How does one test efficiency rating on a power supply? As a consumer just being curious?

[Space Lynx]

I have a few power supplies sitting around over the years, and I was wondering how hard it is to test 80 plus power efficiency? I know JohnnyGuru and others do this, does it require specialized equipment or can an average consumer like myself test to see what performance I am getting at say 50% load, etc. Probably would need to buy a wall outlet watt reader, but what else would I need?

 

[moproblems99]

I have a few power supplies sitting around over the years, and I was wondering how hard it is to test 80 plus power efficiency? I know JohnnyGuru and others do this, does it require specialized equipment or can an average consumer like myself test to see what performance I am getting at say 50% load, etc. Probably would need to buy a wall outlet watt reader, but what else would I need?

You need to know what you are pulling from the wall for one. Then you need to know what all your circuits are pulling. Two ways to do that.

The first is with a volt meter and ammeter, or a handy dandy multimeter. Volts isn't so bad to measure but amps can be a bitch because it needs to be measured inline. Most multimeters have a 10amp fuse which gives you a theoretical max of 120 watts you can measure on the 12V line. That likely isn't going to cut it. So, then you can buy an inductive amp probe which you clamp around the voltage line and you can figure out how many amps you are drawing through inductance. So, with this method, you likely are going to need 2 tools. Volt meter and amp probe.

The other is measuring through software and hope that all your devices/circuits are covered.

After that simple division.

 

[Grog6]

Power factor correction is more likely to make a measurable difference in your power bill.
Most power supplies are about 67% power factor unless it's corrected, and that draws more excess power than the 80% efficiency to 90% efficiency difference.

The rated efficiency only is correct at near 100% load, it drops pretty quick from there.

Like said above, to measure it, you measure power in vs power out on all lines.

 

[Mussels]

i have a wattage wall meter that tells me efficiency readings of devices, seems quite accurate - its over 10 years old so model numbers useless, i'm sure a more modern variety exists for your local power socket designs

 

[Space Lynx]

Power factor correction is more likely to make a measurable difference in your power bill.
Most power supplies are about 67% power factor unless it's corrected, and that draws more excess power than the 80% efficiency to 90% efficiency difference.

The rated efficiency only is correct at near 100% load, it drops pretty quick from there.

Like said above, to measure it, you measure power in vs power out on all lines.

I thought 50% load = the best efficiency? That's what I gather from all of johnnyguru reviews.

 

[Mussels]

I thought 50% load = the best efficiency? That's what I gather from all of johnnyguru reviews.

yeah, if an extra 1-2% matters to you at all

this aint the days of shitty 60% efficiency PSUs

 

[eidairaman1]

Ask @jonnyGURU

 

[delshay]

@jonnyGURU I think I'v seen a specialized unit that has a 24 pin input. Not sure what it does but it looks like you plug the 24 pin output of the PSU directly into it. Someone will need to ask what the unit does.

 

[Frick]

@jonnyGURU I think I'v seen a specialized unit that has a 24 pin input. Not sure what it does but it looks like you plug the 24 pin output of the PSU directly into it. Someone will need to ask what the unit does.

Those would be load testers as used by all serious testers. You can see it as a ... simulation of a computer. Years ago w1zzard used huge resistors to simulate the load.

Anyway, crmaris wrote an excellent article on power supplies a number of years ago. Highly recommended reading for anyome wanting to know about PSU's, or more accurately SMPSs (Switch Mode Power Supplies).

 

[PanicLake]

i have a wattage wall meter that tells me efficiency readings of devices, seems quite accurate - its over 10 years old so model numbers useless, i'm sure a more modern variety exists for your local power socket designs

A wattage wall meter doesn't tell you "efficiency", it just tells you power consumption.
To somehow use it to measure efficiency you have to compare different power consumption of (let say in this case) power supply with the same load situation, possibly measuring different load percentages for each PSU.

Measuring power consumption of just one device, won't give you any power efficiency information at all, unless you have same measurement data (made in the same way) from other sources.

 

[eidairaman1]

P= I*E, E= I*R pyramids

 

[EarthDog]

Why don't we ask @crmaris, TPU's PSU reviewer to answer........

I thought 50% load = the best efficiency? That's what I gather from all of johnnyguru reviews.

At a high level that is true... but the curves are so flat, it doesn't really matter.

In general, I run my PSUs around 70% of their capacity. This gives me room for growth (a more powerful GPU for example) as well as generally quiet operation. To buy a PSU and run it at 50% of its capacity is wasting your money. You can check the math but you will never make up the cost of buying say a 850W PSU versus a 550W PSU running it at 50% load as opposed to 70%.

