And what they tell you about ...let say ...a 60W light bulb? Will they sing with one voice about such load?

I would put my faith in the KAW before a regular consumer-level UPS. But yeah, maybe some testing this weekend with some light bulbs would be a good verification of the thing. Throw some low watt CFLs at it, and the 150W 3 way lamp also.

Kill-a-Watt and Seasonic Power Angel (the same device under the cosmetics) are usually accurate to ~1W from about 15W on up to at least a couple hundred watts. Compared to Wattage of light bulbs, and readings off $600 AC power analyzer.

Incandescent bulb is one thing. Easiest for a meter. "Economic" bulb is another thing. Fluorescent light with a choke - still another.
Edit: I am not saying kill a watt shows wrong numbers. Haven't had one.

psiu
Klusu
Yeah, I agree with Klusu, that the character of a load of a fluorescent lamp can differ a lot from the old type incandescent lamp, if the measurement instrument isn't aware of the different sorts of load (Kill-a-Watt do know). But this can be interesting subject for investigation -- how the power meter reacts on different type of light bulbs

Jay_S
Please, try one fluorescent(economical) and one usual pear-shaped incandescent lamp with your meters. If the meter will give a lie on fluorescent, it isn't aware of low power factor of load, so can't be accurate with a regular PSU.

FYI, KaW and SPA are equally accurate with CFL and incandescent bulbs.

Interesting. Made me wonder why I had not done a formal, systematic check of the Kill-a-Watts and Seasonic Power Angels that we use around the lab.

So I spent the last 40 mins checking the 2 SPAs and 1 KaW currently used actively in the lab against an Extech 380803 AC Power Logger -- this sells for $700~800 and is supposed to be accurate within +,-0.9% for power and 0.5% for all the other measurements. The load was a 75W incandescent bulb, and 23W and 15W CFLs in various combinations. I will do more testing and pull all the data together into a table for a short article on the main site, but here's a preliminary summary.

The KaW always gave lower power (W) readings compared to the Extech -- 92W vs 97.3W, 71-72W vs. 75.4W, 33-34W vs. 36.4W, 13W vs. 15W. It will be pulled from active duty in the lab.

One of the SPA gave power results within ~1W of the Extech -- usually low, except one, which was higher: 98W vs 97.3. The PF reading was consistently 0.2-0.3 low compared to the Extech. This one will be kept in service.

The second SPA was close to the first, but slightly less consistent, with PF readings slightly lower sometimes, and power reading also slightly lower sometimes. This one will also be retired.

I will be checking accuracy at higher power loads as well. There are a couple more Power Angels still unopened; they'll be opened and checked against the Extech. Only the ones that have consistent accuracy within ~1W will be put into service.

All this assumes that the Extech is accurate... which I probably should not assume. But at this point, I have no easy way to check the Extech. I guess I will see if the EE dept at the Univ of BC has something...

Thanks for the inspiration, Ksanderash.

Interesting discussion all! I'll do some light bulb tests tonight if I remember. For those of you who have done them, how close do your measurements come to the rated wattage of the bulbs?

MikeC,

If the university fails you, there are likely 3rd-party calibration facilities in your area. If there is any high-tech industry in Vancouver, there will be calibration shops. Where I work, we annually send out measurement/inspection equipment for calibration to maintain AS9120 quality compliance. If your equipment is too large to send out, the a calibration shop will come to your premise.

I checked my Kill a Watt against some lamps:

Looks OK to me. The incandescent is 1.7% off the bulb's rating. Unfortunately (for testing!), I don't have any other incandescent lamps.

Uh... yeah, right. For about the price of a new Extech, I'm sure. B&K offered to calibrate my old SLMs, too, for many hundreds of $$s.

There is an another E(x)tech, PM300, Kill-a-Watt competitor.



And it has a quite negative feedback.

There is an article (fcenter.ru, remember?), mentioning PM300 as an example of a good meter, working correctly with non-linear loads. I would expect it to, because it is not the cheapest... Too bad if it doesn't.

