration to be outdated. If you want it to be at rated specs, you'll have to send it out to a calibration lab and shell out a minimum of $100.


I am looking for something that is rated a bit higher and is a bit more of a "Professional" tool. So far I have found a few options:

Brultech2020 (http://www.powermeterstore.com/index.php?cPath=112&products_id=4124)

WattsUp Pro with PF logger (http://www.powermeterstore.com/index.php?cPath=112&products_id=1206)

Brultech2020. No idea. It does not give you the frequency bandwidth nor does it give you the accuracy specifications to my satisfaction.

Watts Up: Definitely avoid. Bad experience. After a replacement and unresolved issue, the manufacturer had to buy it back for an issue at engineering level.


Fluke Clamp meter (http://www.powermeterstore.com/index.php?cPath=&products_id=3542&p_tab=accessories) which I could use with my Fluke 177. The only problem with this approach is (if I read right) the Fluke 177 can only read up to 10A accurately if I am reading this correctly:
Fluke 187 Specs (http://us.fluke.com/usen/products/specifications.htm?cs_id=26864(FlukeProducts)&category=HMA(FlukeProducts))

Or a true clamp meter (http://www.extech.com/instrument/products/310_399/380926.html)

Useless. You won't be able to do the basic measurement even the Kill-A-Watt can do.

Volt x Amp = Watts, if and only if you are operating a pure resistive load.

Clean sinusoidal current with phase shift, such as an ideal induction motor
volt x amp x cosφ

Reality:
Combination of phase shift and non-sinusoidal current.
AC power measurement in real life is not a trivial matter. The Kill-A-Watt accomplishes this task using an ASIC DSP.

The sum of instantaneous volt x amp divided by the number of samples during the integration period. The number of samples needs to be high enough to account for every part of the current waveform. So, if the display is to refresh twice a second, the chip would have to take the reading of both the voltage x amperage 500 times every 1/2 second, then divide the sum of the products over 500 to give a true average power(over a half second period). This isn't a hard process for custom made DSP.

PF = the wattage you get above/V*A average
Don't get it?

Draw two waves, 90 degrees out of phase on a sheet of paper. Call them y1 and y2 If the volt =2 and amp = 2 and you simply multiply the two together, you get 4. This is called 4VA.

Now, multiply y1 and y2 along the same x value. No matter how many times you do it, you'll get a 0 everytime. So the result is you have 0W, 4VA, power factor = 0.00.

They do it just like this with a DSP based meter. On an analog process meter, this is done using an analog integrator and the analog output is converted to digital with a A/D converter.


Theory of AC power measurement:
http://www.yokogawa.com/tm/tr/tm-tr0605_01.htm

The best you can obtain with a voltmeter and an ammeter is the VA. If your ammeter isn't a true RMS, you can't even do this. Power factor is predictable and it is around 0.65 on standard power supplies and 0.98 on active PFC power supplies, so you can guess a wattage this way, but this isn't a measurement. This is a rough estimate.


So, what do you guys think? What should I do?
None of the above. Sorry. Buy a REAL instrument.

Brands to look for:
HP-Agilent, Hioki, Yokogawa, Voltech, Dranetz, Fluke and any reputable manufacturer who can actually produce a detailed specifications and make NIST traceable certificate available.

Upper range, but not quite lab grade:
http://www.epd.com/power_meters.html

[Kvar]

I wish I could send you the Watts up Pro that I have sitting in my office right now... BUT... Inventory Item. I have a Uei clamp on also. It has served its purpose of the last little bit. Don't know about accuracy because I have never compared it to anything else.

Ok, now I'm going to talk about the Watts Up Pro. I bought one, and I no longer own one.

After I bought mine, I noticed my unit showed an error that is well beyond the 2% or whatever accuracy they claimed when it was used to measure non-unity (which means power factor is not 100%) loads. I don't have the figures off top of my head, but it was a GROSS error.