The rated efficiency only is correct at near 100% load, it drops pretty quick from there

I don't believe this is true... look at efficiency curves in reviews, bud. like 3% from 20% to 100% load (80plus bronze+)... 10% to 100% on Titanium rated units

 

[Kapone33]

If you have a HX ar AX power supply from Corsair you can use their link software to measure efficiency maybe that is why they are so expansive. I agree with Earth Dog though it is a fail safe way to become an enthusiast by getting more than you need.

 

[delshay]

I generally go for a high end PSU as I don't want to fall short in the future. All my PSUs are either Titanium or Platinum class, ranging from 800w to 1200w, & I have two old Antec truepower which I don't care about which I use for experiments.

 

[FinneousPJ]

I mean you even mentioned @jonnyGURU but did you read this?

Testing Methodology – JonnyGURU.com

www.jonnyguru.com

 

[dirtyferret]

Don't know how many people recall Hardware Secrets but their (his) PSU reviews rivaled JG back in the day; he listed his test methods along with other web sites PSU test methodology

Hardware Secrets Power Supply Test Methodology | Hardware Secrets

www.hardwaresecrets.com

 

[Bill_Bright]

i have a wattage wall meter that tells me efficiency readings of devices

How? A watt meter tells you the total amount of power being pulled from the wall, but it does not tell how much power the connected device is wasting in the form of heat, at that point in time. In other words, I don't see how it can't tell you the efficiency of the connected device. You would have to already know how much power the connected device is outputting (or putting to "work") at that point in time.

For example, assume an 80% efficient PSU. If the computer (CPU, motherboard, RAM, graphics card, drives, fans, etc.) is demanding 300W, it will pull from the PSU 300W (regardless if the PSU is a 350W, 500W or 1000W PSU). But the Kill-A-Wall Meter the PSU is plugged into will show 375W with that 80% efficient PSU (375 x .8 = 300). 75W is being wasted in the form of heat.

Unless you knew somehow already that the computer components were demanding 300W, that 375W reading on the Kill-A-Wall meter only tells you how much the PSU is demanding. It does not tell you the efficiency of the PSU (how much power the PSU is wasting in the form of heat).

Take that same 300W computer connected to a different PSU. Now the Kill-A-Watt meter shows 333W. What does that tell us? If we didn't already know for a fact the computer was demanding 300W, we would not know what that 333W means. It could mean the computer was demanding 233.1 watts and the PSU was an inefficient 70% supply (333 x .7 = 233.1).

But since we know the computer is demanding 300W, the 333W Kill-A-Watt reading tells us that is a 90% efficient supply (333 x .9 = 299.7).

average consumer

There really is no way an "average consumer" can do this. That's why we have to rely on the professional review sites as they have (1) the expertise and (2) the necessary (and expensive!) power supply analyzers that not only read how much the power supply is demanding, but they measure how much power the PSU is outputting too. You have to know both to determine the difference. Ideally, the ratio would be 1:1. But that will never happen. There will always be some energy loss when converting AC to DC.

You can't really use a computer to test a PSU's efficiency either because a computer's demands are constantly varying. You need something that can present the PSU with a pre-determined load (or preferably, a wide variety of expected loads).

Back in the day (before 80 PLUS and good review sites) I would test PSUs with my trusty 300W 10Ω ceramic power resistor ("dummy load") similar to this one (mine is variable with a sliding tap on one end to adjust resistance as desired). We knew how much power the PSU was demanding from the wall, we knew the exact load, the exact output voltage, and the current. So it was not hard to determine how much power was being wasted. BTW, that dummy load would get HOT!!!

Another problem for the "average consumer" is PSUs output 3 voltages (+12VDC, +5VDC, and +3.3VDC) simultaneously on multiple rails. Some PSUs even have multiple 12V rails. To properly test the efficiency of the PSU, you really need to test all the rails at once, or at least properly terminate with a known value the unused rails while testing one of the rails. This is because some power is always consumed on every rail, even when no load is presented to that rail. So even though it would be a tiny bit, that would still skew your efficiency results if you did not properly take them into consideration too.

***

@lynx29 Not sure why you want to measure efficiencies here - if just curious or if you really want to know the status of these PSUs. I recommend you not worry about efficiencies because it is just too complex (and expensive) for the "average consumer" to test for. But I think it would be a good idea to test these old PSUs to make sure they still work.

A PSU Tester can be real handy for this. The advantage of this model is that it has a LCD readout of the voltages. With an actual voltage readout, you have a better chance of detecting a "failing" PSU, or one barely within the required ±5% tolerances (at least with the tester’s internal load) as specified by ATX Form Factor PSU standard. Lesser models use LEDs to indicate the voltage is just within some "range". These are less informative, considerably cheaper, but still useful for detecting PSUs that have already "failed". However, none of these testers test for ripple and other anomalies that affect computer stability. That again, requires sophisticated and expensive test equipment, and the knowlege to use them and properly interpret their results.