Klusu
Yeah, here is that article, cloned at Xbitlabs for English speaking people.

We’ve got one, too [PM300]. However, we use it only when we need to quickly estimate the power consumption of a computer or some other device (a consumer wattmeter is most handy then because it needs no kind of preparation) but not for serious tests.

It's OK. These are produced no so high-precisionly, that you can check measurement equipment by them

re our Extech -- 380803: True RMS Power Analyzer Datalogger:
http://www.extech.com/instruments/produ ... prodid=206

Wow...sorry, feel like I put a penny on the tracks watching this thread derail onto something completely different, yet so awesomely SPCR.

Good thread. Just what I was searching for.

Last year I bought a GA-MA74GM-S2H rev.1, I paired it with ECC memory, and everything seemed to be working.

So last week I bought a GA-MA74GM-S2H rev.3 to replace my rev.1 (yeah... no comment), my power consumption drop from 28W to 27W and I was quite happy.

Then AMD announced its will to drop DX9 VGA support. And I also found that the ECC MODE on GA-MA74GM-S2H is NOT working at all! Sad day

Our CPUs DO support ECC memory... So motherboards support ECC memory, too... But for ECC memory to fullfill its job, the chipset has to be instructed. There should be a "ECC MODE ON/OFF" BIOS entry.

No entry? No ECC MODE. That's it.

The quickest way to check if ECC MODE is enabled on your system is to fire up MemTest86+. As already said, you can find it in the Ubuntu/Kubuntu CD-ROM. Start it from the boot menu.

Right now I am very temped by this Biostar A760G, but also the Asus M4A78L-M seems to be a nice choice...
The thing block me is that revenue of Biostar is 300'000 $, revenue of Gigabyte is 1'500'000 $, and revenue of Asus is 23'000'000 $ ...

Just 2 question for anyone has this BIOSTAR:
a) Can I undervolt the CPU?
b) Does MemTest86+ say "ECC enabled"?

Measuring a resistive load is not a goog test. These Cheap Power Meters show their weakness with our PSUs, it is not uncommon for them to measure a value lesser than the real one.

A CFL is not resistive. PF about 74%. 59W for a 60W bulb is very close (I measured one 55W).
I bought this, cheap, measures real power and PF, claims ±0.5% power accuracy, seems quite good:
http://www.nedis.com/datasheets/PRODUCT ... HQ_ENG.PDF
What I wrote earlier about 33W 690G was wrong. Too good to be true...

That doen't mean anything. Look at Asus' budget boards -- they often comes with missing components, like mosfets, caps... I haven't seen this on Biostar. Btw, you forgot to mention ECS, which is as powerful as Asus, and how about the quality of ECS boards...

Klusu
Can't find any power meter here in Moldavia )) I'm feeling like in a Stone Age here.

I posted these images over at [H]ardforum, in this thread: Any way to confirm ECC functioning?. These are shots of my working system with the Biostar board.



ECC enabled:


Memtest 86+ confirms it:


There's more discussion over there if you're interested. And in this post by another [H]ardforum member.

I am very interested in motherboard with bios undervolting and ecc support.

It is ideal for a low power vmware ESXi server, I would like to pair it with a low power quad core (when you have 4 tasks to do probably draws less power a quad core at a the minimum freq than monocore at highest freq).

Can someone say if these motherboards support ESXi????


Thanks in advance!

Mario

Thanks to everyone for all the great info in this thread. I just got a new server up and running using this motherboard.

AMD Sempron 140
2GB Kingston DDR2-800 ECC non-registered
Antec NSK6580
Antec Earthwatts 430W
OCZ Vertex 30GB SSD
3x WD15EADS

With the hard drives not powered up, the system is idling in Windows at 34w, according to my kill-a-watt (800MHz, 1.056v according to CPU-Z). Seems nice, but I'm sure the CPU could probably go lower on the voltage.

