Discussion PSU tier list discussion thread

Page 3 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.

Juular

Respectable
BANNED
Mar 14, 2020
1,061
258
1,940
How does the Wentai Aidan end up on tier one, again, as only one listing, when there are 7 different models, and when Aris had THIS to say about the Wentai Aidan 1616.
Good point, will be specified as 1616W. On it's performance, again, nothing of prevents it from being placed in the tier A according to current methodology. It's meh unit but that's all.
 

daris98

Honorable
Jun 8, 2016
8
0
10,510
How does the Wentai Aidan end up on tier one, again, as only one listing, when there are 7 different models, and when Aris had THIS to say about the Wentai Aidan 1616.

  • Increased EMI emissions at low frequencies
  • Short EPS and ATX cables
  • DC Power sequencing has problems (3.3 V rail higher than 5 V)
  • Power-on time is greater than 150 ms
  • The PSU cannot handle the full load with 100 V input and 110% load with 115 V input
  • Weird efficiency drop with 20 W load
  • Noisy operation
  • Low PF with 230 V input
  • Load regulation at 3.3 V and 5VSB not that tight
  • No selectable semi-passive operation
  • No digital communication with the system
That absolutely does not look like Tier A quality or performance to me. There will be others I'm sure once I actually start looking at them more closely.
Aris was writing those cons because he compared it to the AX1600i, which is the direct competitior of the Aidan 1616. Otherwise most of those cons wouldn't be written, like the Vreg was specifically written "not that tight" because it was compared to the AXi, not because it was loose in the grand scheme of things.
 
I am surprised the Vengence Silver is in the top tier and is in a tier above the TXM Gold.

I am having trouble finding a review. Can anyone link one?

Vengence specs
https://www.tomshardware.com/news/corsair-sfx-platinum-vengeance-computex,37314.html

Just looking at the specs, It seems like a unit that would be below the TXM in Corsair's lineup given it has shorter warranty length and 80+, however, I understand that doesn't mean much.

The rifle bearing fan doesn't exactly scream "tier A" to me either. I guess the 50c temp rating is nice, but the TXM has the same rating IIRC.
 

Juular

Respectable
BANNED
Mar 14, 2020
1,061
258
1,940
I am having trouble finding a review. Can anyone link one?
Reviews that are listed in the spreadsheet :
ftp://members.cybenetics.report/PDF_Reports/d/cybenetics_sbN.pdf
The reason why TX-m is positioned lower is because it has ripple higher than limits we set for tier A.
 
Aris was writing those cons because he compared it to the AX1600i, which is the direct competitior of the Aidan 1616. Otherwise most of those cons wouldn't be written, like the Vreg was specifically written "not that tight" because it was compared to the AXi, not because it was loose in the grand scheme of things.
Aris doesn't offer bullet points as a point of comparison between models. Anything he notes would be noted because it simply doesn't meet the expected capability or criteria that he believes a unit SHOULD meet. That criteria doesn't change from review to review, or model to model. I realize there WAS some comparative notation happening there, but it was not the focal point of the cons.
 
Reviews that are listed in the spreadsheet :
ftp://members.cybenetics.report/PDF_Reports/d/cybenetics_sbN.pdf
The reason why TX-m is positioned lower is because it has ripple higher than limits we set for tier A.
I also do not see that review listed on the spreadsheet.
 

Juular

Respectable
BANNED
Mar 14, 2020
1,061
258
1,940
So we put "meh" units on Tier A? 🙄
As long as they qualify for tier requirements - yes. If they have some really major issues we gray them out. We also gild units that can be considered the best ones in the tier.
I also do not see that review listed on the spreadsheet.
cFBOrhg.jpg
 
As long as they qualify for tier requirements - yes. If they have some really major issues we gray them out. We also gild units that can be considered the best ones in the tier.

cFBOrhg.jpg
The list says the Wentai Aiden is a unit "with various issues..." but yet the list recommends it "for high-end systems."

And of course, there are other units marked similarly.

This doesn't make sense at all.
Exactly right. It doesn't. At all.
 
How does the Wentai Aidan end up on tier one, again, as only one listing, when there are 7 different models, and when Aris had THIS to say about the Wentai Aidan 1616.

  • Increased EMI emissions at low frequencies
  • Short EPS and ATX cables
  • DC Power sequencing has problems (3.3 V rail higher than 5 V)
  • Power-on time is greater than 150 ms
  • The PSU cannot handle the full load with 100 V input and 110% load with 115 V input
  • Weird efficiency drop with 20 W load
  • Noisy operation
  • Low PF with 230 V input
  • Load regulation at 3.3 V and 5VSB not that tight
  • No selectable semi-passive operation
  • No digital communication with the system
That absolutely does not look like Tier A quality or performance to me. There will be others I'm sure once I actually start looking at them more closely.

I agree.

Besides... you can't actually buy the PSU. They're looking for someone to relabel it and sell it.

I've considered it, but it's not even HX quality, but for twice the money. GaN or no GaN, it's a meh product.
 
