TechPowerUp is Hiring a Power Supply (PSU) Reviewer

[boomheadshot8]

sadly, i have poor qualifications in electronics
but it's interesting

 

[pk67]

The goal is NOT to have reviews that look extremely technical, super complicated, that you don't understand, to make you assume "they are doing the right thing".

Seems like you are looking more for marketing poetry writer than pure technical reviewer.
Is T-T(ambient) = f(Load) where Load = Paverage/Pmax.nominal is super complicated for you ?

I'm asking cos for avearge reader temeperature of hotspots of critical parts like electrolytic capacitors is meaningless i'm sure but from designer perspective such details are crucial for the expected lifespan of final product (at given conditions at least)

I also don't want people that just blindly follow a test protocol, I can automate that, and feed the numbers into ChatGPT to generate the review text, no reviewer needed.

Without extensive knowledge of reliability of used components in tested PSU review would be still based on blind taste of reviewer than on hard facts related to big statistics ( which are hard to understand and evaluation for average reader without strong technical background ).

 

[tpuuser256]

Wish I had the space and time for this... Good luck !

 

[jonnyGURU]

Could you share you vision of useful PSU review format, because, at current form, 99% of them fail to deliver useful data that actually matters for PC builders. We are too much into wanking around useless numbers, basically if a PSU fails to meet some basic ATX specs, it should be banned either way, clients do not need to think about specifics, they do not understand either way, like it is 1970ies esoteric analog tech magazine, but the useful question will my PC with CPU X and GPU Y work without issues, without compromises on protections, imho are being too lax lately, that is a different topic. The spikey mixed load current behaviour is not that reviewed with simulated loads, I fail to see how on earth some more badges can help buyer to answer if my PSU enough. Who says that we actually need so sophisticated supplies, considering the on board VRM power delivery has gotten better itself, not needing so many aspects the reviews cover. 90% of office PC using cheap and cheery OEM PSUs work, the overkill gamer logic stigma, it ventures into stupid... well yeah... money, I understand.

I don't necessarily agree with you. Sure, a PSU should be a pass/fail. It should either pass all of the tests or not. But the reviews are typically done very quickly, with one unit and often incorrectly. When I test a PSU after an EVT, DVT, PVT, I test five units on the ATE. At 90V and 264V. At 25°C and 50°C. One unit typically does fail and when that happens, everything stops until a failure analysis is done. There are thermal tests and PC compatibility tests. Units are sent to Intel for validation, Cybenetics, PAL labs, etc. Nothing moves to the next phase until every unit passes. This is why it takes me almost a year to put out a new PSU. I have failed SO MANY PSUs that reviewers have passed and slapped "recommended" tags on, it just kills me.

It should be a very objective process. But sadly, it ends up being quite subjective. There are "hard and fast" rules to PSU testing. Sadly, too often opinions get in the way.

"The caps aren't Japanese" You do realize those Japanese brand caps are made in the same Chinese factory as the ones you "don't like" and know nothing about, right? "There is only 100mm between the SATA connectors" So? "The exhaust temperatures are very high" Yet the efficiency isn't impacted. This only means that the airflow through the PSU is good. Not that the PSU is bad. And then, as I stated earlier, the testing just isn't thorough enough.

And then there's reviewers with 100K worth of equipment that don't know how to use it.

Unfortunately, once you're on the inside for over a decade like I have been, you become very jaded. I learn something new every week. Even Aris is learning as he goes and will eventually get to the point where he won't want to review PSUs anymore. I know I don't. And Oklahoma Wolf said the same thing.

If you want to "keep it simple stupid", you could do all of the tests and ONLY point out where the PSU failed. At the end of the day, that's what a lot of people do when they jump to the conclusion page of a review. And while this may work for most, I still feel a lot of information is lost not only on the end user... but also the reviewer!

 

[pk67]

"The exhaust temperatures are very high" Yet the efficiency isn't impacted. This only means that the airflow through the PSU is good. Not that the PSU is bad. And then, as I stated earlier, the testing just isn't thorough enough.

I wouldn't said exhaust temperatures doesnt matter until efficiency isnt impacted. It is shortsighted approach imho. Cos electrolytic capacitors lifespan strongly depend on temperature. I would prefer louder PSU but with cooler exhaust than silent or in idle/turn off fan mode but way more hotter.
Efficiency is easy to measure but expected real lifespan is a lot way harder. If protections will act as designed than only caps will have to be refreshed if they dry out prematurely for the sake of high temperatures. Sometimes high temperature + el. caps could work as a timed bomb in PSU.

