Teardown: AnTai ATX.9801 handheld plug power meter

[djacobow]

AnTai ATX.9801 handheld single-phase power meter teardown


Hey guys, I've been on the prowl for a low cost power meter that can do a decent job with small loads with high crest factors (like you might get from an unloaded iThing charger) for awhile.

I found this guy on Fleabay for $65, shipped. http://www.ebay.com/itm/311283405607

It came well-packed in a decent box, and shipping was quick. It came with a Chinese power connector and a very low quality US adapter. I will probably just snip the cord and put on a proper US socket as the adapter does not even pass through ground.

A two-sheet owner's manual was included, all in Chinese. Too bad, too, as there is a button that steps the unit through some modes. I don't know what they do, though. I suspect they let you set your electric billing rate, etc.

Overall, the quality seems fine for the price. There are fuses, there is some attempt at cutouts to keep the HV and DC sections separate, but this is no work of art.

The unit is built around a CS5463 power/energy IC: http://www.cirrus.com/en/products/cs5463.html. MCU is some kind of 8051 from STC (15W1K16S).

When I first plugged it in, it measured 129Vac, which let the wind out of my sails immediately. But just to see how far out it was on voltage, I measured the mains with both my 87V and 34401A and it turns out that my sockets really are that hot. (If they stay like that perhaps I should have a chat with the local utility.) Anyway, all three meters were within 0.1V of each other. Not bad.

I also compared the frequency readout to what my HP gives. Not that I've ever calibrated the HP, but just for rough quick-n-dirty: the AnTai reported 59.94Hz while the HP reported 59.96.

Unfortunately, for measuring power I don't have any calibrated equipment against which to compare.

I plugged in a 43W incandascent lightbulb to the unit. It reported a pf=1.00, real power = 48.93 while the Kill-A-Watt reported pf=0.96, 41.2W (real). Those are just too far off to make much comparison. I suspect it's the K-a-W that's way out.

I also plugged in a project I'm working on which has a lightly loaded ST Viper PSU. It read 4.3 W (real), but I don't have a suitable tool to compare. The Kill-A-Watt read 2.9W (real)

Overall, for $65, it's not that bad, but I still want a proper power meter. I've been eyeing the Chromatech units that show up used on Ebay. Anybody have an opinion?

[NiHaoMike]

If your electronics are universal 100-240V (as most are nowadays), they will actually enjoy the little bit of extra voltage.

With a purely resistive load, you can calculate wattage using voltage and current measurements that can be done with a standard multimeter.

Does it have the capability to average readings? If not, it would be largely useless for determining actual energy use of loads that vary a lot.

[coppice]

Overall, for $65, it's not that bad, but I still want a proper power meter. I've been eyeing the Chromatech units that show up used on Ebay. Anybody have an opinion?

$65 seems a lot for that device. If the big utility meter makers could get that much for a simple single phase power/energy meter they'd be over the moon. :-)

If you want a lab grade power meter the most common ones in power labs seem to be the Chroma and Yokogawa ones. Both make very satisfactory single and 3 phase meters. A big pitfall with these things is their bandwidth. Some power measurement specs call for agreement with a particular model of power meter, because no two commercial power meters measure over the same bandwidth. For a pure sine wave load this makes very little difference, but with a typical distorted electronic load the readings can vary a lot as you select different bandwidths. If you want serious power analysis the Kikusui machines are very good, but rather expensive.

[djacobow]

If your electronics are universal 100-240V (as most are nowadays), they will actually enjoy the little bit of extra voltage.

Sure, I don't think most stuff with a switcher will mind, and maybe they'll be a bit happier to sip less current. But there are other things in the house like motors and unregulated loads that might not be happy. 129V is out of spec for North America, by any measure. Here's a surprisingly complete doc from the local utility that describes what standards are in effect:

http://www.pge.com/includes/docs/pdfs/mybusiness/customerservice/energystatus/powerquality/voltage_tolerance.pdf

With a purely resistive load, you can calculate wattage using voltage and current measurements that can be done with a standard multimeter.

Yeah, I'll do just that if I get a chance soon. The incandescent bulb is pretty close to pure resistive, and I believe I have some power resistors around somewhere.