The Watts Up series uses a current transformer while the Kill-A-Watt use a shunt. Current transformer is more vulnerable to phase error and would not have a linear response to harmonic current which can extend into thousands of heartz. I believe the core was made of iron, so I would expect it to truncate much of the higher frequency harmonics present in a computer power supply type load and add to inaccuracy.

It measured voltage and current reasonably well, however the problem was that it showed a significant amount of power factor measurement error. This was confirmed using a lab grade power meter daisy chained with the Watts Up Pro. The lab power meter is not powered directly from the power source under test, so connecting it downstream of the Watts Up does not create an error.

The two meters agreed well with a resistive load. With an induction motor load drawing around 150W, the two meters showed a significant discrepancy. Amp and volt readings were close together, which means the Watts Up had a phase error. I contacted the manufacturer with my concerns.

I provided them with data reported from both instruments, as well as the data from my Kill-A-Watt. I heard back from the company president Mr. Brad Volin and he asked me to return the meter. I got a replacement meter. Same exact problem, however the absurd skew was in the opposite direction. The meter went back to Doubleed as soon as I got the return label from them. I requested that they repair my unit or provide me with yet another replacement that meets published specs.

Mr. Volin reported back to me that they were able to duplicate the problem and that their engineering team was looking into it. I was told there was an issue beyond the specific units I have handled. I didn't need the thing right away, so I gave them some time, but after a few months, they could not provide me with a unit that meets the specs and offered to buy back my unit.

This was in late 2005 or 2006. So unless they've come up with a fix since then, I would avoid it. The actual data got buried somewhere, but the amount of error was obscene by my standards. If you contact Mr. Volin, he maybe able to provide you with the actual data so you can see if the error is acceptable or not yourself.

To their credit, the computer software features are nice, but with a serious flaw in the fundamental operations, it was not practical for me.

[Spectre]

The funny thing is with the Kill-A-Watt it occasionally gives readings of over unity with certain power supplies.

[Kvar]

The funny thing is with the Kill-A-Watt it occasionally gives readings of over unity with certain power supplies.

By how much?

This can happen when there's a slight error in calculated watt and VA. Input on power supplies contain harmonics, so it's advisable to use something with specified bandwidth of no less than 2KHz to avoid this kind of error.

http://www.evaluationengineering.com/archive/articles/0598power.htm

[Spectre]

By how much?

Depends on the unit but on some untis it gets as bad as saying a unit is drawing 100w when the DC load is 105w but the fact that it does it at all makes it useless for determining efficiency.

Then there are other units that it never has a problem with. :crazy:

[Kvar]

Unity has nothing to do with efficiency. Unity means that the ratio of input watt divided by the product of input volt and amp = 1.00.

A light bulb on 120v circuit that takes 1A and uses 120W is 120W/120VA, or 1.00 PF.

A computer that takes 120V 2A and uses 120W is 120W/240VA, or 0.50 PF.

The symptoms you're seeing seems like a measurement error. What is your load and how do you know that it is 105W? Use a steady load that doesn't fluctuate, such as a 150W rated 1 ohm resistor or 2 car headlamps.

Measure the current going to the load. Measure the voltage of that rail under load.

Multiply the volt x amp to get watts. (you can do this with a steady state load running off of DC.

Divide this by the watts reading you get on Kill-A-Watt. Regardless of power supply you should between 0.65 and 0.85 efficiency(not to be confused with power factor)

[Spectre]

Unity has nothing to do with efficiency. Unity means that the ratio of input watt divided by the product of input volt and amp = 1.00.

A light bulb on 120v circuit that takes 1A and uses 120W is 120W/120VA, or 1.00 PF.

A computer that takes 120V 2A and uses 120W is 120W/240VA, or 0.50 PF.

The symptoms you're seeing seems like a measurement error. What is your load and how do you know that it is 105W? Use a steady load that doesn't fluctuate, such as a 150W rated 1 ohm resistor or 2 car headlamps.

Measure the current going to the load. Measure the voltage of that rail under load.

Multiply the volt x amp to get watts. (you can do this with a steady state load running off of DC.