 

[jonnyGURU]

I wouldn't said exhaust temperatures doesnt matter until efficiency isnt impacted. It is shortsighted approach imho. Cos electrolytic capacitors lifespan strongly depend on temperature. I would prefer louder PSU but with cooler exhaust than silent or in idle/turn off fan mode but way more hotter.
Efficiency is easy to measure but expected real lifespan is a lot way harder. If protections will act as designed than only caps will have to be refreshed if they dry out prematurely for the sake of high temperatures. Sometimes high temperature + el. caps could work as a timed bomb in PSU.

See, this is a perpetuation of a misconception.

AC not converted to DC is converted to heat. If the PSU is more efficient, it generates less heat. If you have two PSUs with the same efficiency at the same load and one has hotter exhaust than the other, that is not because on PSU is running hotter than the other. It's because it's better at moving the heat AWAY from the PSU. Whether it's because of the fan design or PCB layout.

When I was at BFG, we designed a PSU with the components set perpendicular to the main PCB as opposed to putting them on the main CPB itself. This allowed the air to flow across the components instead of just "pushing air down" onto the tops of the components. The results was very low internal temperatures, but higher exhaust temperatures. Why? Well... the heat has to go somewhere and if it isn't, then you're cooking your components!

So what does Paul at HardOCP say? "Con: Exhaust temperatures very high."

THIS is how you do a proper thermal test:

It's NOT done by just measuring the delta between intake and exhaust temperatures.

 

[chrcoluk]

I would expect a PSU review to consist of two things right?

Validating the components inside, so this needs the PSU opened up and examined, might need some knowledge of these type of components, otherwise will spend hours researching components.

The other bit is the testing, and usually this requires specialist equipment, to test different loads, and taking specific measurements?

Ultimately the combination of the two then allows the reviewer to provide a educated assumption of the likelihood of the PSU lasting a long time and risk of damaging the PC.

 

[Shrek]

Why is the pay never mentioned?

 

[Cheeseball]

When I was at BFG, we designed a PSU with the components set perpendicular to the main PCB as opposed to putting them on the main CPB itself. This allowed the air to flow across the components instead of just "pushing air down" onto the tops of the components. The results was very low internal temperatures, but higher exhaust temperatures. Why? Well... the heat has to go somewhere and if it isn't, then you're cooking your components!

That was that 1000W model from 2007 right? Hot exhaust but lasted so long that I used that shit until 2014. Good times.

 

[Airisom]

OklahomaWolf pleeeaaasseeeee

Loved the death of a gutless wonder reviews on jonnyguru

 

[pk67]

See, this is a perpetuation of a misconception.
...
When I was at BFG, we designed a PSU with the components set perpendicular to the main PCB as opposed to putting them on the main CPB itself. This allowed the air to flow across the components instead of just "pushing air down" onto the tops of the components. The results was very low internal temperatures, but higher exhaust temperatures. Why? Well... the heat has to go somewhere and if it isn't, then you're cooking your components!

So what does Paul at HardOCP say? "Con: Exhaust temperatures very high."

THIS is how you do a proper thermal test:

It's NOT done by just measuring the delta between intake and exhaust temperatures.

I see your point - in some pathological cases significant amount of heat still bypassing exhaust flow lowering exhaust temperature. As are result we could come to missleading conclusions.
But my point was quite different in fact.
Low exhaust temeperature - still need further investigation but high exhaust temperature would immediately disqualifying some PSU as are result of too low air flow.
I wouldnt agree with you - that you are measuring temperature in most important places at your picture.
Right designed HF transformer is way less prone for aging than el.caps. So located at right places temperature sensors and isolated from impact of direct cold airflow are important for sure. But I dont see sensors close to el. caps , so even what you are measuring at your picture is not a perfect but way better than raw conclussions based just on delta T at exhaust airflow.
But keep in mind high delta T at exhaust can say as something alone too cos delta T of hottest components must be even higher than that. So for sure it is worth of closer inspection and placing sensors at parts most prone for aging like el. caps. Healthy caps are not sources a lot of heat but they are prone for aging by drying out so checking thier temperatures is important and can say as a lot about possible lifespan of PSU.

 

[Shrek]

High exhaust temperatures don't bother me if the capacitors are situated at the inlet.