  • Like
Reactions: ProfSeaGoat

Juular

Respectable
BANNED
Mar 14, 2020
1,061
258
1,940
The list says the Wentai Aiden is a unit "with various issues..." but yet the list recommends it "for high-end systems."
It's still a better unit than pretty much anything from lower tiers. We could gray it out ... for something but do you have that something in mind ? All issues Aris were talking about are minor. If we were to gray out units for minor issues even stuff like Seasonic Focus would end up there (Gold SKUs, even renewed ones have rather meh 3.3V rail transient response for example).
I've considered it, but it's not even HX quality, but for twice the money. GaN or no GaN, it's a meh product.
Corsair HX for that matter is gold color, i.e it's recommended over just white units like Wentai Aidan 1616. And Wentai Aidan isn't merely because it doesn't fit 30mV ripple requirements on 110% load, should've they marketed it as 1.3kW unit which it really is it would've been gilded. And again, we don't tier cost either, that's up to consumer to decide.
 
Reviews that are listed in the spreadsheet :
ftp://members.cybenetics.report/PDF_Reports/d/cybenetics_sbN.pdf
The reason why TX-m is positioned lower is because it has ripple higher than limits we set for tier A.

That doesn't make sense. TX-M is better than Vengeance in every aspect, but since the ripple is higher, but well within specification, it's a tier below.

See... here's why I don't like tier lists.
 
  • Like
Reactions: ProfSeaGoat

Juular

Respectable
BANNED
Mar 14, 2020
1,061
258
1,940
TX-M is better than Vengeance in every aspect
Care to elaborate in which aspects it's better then ? And why these aspects are better than ripple which may be well within specifications but it's obviously a good thing to have ripple as low as possible when all other performance metrics are satisfactory ? The only thing we don't have data on Vengeance Silver is transient response, i don't know who to blame on this, Aris or Corsair but Aris didn't publish a review for it with this data, only a report which doesn't include that.
 

Juular

Respectable
BANNED
Mar 14, 2020
1,061
258
1,940
No. I don't. Because it's not worth my time.
You ain't helping ...
You said that TX-m is better than Vengeance Silver in 'almost all aspects'. We however do know from reviews that it (TX-M) :
  1. it has higher ripple, higher than all tier A units for that matter
  2. it's more noisy
  3. Edit: It's single rail in comparison with selectable mode on Vengeance Silver
And ... that it? In other aspects they are very close to the point of no difference. Voltage regulation are good enough. Both are modern LLC + DC-DC units. Both use full Japanese electrolytics with side of polymers. Both are manufactured by fairly good OEMs, sure, you can hear people saying that both GW and HEC are 'meh' OEMs but i'm sure you're not one of them. Transient response on TX-m isn't particularly fantastic and we lack info on Vengeance Silver although we currently don't have transient response in the methodology anyway. What else ... protections ? I'm fairly sure that Vengeance Silver have passed all protection tests by Aris, but again, we don't have that info and Corsair claims entire protections set anyway. Efficiency ? Come on. As a result of better efficiency it possibly runs cooler and thus has better components longevity ? Perhaps, barely. But it has 5 years warranty which is long enough to consider it being a good choice.
So in which aspects TX-m is actually better ?
But if you don't have a time to actually elaborate on your claim i understand that, although you should understand that we can't take your word as a given even with all due respect, this tier list isn't based on words but hard data.
 
Last edited:
You ain't helping ...
You said that TX-m is better than Vengeance Silver in 'almost all aspects'. We however do know from reviews that it (TX-M) :
  1. it has higher ripple, higher than all tier A units for that matter
  2. it's more noisy
  3. Edit: It's single rail in comparison with selectable mode on Vengeance Silver

I have an entire PRD for both TX-M and Vengeance Silver that outlines what components should be used, what margins there needs to be for efficiency, voltage regulation, etc.

Yes. Vengeance has a single/multiple +12V rail switch. Who does that matter to? Absolutely nobody because hyper-marketing has convinced people it's not necessary.

Yes. The TX-M is loud because of an air gap between the fan frame and the housing.

But the regulation and component selection inside the TX-M is superior to Vengeance. Which is why Corsair gives the TX-M a 7 year warranty instead of 5 years.

You do realize that Vengeance is only rated at 40°C while the TX-M is rated at 50°C, right?

PSU reviews, including Aris's, simply are not that thorough. It's like the whole CWT GPU vs. GPX. I can leverage the GPX platform and save $10 on the BOM, and although Aris might say it passes all of his tests and is a better value proposition than anything based off the GPU platform, I wouldn't feel confident using it because of the cheap components used inside. I literally have a lab full of GPX's that have blown up for a number of reasons, yet reviews are saying "it's good enough".
 