Also, these testers only provide a small "dummy load", not a variety of "realistic" loads. So while better than nothing (or a paper clip!), using one of these testers is helpful, but still not a conclusive test.

You can use a multimeter, but again, to get conclusive measurements, the supply needs a proper load. This means it would have to be connected to a working motherboard. This is very inconvenient when testing several PSUs. One of those testers make it easy.

Acceptable Tolerances:

12VDC ±5% = 11.4 to 12.6VDC​

5VDC ±5% = 4.75 to 5.25VDC​

3.3VDC ±5% = 3.14 to 3.47VDC​

 

[delshay]

PSU also has -12v & sb5v. My motherboard is set-up to use the sb5v at all USB port as I need it to charge the mouse or any other device when the computer is turned off.

 

[Grog6]

If you really want to know, you can buy a tester that loads whatever you want to measure; if your wallet is thick enough, go for it.

Or, You can build one out of large resistors, relays to switch loads, and a multichannel data aquisition system to compile all the data.
I could probably put this together out of my junk box in a week or so.

A decent wall outlet based wattmeter that also tracks power factor, hopefully with a usb port for output, to measure the input power will be necessary.

You can probably build all this for ~$1500, if you can write the DataAq software yourself.

The thing is, Do you Want to Know bad enough to do all that?
If the answer is Yes, then go for it; learning how to do a new thing is always good for your brain.
I'll even help with schematics and calculations.


I personally just buy well performing power supplies tested by someone else; the only thing I really care about from a power supply is good power factor, which I can measure with tools I have, and how much power it puts out, which I trust the reviewers for.

I run enough computers that power factor gets important, an uncorrected power supply can be really bad; switching power supplies are a capacitive load, and a bunch of those are not good on your power bill.
I'd pay the extra dollars for PFC.

Don't listen to the luddites that say everything should be left to professionals; a professional is just a guy who read a book and passed a test, you can read the book for the same knowledge.
The most dangerous guy I ever met was a freshly graduated Electrical Engineer; he destroyed most of the lab within a week, lol.
They don't teach anything practical in School these days, just theory.

 

[R-T-B]

uncorrected power supply

They still make those?

Don't listen to the luddites that say everything should be left to professionals; a professional is just a guy who read a book and passed a test, you can read the book for the same knowledge.

Yes, and spend the same $1000 on equipment. This isn't just a degree thing.

 

[Bill_Bright]

Some PSUs have -12V. Many no longer do. That is now optional as seen by the June 2018 ATX Form Factor Design Guide, Table 3-2, Note 4.

Yes, they have +5Vsb, but that, by definition is the "standby voltage" where the power consumed in standby is so minuscule (in comparison to normal power modes) any way (even when charging phones). You would only be talking a couple watts difference here and there depending on the efficiency.

Still, your point is valid. Any output affects, and is affected by the over all efficiency of the device.

Don't listen to the luddites that say everything should be left to professionals; a professional is just a guy who read a book and passed a test, you can read the book for the same knowledge.
The most dangerous guy I ever met was a freshly graduated Electrical Engineer; he destroyed most of the lab within a week, lol.
They don't teach anything practical in School these days, just theory.

This, of course is nonsense. It is naive to suggest schools don't teach practical electronics. Practical electronics is exactly what tech schools, community colleges with AS curriculums, and trade schools teach.

Luddites? LOL. It would seem someone needs to look up the definition of a Luddite.

I will agree with you on one point, however. There is a difference between a formally trained and educated "technician" and a formally trained and educated "engineer". Engineers learn theory. They learn design based on theory. And we need more EEs. But they tend to lack "real-world" experience and an understanding of "applied" electronics - that is, how it really works out in "the field". So I too have seen some newly hatched engineers who didn't seem to have a clue how electronics really worked - except on paper.

At the same time, I have seem many self-taught folks who thought they were God's gift to electronics who didn't have an understanding of Ohm's Law, or electronic safety, for example, who took shortcuts, dismissed out of ignorance accepted practices, and worse, gave bad advice - forgetting completely that anything that plugs into the wall can kill. That is the problem with just about all "hobbyists" and self-taught "experts" - they don't get the well rounded education needed to differentiate between the dime-a dozen-expert and the professional expert. But of course, education and experience alone, regardless how extensive and complete, still does not automatically make one a professional.

FTR, though formally trained, educated, and certified, I am not an electrical engineer.

They still make those?

I am sure they do but I am not familiar with any that are not at least passive. But I note the EU, for example, requires all computer PSUs to have active PFC.