Does it have the capability to average readings? If not, it would be largely useless for determining actual energy use of loads that vary a lot.

It does not. It has a window labeled "annual kWh" but I don't know if its applying some kind of smoothing/averaging in the calculation or not. Bummer.

It might be fun to hack out the 8051 and hack in something else to talk SPI to the Cirrus part, then I can do whatever I like. The part itself is actually pretty capable. A more involved project that I just don't have time for these days.

[coppice]

The incandescent bulb is pretty close to pure resistive, and I believe I have some power resistors around somewhere.

That's a common misconception. Something like a water heater is close to a resistive load. An incandescent bulb is rather non-linear. It's temperature varies so much during the mains cycles that with a really pure sine wave voltage waveform the current waveform might have 10% THD. The harmonics in the current waveform contribute nothing to the power consumption, if the voltage waveform is really pure. How pure will the voltage waveform in your house be? Somewhere between zero and 20%THD is common. Its complicated. 

[djacobow]

If you want a lab grade power meter the most common ones in power labs seem to be the Chroma and Yokogawa ones. Both make very satisfactory single and 3 phase meters. A big pitfall with these things is their bandwidth. Some power measurement specs call for agreement with a particular model of power meter, because no two commercial power meters measure over the same bandwidth. For a pure sine wave load this makes very little difference, but with a typical distorted electronic load the readings can vary a lot as you select different bandwidths. If you want serious power analysis the Kikusui machines are very good, but rather expensive.

My personal interest is actually in "vampire" or standby power, particularly from devices that use "almost no" power but of which there are many in the house. As you know, these devices do have very nonlinear current waveforms, so yeah, bandwidth is a concern. I've used some nice Yokogawa gear and would love to have one, but the Chroma units seem to cost less and show up used on Ebay more.

$65 seems a lot for that device. If the big utility meter makers could get that much for a simple single phase power/energy meter they'd be over the moon. :-)

My interest is piqued. What do the utilities pay for single-phase "smart" meters they are putting on residences?

[djacobow]

OK, so another check-in:

+ yes, it does integrate energy

- my units locks up and becomes completely inoperable after a few minutes.

So, for $65, complete garbage.

[WackyGerman]

Well here in germany you can get a power meter from Christ Electronics which measures a lot of stuff for roundabout 200 € . It s a step between your antai meter and the real expensive ones from yokogawa . Here you can loan one of these Christ meters from the local power provider to check all you electric units of the power consumption for free . Older ones like the Christ CLM 200 which are rejected by the providers can be bought cheaply at ebay so as I did .
But unfortunately I don t know if they have versions for the US outlets .

[djacobow]

But unfortunately I don t know if they have versions for the US outlets .

That's not a problem. I can easily hack on a US connector if the voltage range is handled. I don't read German, but it looks like the power sensitivity is only down to 0.1W which is probably not good enough for my purposes.

[Engineer1]

- my units locks up and becomes completely inoperable after a few minutes.

So, for $65, complete garbage.

Sorry for the thread archaeology...

I bought one of these (thanks a lot for putting me on to it) as I needed to measure the power taken by a simple transformerless power supply design (capacitor, bridge and Zener type). It worked very well. The voltage, current and frequency accuracy were all within 0.1% of my other instruments (HP meter, Fluke frequency counter).

I checked the power measurement with a resistive load set to dissipate 1W (based on the voltage measured). The unit measured at about 1.02W, with a PF of 0.999.

I didn't experience any lockups. Left it running for over 24h, and it was still ok.

The adapter plug supplied is horrendous, though - I just cut the plug off and fitted a standard UK one.

So, as a budget unit to give you more than the Kill-a-Watt type devices, this is pretty good. It certainly solved a problem for me.

In an ideal world, with time to spare, I'd take voltage and current waveforms, and use Matlab to work everything out accurately. Sadly, my world isn't ideal...

Cheers.

[Uwe Quast]

Hy djacobow

you'll find english info about that meter directly on AliExpress's : HOPI HP-9800 Handheld Power Meter page

best regards
Uwe