Juular

Respectable
BANNED
Mar 14, 2020
1,061
258
1,940
I have an entire PRD for both TX-M and Vengeance Silver that outlines what components should be used, what margins there needs to be for efficiency, voltage regulation, etc.
I understand that and i kinda expected that this would be your answer. But how do we determine that looking at the data we have access to ? We can't. And can't tier PSUs only on the word even if it's the head of RND department for it's brand.
PSU reviews, including Aris's, simply are not that thorough.
I understand that too. Perhaps you meant not 'not thorough' because Aris reviews are the most thorough reviews out there, but that Aris doesn't have big enough sample size to account for variations between batches and catch all issues. But even if he did, there are simply not enough Aris Mpitzoupulos to test them all, from his words, he has a queue of units to test for years ... So again, we operate with the data we have.
 
I don't mean to challenge you, but most of the documentation i have seen says the vengence is rated at 50c, such as Corsairs own web site:
https://ibb.co/S3CWFcr
https://www.corsair.com/us/en/Categ...ngeance-Series/p/CP-9020175-NA#tab-tech-specs
Or this toms announcement
https://www.tomshardware.com/news/corsair-sfx-platinum-vengeance-computex,37314.html

Challenge away! It gives me the opportunity to tell marketing they screwed up and put the wrong thing on the website. :D

From section 4.2.2 of the Vengeance Silver PRD:

DC Output Power

450. Continuous DC Output Power: 450W at 40 degree Celsius.

550. Continuous DC Output Power: 550W at 40 degree Celsius.

650. Continuous DC Output Power: 650W at 40 degree Celsius.

750. Continuous DC Output Power: 750W at 40 degree Celsius.

Peak DC Output Power: 40 degree Celsius, 115Vac 60Hz, 230Vac 50Hz. 17 Seconds maximum, one occurrence maximum per minute.
 
  • Like
Reactions: Juular
I understand that and i kinda expected that this would be your answer. But how do we determine that looking at the data we have access to ? We can't. And can't tier PSUs only on the word even if it's the head of RND department for it's brand.

Except you CAN, because you just GOT the damn information from the only-est person in the world you COULD possibly take their word for it on without seeing it expressly spoon fed to you in a review. If you can't take the word of the guy from research and development for the company that sells BOTH models, then there isn't really anybody at all you can take their word for it from, even if they reviewed the unit. It's one thing to not take my word for something, or even entertain suggestions for changes on the list, but when you have to ignore the person who is one of maybe five or ten people in the world who belongs in a room having a conversation with the other 9 people who can discuss power supplies at that level of conversation, then you're just being a rotten onion by design.
 
I understand that and i kinda expected that this would be your answer. But how do we determine that looking at the data we have access to ? We can't. And can't tier PSUs only on the word even if it's the head of RND department for it's brand.

I understand that too. Perhaps you meant not 'not thorough' because Aris reviews are the most thorough reviews out there, but that Aris doesn't have big enough sample size to account for variations between batches and catch all issues. But even if he did, there are simply not enough Aris Mpitzoupulos to test them all, from his words, he has a queue of units to test for years ... So again, we operate with the data we have.

No. I mean not thorough as in doesn't test as many data points as Corsair does. I've blown up a lot of the PSUs that Aris has passed. Heck.... Jeremy was even worse. He'd give 8 and 9 scores to PSUs that couldn't even pass half the tests we throw at them.

I still use Aris's Cybenetics beta testing services through all stages of R&D, though. Because having an extra set of eyes on a project is always a good thing. But it's the same reason I have Allion Labs testing products through all stages in parallel with the OEM and Corsair itself. If you rely on one person to do all of the testing, you're going to miss things.

And it is good that Aris does a component break down. More people should look at those. For example: If you have two PSUs of the same 850W wattage, both Gold and they both seem to perform the same, but one has a 800V 20A bridge diode and the other has two, which one is better? If one has four 60V 140A MOSFETs for synchronous rectification and the other has six, which one is better?
 
Last edited:

Juular

Respectable
BANNED
Mar 14, 2020
1,061
258
1,940
No. I mean not thorough as in doesn't test as many data points as Corsair does.
If only we had access to Corsair data.
If one has four 60V 140A MOSFETs for synchronous rectification and the other has six, which one is better?
On paper, yes. One is better. But just as you're saying that PSU isn't just a box of capacitors, i'm sure you can say the same about MOSFETs. It's not as simple as choosing a PSU with highest rated components. And either way, how you're gonna translate that into methodology ? It's entire design that matters and reviews are closest thing we have to showing the result.
Except you CAN, because you just GOT the damn information from the only-est person in the world you COULD possibly take their word for it
It's good to have a person to provide such data for Corsair PSUs for sure, and we could perhaps ignore results we see for TX-m in the reviews and tier it back just on the basis that Jon says so. But how about other brands ? We don't have a person as open for discussion as Jon for any of other brands. And even if we suddenly did, imagine, comes Thermaltake contact and says that PSU X is better than Y, while in reviews we clearly see that Y performs better than X which puts them in separate tiers, and we have no idea who that person is, how strict their testing process (even if i don't exactly know that for Corsair either but we trust Jon and he gave an impression that it's very strict). So what do we do ? Make an exception only for Corsair ? That doesn't sound neither fair nor unbiased to me.
 
Status
Not open for further replies.