 

[Grog6]

All the AT power supplies I have are uncorrected; I'm pretty sure everything I've bought in the last 10 years is PFC.

Looking at newegg, yes, there's a lot that are not PFC.

It looks like a ~$10 option, comparing similar ones. It will save over $10 in a year, if it's on the 70% power factor end of things.

If businesses here go under 70% power factor, there's a $2000 kicker that gets added on the bill.
It was such a drop in the bucket on the power bill, no one noticed it until it caused other problems, like taking out a transformer switch.

I never saw the whole bill, but it had 60 growth stations rated at 50kW each, lol.

 

[Bill_Bright]

Looking at newegg, yes, there's a lot that are not PFC.

Not sure I'm buying that. If you look at Newegg here and look under PFC on the left panel, you will see 19 are passive, 999+ are active, and 214 are reported to have no PF correction. So to start, clearly most have PFC. However, if you look at those 214 that claim to have none, then visit the maker's webpage, or review site, you will see many really do have PFC. For example, Newegg reports this Gold Certified Raidmax does not have PFC. But if you check out the HardOCP review, it has Active PCF.

Same story with the Antec Earthwatts supply listed on Newegg as No.

All the AT power supplies I have are uncorrected

AT? Did you mean to say AT? AT power supplies were superseded by ATX power supplies almost 25 years ago. I would not expect AT supplies would have it. But for sure, most ATX do, and all do if they are intended to be sold in the EU.

 

[Grog6]

I said AT, I meant AT.

The power supplies I checked were cheap ones, that would not have PFC; If you're advertising 80% efficiency, you're going to add the chokes to do some passive correction.

I got my CET cert in 1984; I got my EE in 2005; I do know how this shit works, as well as a lot of other things, lol.

And as far as "God's gift"; you made me lol on that one, ever read your posts?

 

[John Naylor]

I use these to determine power at the wall for UPS sizing ...

Kill A Watt Meter - Electricity Usage Monitor | P3

On a side note ... had a bit of luck recently ... both good and bad. My son decided he needed an extension cord and removed one in m y office that fed a 550 watt PC speaker system ... and plugged it into the UPS. The puter would draw about 775 watts under stress testing, but more like 670 instantaneous when gaming so was comfy using an old (pre-Schneider Electric) 1000 VA / 700 watt APC unit since 2013. Well when he plugged the 550 watts speaker system in, it was the proverbial straw on the old battery I guess.

In the recent past, we had been using the Cyberpower PR1500LCD 1500VA 54 pound model since APC was taken over by Schneider; but with the price on these rising to $400, took the opportunity to try the half the weight, saller form facctor Cyberpower CP1500 PFCLCD which gives a bit more headroom and battery life. Been installed about a week and this morning I saw something I haven't seen up my way since I was a kid ... "suns howers w/ heavy thunder". Lost power 3 times over the course of 3 hours, each time the battery providing more than enough battery power to last till power returned. All the digital clocks and multimedia equipment are blinking but the two computers in use didn't need to be shut down.

Oh almost forgot ... replaced the battery in the old unit and it's now serving that 2nd computer. One of the other 4 PCs were on as well as a few laptops. The lappies were fine, but my son's PC, w/ just power line surge protection, ... he lost power twice and then decided to take a break till thunder passed.

But the P3 units are a good investment worthy of everyone's tool box. I use them to read line voltages at the wall, but I also test the outputs every year or so to determine any voltage variances for the same reason I redo stress tests every year to see if cooling systems are still working optimally.

And it's not just about "meeting spec". The ATX spec says that in order to maintain ATX compliance, the voltage must be maintained within +/- 5%. Just about any cheap PSU will do this and 5% is more than fine for an office PC. But overclocking is very much affected by voltage stability. The VRMs on your system components have to work much harder with PSU voltage varions of 5% than they do at 2% or 1%. If the user is intersted in moderate overclocking, I will use a PSU that can maintain less than 2% variance. If one is looking for maximum overclocking, then I'll look for less than 1% variance. There are other factors in play here such as how fast it can respond to chnages as well as electrical noise. But as with anything else, your componentry should be chosen to meet the specific needs / goals of each build.

As for "how hard it is to test 80 plus power efficiency? " ...

Here's what ya need .... https://tpucdn.com/review/super-flower-leadex-iii-850-w/images/setup.jpg

But my answer is why bother ? If it's been done, why would I invest my T & E to do it again ?

Super Flower Leadex III 850 W Review

The new Leadex III unit with 850 W capacity has excellent performance and enough cables and connectors to power a potent gaming system. The cherry on top is the silent operation under even tough conditions. While Super Flower continues the good tradition of the Leadex name, it'll be hard to find...

www.techpowerup.com

Here we see that the efficiency of this un it is 91.960% @ 50% load.