Is 80 PLUS Broken? How To Make It A More Trustworthy Certification

[Sakkura]

Spoiler

I could not possibly care less how efficient a PSU is; what I care about is how reliable it is.

Actrally it does matter...That saved power that save money on your power bill would allow you to buy more computer parts if you missing like few bucks or few cents that power saved PSU help you save money.

I would recommend punching in the numbers based on local electricity costs and your rig/amount of use and seeing how much this actually adds up to.

For example, if I take a 500W load (which is higher than most single GPU systems would ever draw even under max load), look at the worst case efficiency difference between 80+ and 80+ Titanium (14% better effeciency at 50% load), and say I'm running it 40 hours a week, every week of the year, it works out to a whopping ~$7.50 a year where I live. Admittedly, my local rates are much lower than a lot of people pay (~$0.05//kWh), but on the other hand the values I used for power consumption are definitely on the high side.

I think a lot of people overestimate the amount of money they're going to save with an efficient PSU.

If I run a more realistic average 300W load on a 600W PSU for 40 hours per week, then the difference between 80+ and 80+ Titanium over a year would be about 84 kWh. That would cost me about $24.

Call it a 5 year service life and that's total savings of $120. Pretty substantial.

I don't think it's enough to make 80+ Titanium worth it though, since the price premium for Titanium is much bigger than for Platinum or Gold. These days you can get a Gold rated unit for just a small premium over Bronze or 80+, so that's strong value for money.

But even if we assume you have a high end graphics card (system power draw of 300 W) and game 40 hours a week, those numbers would still be on the high side as neither 80+ or 80+ titanium are common. If we look the much more common Bronze vs Gold, difference in efficiency is now down to 5%. So your yearly cost difference is about $8.60, or ~$43 over 5 years.
I'm not trying to argue that efficiency is irrelevant in terms of utility bills, just saying that it depends on a lot of factors and should not be assumed to make a meaningful impact.

Your numbers are off. The difference between Bronze and Gold is 5.9% in the US and 4.5% in the EU, at 50% load. It's higher at the 20% and 100% load levels.

Besides, 80+ standard is common enough.

 

[TJ Hooker]

Your numbers are off. The difference between Bronze and Gold is 5.9% in the US and 4.5% in the EU, at 50% load. It's higher at the 20% and 100% load levels.

Besides, 80+ standard is common enough.

Where are you getting your numbers from? According to both Wikipedia and the table in this article, Bronze vs gold is 5% different at 20, 50, and 100% load for 115V, and 4% for 230V (non redundant).
80+ may not be rare, but there are far more 80+ Bronze, at least going off Newegg/PCPartPicker. Also, I don't think I've ever seen someone recommend an 80+ PSU, while there are a number of good 80+ Bronze units I see being recommended on a regular basis.

 

[Sakkura]

Your numbers are off. The difference between Bronze and Gold is 5.9% in the US and 4.5% in the EU, at 50% load. It's higher at the 20% and 100% load levels.

Besides, 80+ standard is common enough.

Where are you getting your numbers from? According to both Wikipedia and the table in this article, Bronze vs gold is 5% different at 20, 50, and 100% load for 115V, and 4% for 230V (non redundant).
80+ may not be rare, but there are far more 80+ Bronze, at least going off Newegg/PCPartPicker. Also, I don't think I've ever seen someone recommend an 80+ PSU, while there are a number of good 80+ Bronze units I see being recommended on a regular basis.

You can't use those numbers for that kind of comparison - they're percentage points, not percentages. You have to convert them to relative terms. If the system is drawing 300W from a 600W power supply, and it's Bronze rated, that means the PSU is drawing 340.9W in Europe or 352.9W in the US. If it's Gold rated instead, power draw drops to 326.1W (EU) or 333.3W (US).

340.9W is 4.5% more than 326.1W and 352.9W is 5.9% more than 333.3W.

Or in unitless terms; (1/0.88)/(1/0.92) - 1 = 4.5% rather than the 4% you'd think. And (1/0.85)/(1/0.90) - 1 = 5.9% rather than 5%.