UPDATE: using the motherboard VID control, I am getting actual voltages lower than 1.075v, contrary to what an earlier user with a Sempron 140 reported. I am using the latest BIOS as of today's date.

I applied some of the tricks discussed earlier (disabling sound, lowering graphics clock to 150, dropping memory speed to 533), and I am currently ORTHOS testing at 800MHz, seeing how far under 1V I can go.

2ND UPDATE: Despite showing values lower than 0.8v, it won't post. So I'm stuck at 0.8v for the time being. Idle power is at 25W w/o hard drives.

I just picked up this board for a NAS build. If you have a Fry's near you, they have a combo with this board + a Athlon II X2 250 for $50AR. The $10 rebate might be a bit flaky, so actual cost might be $60.

http://slickdeals.net/forums/showthread.php?t=1756164

The Athlon II 250 is overkill for a file server, so I will do a swap with the Sempron 140 from my HTPC. Glad to hear the BIOS was able to get it down to 0.8v. I have been using k10stat on the HTPC with the Sempron 140 and have been able to get it down to 0.7125v (CPU-Z shows 0.720v) @ 800Mhz.

Sadly, this board has been deactivated on newegg (although you can still buy it open box; also available on plenty of other online retailers). Perhaps Biostar is stopping production of them?

Anyway, I looked at the BIOS manuals for a few other Biostar motherboards as well. All the boards based on the 770 chipset have the ECC BIOS options. And the A770E also appears to have a three-phase VRM (plus it only supports <=95 watt CPUs). Newegg doesn't sell it, but the A770E3 looks like the exact same thing, only with DDR3 support (i.e. strictly AM3).

The 770-based boards are full ATX, rather than microATX, which is good or bad depending on your situation. Bad, because it's bigger, and won't work in microATX-only cases; good because you get more PCI slots and an additional PCIe x1 slot.

The 770 chipset does not include integrated graphics. You could probably get virtually the same power consumption using an old, fanless and heatsink-less PCI-based graphics card. However, the BIOS also has a "headless" mode option. According to the BIOS documentation, this is exactly what it sounds like: allows running without a keyboard, mouse or monitor. I'd like to think that this means, after installing your operating system and some kind of remote management tool, you could completely remove the video card and save a couple watts or more.

However, the A760G also has the "headless mode" BIOS option. When I first got the board, I tried disabling the internal GPU while enabling the headless mode. But the board wouldn't boot. I didn't work at this very hard, because I figured the internal GPU probably didn't use much power anyway.

So, it's unconfirmed whether or not headless mode lets you run without a graphics card.

Finally: according to Wikipedia, the 760G chipset is manufactured on a 55nm process, while the 770 is a 65nm process. In theory, that could mean higher power consumption for the 770, but in reality... who knows!

I might eventually snag one of these 770-based Biostar boards and see how they compare to the A760G... first I need to get rid of a lot of old junk, and also wait for a deal/open box.

Mine was an open-box pickup from Newegg and works fine BTW. No I/O panel shield or accessories but works great.

What about the Biostar TA760G? It's more expensive and has 2 more memory slots, but appears to have more BIOS options.

I don't have enough posts to post a URL, but the Newegg number is N82E16813138137

I have it, the differences are the one you have said. Does have ECC support. Can undervolt. Good board.

I have it on order...

Thank you everyone for pointing out this board.

I replaced the popular Gigabyte 740g motherboard with a Biostar TA760g in my WHS box. According to my Kill A Watt, it now idles at ~27 watts when the hard drives spin down.

Biostar TA760g motherboard
BE-2400 processor undervolted to 1.075V
1 2GB stick of ram
1 WD 250GB laptop system drive
2 1TB WD Green drives
Antec 380 watt power supply

The 740g board is moving to a different PC that doesn't run all day long. It previously idled around 40-45 watts. I was able to swap just the motherboard without changing any other components and WHS did not complain. I had to install a few drivers for the new board, but that was it. This was after taking many backups and preparations, anticipating a WHS re-install.