Divide this by the watts reading you get on Kill-A-Watt. Regardless of power supply you should between 0.65 and 0.85 efficiency(not to be confused with power factor)

I am well aware how to get efficiency, and how to calculate the wattage via voltage and current. I know the DC load because I am programming the load into the tester.

I only mention the situation in case you actaully knew why the Kill-A-Watt would do so only on particular units and not all. The problem seems completely unrelated to the units inability to measure above 1500VA accurately.

Part of the reason I dropped using the Kill-A-Watt (and I blew a couple up).

[Kvar]

I am well aware how to get efficiency, and how to calculate the wattage via voltage and current. I know the DC load because I am programming the load into the tester.

I only mention the situation in case you actaully knew why the Kill-A-Watt would do so only on particular units and not all. The problem seems completely unrelated to the units inability to measure above 1500VA accurately.

Part of the reason I dropped using the Kill-A-Watt (and I blew a couple up).

What power supply causes the Kill-A-Watt to read beyond unity and by how much? What does it display for power factor? I asked this initially and you talked about efficiency, but didn't provide an answer in regard to "beyond unity". With purely resistive load, mine can sometimes read something like 150W/151VA, which is not possible, but such a small error is from the measured data not adding up perfectly.

[Spectre]

What power supply causes the Kill-A-Watt to read beyond unity and by how much? What does it display for power factor? I asked this initially and you talked about efficiency, but didn't provide an answer in regard to "beyond unity". With purely resistive load, mine can sometimes read something like 150W/151VA, which is not possible, but such a small error is from the measured data not adding up perfectly.

I'll have to go back and look which ones but when it does read better than 100% efficiency after calculating it stays at 1.00 or .99 most of the time. The Coolermaster Real Power Pro, the Koolance, and SilverPower were reading above 1 by how much I am not entirely sure since I pulled the Kill-A-Watt and those tests were done ~6 weeks ago. I may have written it down in my note book but I changed power meters after the the 450w-500w roundup when I lost 2 units to the Powerteks.

Part of the point of this is while cool....the Kill-A-Watt can really bite you on the ass if you don't pay attention to it. Why it gets so buggy I don't know, but there seems to be a good amount of unit to unit variation on what you load it with from my expereince. Some power supplies it reads just fine but others it is worthless.

[Kvar]

So the time has come to upgrade my test system to something that can can stress 700W+ PSUs.

The problem is that I am a bit worried about the accuracy of my UPM power meter above 500W after reading Jonny's post about how the Kill-A-Watt became inaccurate at higher loads.


I have a Kill-A-Watt meter myself. I ran a test at said higher loads. Test done at about 780 & 1450W. Since I don't have an electronic load, I used an oil heater as a load.

Load one read 760 and 800watt and load two read 1440 to 1460W. The reading was rather jumpy within the range.

Since the load was resistive, I can use V*A =W so i tested it using my DMMs with current calibration tag.

Voltage test = Fluke 87V (+/- 0.7% + 0.2v)
Current test = Fluke 187 (+/- 1.5% + 5mA )

121.0 x 6.47A =782.9W +/- 2.2% or 17W
119.4 x 11.91A = 1422W +/- 2.2% 31W

For "sizing" stuff, do you really need something much better than +/- 10% accuracy? Be glad you can get that close with a $30 instrument.

If you want something well better than 1% accuracy, you can get a Voltech PM100, which is around $2,000 new, maybe 1/2 to 1/3 that used, but you'll have to pay about $200 to have it calibrated.

[ianm2]

I was going to mention that, I used to be into valve, or as you say, tube electronics, but I did most of it all without most lab equip., and I could never really understand what I was doing, better than some, but not 'there' if you know what I mean....... loads about some things, gaps with other basics.

So I am relearning ac theory, and this is coinciding with that, as well as a thanks to jonnyguru, an interest in pc power supplies.

I have a farnell catalog, big components supplier, I am sure you have one there a big trade supplier or a few, and started looking at all the test and measurement equipment, and what it does.