Plenty of people buy 80+ PSUs. The EVGA W1 units for example.

 

[TJ Hooker]

You can't use those numbers for that kind of comparison - they're percentage points, not percentages. You have to convert them to relative terms. If the system is drawing 300W from a 600W power supply, and it's Bronze rated, that means the PSU is drawing 340.9W in Europe or 352.9W in the US. If it's Gold rated instead, power draw drops to 326.1W (EU) or 333.3W (US).

340.9W is 4.5% more than 326.1W and 352.9W is 5.9% more than 333.3W.

Or in unitless terms; (1/0.88)/(1/0.92) - 1 = 4.5% rather than the 4% you'd think. And (1/0.85)/(1/0.90) - 1 = 5.9% rather than 5%.

Ah yes, you're right, I was calculating it wrong. The way I was doing it would only apply if you assumed a constant power at the wall resulting in varying power delivered based on efficiency, rather than constant system power draw and varying power at the wall. To find difference in power consumption at the wall between Bronze and Gold for 50% load, I should have done (1/0.85)-(1/0.9) = 6.5%
300x0.065 = 19.6 = 352.9 - 333.3, which matches your calculations.

 

[Onus]

You can check on thorough PSU reviews, like the ones we do here at Tom's.

Don't get me wrong, 80 PLUS remains of high importance and in general everyone should stay away from PSUs that haven't been evaluated by an independent lab, however as everything in this life it also has its deficiencies which we discuss in this article.

Any brand/OEM that is willing to send a product for evaluation to any independent lab voluntarily, shows that is sure about its product and doesn't have anything to hide. Nonetheless, still the evaluation process should be proper in order to expose possible golden samples and provide a full and accurate data report. If the evaluation process is flown or semi-complete then it can easily be fooled and on top of that it won't reveal the whole truth about a product's performance.

This. Good reviews expose the details, but I continue to rely on 80plus for a simple junk/not junk rating. If it wasn't submitted, chances are it wouldn't pass, and is probably junk. If it is certified, it was run at 100% for at least half an hour; no PSU-shaped object could withstand that, even at only 23C ambient. Competent reviews will still expose the turds-in-a-box (e.g. HardOCP's review of the Coolermaster GX), but even that one isn't going to kill your system the way an Apevia, Logisys, or Diablotek might.

 

[AldoGG]

I applaud this article!

At least we should ask for the next certification:

1.- Increase the measurements to 7: levels 5%,10%,20%,40%,50u%,100%
2.- Test at 32c +- 2 degrees (Ambien temperature on some places in the world is that all the year)
3.- Include 5VSB Efficiency too
4.- Certification needs to reapply if any component changes

Lets cal this new certification 80 PLUS+, 80 PLUS+ Bronze ...

 

[krb1945]

I notice some posters seem to be saying you can tell the quality of a PS by the length of the warranty, and they may or may not be correct. I see some manufacturers place length of warranty right on the device, 5 years, 7 years etc. The problem is anyone can state a warranty period, but that does not mean they are going to honor the warranty.

 

[Juventas]

I can't agree. Until 80 PLUS became common blue screen errors due to dirty ass power being delivered to components was the norm. Even so called premium name brand PSUs suffered from this problem.

Yep. 80 PLUS might not be perfect, but it's done an excellent job at creating a line between what's junk and what's not. Premium brands like Antec and Thermaltake were selling models for years that couldn't meet 80 PLUS standard and failed like crazy. Lesser-known names like Fortron (FSP) and Power Man (In Win) had models that were better and cheaper and easily got 80 PLUS.

I'm not a system builder anymore, but when someone asks me for buying advice for a basic power supply, I don't recommend brands, I recommend 80 PLUS.

 

[Juventas]

Aris, thank you for this article!

Desktop sales are declining and a growing portion of those remaining use external AC-DC power supplies (all-in-ones, mini-itx/stx, etc.) in the same way that laptops do. I think the next direction 80 PLUS needs to go is certifying external power supplies. I realize these are relatively simple, but I've read their efficiency can differ dramatically as well.

 

[Juventas]

Lets cal this new certification 80 PLUS+, 80 PLUS+ Bronze ...

Plus-plus? This was meant as a joke, right...

 

[FancyAvatar]

Please, forgive me for the small OT but i need to ask Tom's staff a question.

In the next few weeks, are you going to review the 550w/650w unit of the new Evga G3 power supply ?

Thank you.

By the way, your PSUs reviews are a pleasure to read and very instructive too.

 

[WyomingKnott]

It's almost as though you are inventing things we don't ...

How did you get that name? Are the other 25 letters taken? Did I even Try, oh, I'll read your post in a second, I never met a person with a one digit user name before. I'm star struck.

It's kind of clever that when you quote him/her the heading is "I said."

 

[Aris_Mp]

@FANCYAVATAR I have two G3 samples but of higher capacity, which I plan soon to schedule for a review.

 

[Olle P]

I can't agree. ... I applaud this article!

Very confusing comment...

10 Years ago 80+ Bronze was a big push, now Gold is becoming mainstream. ...

I wouldn't consider the Gold rating "mainstream". Gold rated PSUs are still much higher priced than those with a basic or Bronze rating.

The only big issue is the fakes that come primarily out of China. ...

While I agree that this is the main issue (in combination with the fraud not being enforced) we have to realise that the vast majority of PSUs are made in China.

I could not possibly care less how efficient a PSU is; what I care about is how reliable it is.

I would recommend punching in the numbers...
... look at the worst case efficiency difference between 80+ and 80+ Titanium...

If you couldn't care less, why be so narrow-minded?
Compare instead a very low efficiency (~40%) PSU to basic 80+ efficiency.
500W @ 80% efficiency means 125W loss.
500W @ 40% efficiency means 750W loss.
To me that's definitely something to care about!

I do agree that the difference between 80+ and 80+ Titanium is rarely worth the price premium, but that's way off what this article is about.

... I continue to rely on 80plus for a simple junk/not junk rating.

I totally agree. For the vast majority of low cost (<$60) PSUs the 80+ stamp is the only indication whether it's junk or functional.
For me brownouts and power surges are not an issue. I want ripple and electric noise to be within specs. Efficiency must be "reasonable". I don't want to spend extra money to get better specs.
Very few reviewers provide extensive cover of that niche!

Now for some comments on the article itself:
1.Small number of measurements in order to classify a PSU into one of its categories.
Agreed! Most PSUs do in real life operate mostly at <50% load. We need more measuring points there. For high power consumer PSUs it's reasonable to have some fixed wattages up to 100W where measurements are done.

2.The methodology allows for manufacturers to submit golden samples.
Additional annual tests on a random production sample should be part of keeping the certificate.

3.Very low ambient temperature at which the measurements are conducted.
Agreed. ~30C should be the normal test temperature, plus one 1h test running 100% load at the maximum operating temperature defined by the manufacturer.

4.It doesn’t measure the standby power consumption ...
That's a valid point!

5.It doesn’t take into account the 5VSB rail’s efficiency.
Also a valid point!

6.There is no clear mention of testing equipment.
I oppose a demand for specific make/model of equipment, but there should a) be some specified minimum requirements for all test equipment used, and b) the test equipment used for a certification should be listed in the test protocol, so that it can be identified later on if required.

7.80 PLUS doesn’t effectively deal with fake efficiency badges.
This is really bad! While individual markings for each certificate would be great I think we could come some way in the right direction by having the badge identify the test lab used, much like there's a four digit number identifying a responsible "notified body" when applicable to CE markings.

 

[FancyAvatar]

@FANCYAVATAR I have two G3 samples but of higher capacity, which I plan soon to schedule for a review.

Awesome.
Thank you Aris

 

[Lutfij]

Hmmm, now we have to second guess the standard and pretty much all the PSU's out there. Well if anything the truth will/shall be revealed and we can breath a deeper sigh of relief...until then I'm going to go along the badge.

:/

 

[prashant393]

Many people think that a psu is not an important part of their rig but it is... after all we dont want our $1000 rigs blowing 100 fee up in the air... it would be a great fireworks show though...XD

 

[Lutfij]

^ That line looks like it was quoted from another post...