And indeed, there are power meters there, some multimeters have them, and there are more expensive lab type bench ones, kvar has already mentioned them, so I won't go further

hameg have one, and the pdf is listed online, so you can learn a little more if you don't know enough, the manual could be better written, but its got some theory

its a great place to go, makers websites and download manuals/data sheets/tutorials, if they are well written, you learn more applicable stuff than a textbook will teach

oh and some can do 8000watts

hint, if you want the inside knowledge, go to the trade suppliers/makers. there are always pro people and suppliers who know loads of stuff.

one of the keys to understanding this ac stuff is that waveforms are linked to circles and triangles, so you have to understand angles, remember 360 degrees in a circle, 360 degrees ( the circumferance ) is 2 x pi x the radius, so to get the radius in degrees, you divide 360 by 2 x pi, and you get 57 degrees, which is called one radian.

180 degrees inside a triangle, right angle is 90 degrees, so the other angles add upto 90.

sines, cosines, tangents come next. soh cah toa...links the angle to the sides of a right angled triangle.

why is it importatant? because capacitors and inductors, REactive components, shift the position of the current relative to the volts wave, phase.

C I leads V , V leads I in inductors, CIVIL.

inductors and caps have resistance, its called reactance, combine that with a dc resistor, and you have to find the total which is called IMPEDance, the diagonal or the triangle, but that's trigonometry, or geometric sum,also called a vector, not arithmetic adding.

ditto with power, you get dc power( called real, active, or true power) , reactive power, and the geometric sum, which is apparent power( neither dc or ac but a combination of the 2), this is the diagonal of the triangle, and as its the longest side, its more power. power factor is the angle between dc and apparent.

so your power isn't simply VxI as you only have bits of V and I at the same time. And then, you get into calculus, and numbers called complex ones, again which aren't hard, just to do with -1, a special number) which isn't hard really it just scares people.

yes its all inextricably linked. dunno why I couldn't understand it at school. I couldn't link the bits together.

sorry if its sucking eggs, but I bet many have no clue, so start drawning triangles, no I am far from there yet, but getting there

[Kvar]

I can't believe I didn't state it in earlier posts. I'm using a Kill-A-Watt and a Valhalla Scientific 2101 as the other meter.

http://img201.imageshack.us/img201/1377/instrumentmg5.jpg
If calibrated annually, its accuracy for up to 5KHz is: ± 0.25 % of reading ± 6 counts

[Kvar]

I have a farnell catalog, big components supplier, I am sure you have one there a big trade supplier or a few, and started looking at all the test and measurement equipment, and what it does.

And indeed, there are power meters there, some multimeters have them, and there are more expensive lab type bench ones, kvar has already mentioned them, so I won't go further

hameg have one, and the pdf is listed online, so you can learn a little more if you don't know enough, the manual could be better written, but its got some theory

its a great place to go, makers websites and download manuals/data sheets/tutorials, if they are well written, you learn more applicable stuff than a textbook will teach

oh and some can do 8000watts

You really don't direct connect a power analyzer for high power stuff so most power analyzers can do tens of thousands of kW using a potential transformer and a current transformer.

The power company's meter at commercial facilities can read kW usage, but the current doesn't actual travel through the meter. They use a current transformer that wraps around the big fat wires and gives a proportional output, i.e. 1A through the meter for every 1000A of current in the wire wrapped inside. Newer software based meters let you program in the ratio so the read out is direct, including prefixes.

[Travis]

After reading the whole thread I think I'll carefully choose the equipment to measure input active power. Even a shunt based Kill-A-Watt power meter may have an unwanted error large enough (5~10%?) to make the efficiency numbers I calculated meaningless.

So how much will a capable power analyzer cost me if I want an accuracy of ~2KHz and ~1%? Does this bench-top meter meet my need?
http://auction1.taobao.com/auction/item_detail-0db2-fba55d50bca3e501b18a590775cf434e.jhtml

Plus a question for Jonny: How much is your Weibo power meter?