 

[jonnyGURU]

I see your point - in some pathological cases significant amount of heat still bypassing exhaust flow lowering exhaust temperature. As are result we could come to missleading conclusions.
But my point was quite different in fact.
Low exhaust temeperature - still need further investigation but high exhaust temperature would immediately disqualifying some PSU as are result of too low air flow.
I wouldnt agree with you - that you are measuring temperature in most important places at your picture.
Right designed HF transformer is way less prone for aging than el.caps. So located at right places temperature sensors and isolated from impact of direct cold airflow are important for sure. But I dont see sensors close to el. caps , so even what you are measuring at your picture is not a perfect but way better than raw conclussions based just on delta T at exhaust airflow.
But keep in mind high delta T at exhaust can say as something alone too cos delta T of hottest components must be even higher than that. So for sure it is worth of closer inspection and placing sensors at parts most prone for aging like el. caps. Healthy caps are not sources a lot of heat but they are prone for aging by drying out so checking thier temperatures is important and can say as a lot about possible lifespan of PSU.

Why are you assuming that only some parts are tested and not others?

Here is an example of one of my thermal test reports:

And yes... some components, like capacitors, can become hot by proximity. But that has nothing to do with high exhaust temperatures being a negative. If your PSU is creating X amount of heat and that heat is making it's way out of the exhaust grill, you're doing it right.

Note that some components are as hot as 117.8°C (bridge diodes at low mains are nasty space heaters). Naturally, your exhaust temps aren't going to be 117.8 (in this case, they were 60°C), but if you have parts creating that much heat, it needs to go somewhere. Preferably out of the back of the PSU.

High exhaust temperatures don't bother me if the capacitors are situated at the inlet.

Inlet? Are we talking an old school PSU with two 80mm fans now?

BTW: The design is and always will be best for keeping the PSU cool. Sadly, low noise won that battle.

 

[pk67]

Why are you assuming that only some parts are tested and not others?

Here is an example of one of my thermal test reports:
View attachment 370943

And yes... some components, like capacitors, can become hot by proximity. But that has nothing to do with high exhaust temperatures being a negative. If your PSU is creating X amount of heat and that heat is making it's way out of the exhaust grill, you're doing it right.

Note that some components are as hot as 117.8°C (bridge diodes at low mains are nasty space heaters). Naturally, your exhaust temps aren't going to be 117.8 (in this case, they were 60°C), but if you have parts creating that much heat, it needs to go somewhere. Preferably out of the back of the PSU.

Yes you are right i assumed you are not measuring el. caps temperature cos i dont see sensors on them at your picture - but i could be wrong cos you could add sensors at holes in pcb under the caps.
But your report still missing two important params ( for me at least ) Pout and T ambient.
Hot spots at bridges telling us this one design asking for some changes. Cos FR4 laminate is aging too and will not last forever imho.
But generally speaking I like your detailed approach.

Such hot spots for 40 bucks PSU would be acceptable, but for 200-250 bucks PSU i would expect and demand something more sophisticated design (heat spreading from hotspots).

 

[jonnyGURU]

But your report still missing two important params ( for me at least ) Pout and T ambient.

750W. And I did state the exhaust temp was 60°C. And the report says that the chamber temperature is 40°C. But, for the record, the actual temp measured at a thermistor at the intake was reading 40.10°C.

Hot spots at bridges telling us this one design asking for some changes.

Welp. Sounds like you're the next PSU reviewer for TPU. You can have my job too.

BDs typically get this hot, around 75 to 80% within spec. They are, without a doubt, the hottest part of an SMPS. You should know that. If you really think this is just a $40 BOM, that's hysterical. Yes, a bridgeless design will bring these temps down to around 75 to 80°C, but will also bring the cost up another $8 to $10. Oops.... I guess that would be a $50 BOM.

 

[Shrek]

I thought the main high frequency transformer could run at over 100°C

 

[pk67]

BDs typically get this hot, around 75 to 80% within spec. They are, without a doubt, the hottest part of an SMPS. You should know that. If you really think this is just a $40 BOM, that's hysterical. Yes, a bridgeless design will bring these temps down to around 75 to 80°C, but will also bring the cost up another $8 to $10. Oops.... I guess that would be a $50 BOM.

Personally I would like to spend even 20 bucks more to avoid such hot spots.
For 100% Load and at 40degC ambient it is not bad. I would be more interested in report at 60% load at the same ambient Temp.
I'm not saying it is bad design or extremly cheap but even you admitted that it could be better at bit higher BOM.
I'm not saying report at 100% load is meaningless for me. It is usefull as a point of reference - but for practical reasons I assume 60% load as an upper boundary of practical load range for such PSU.

I thought the main high frequency transformer could run at over 100°C

Yes it could for a while. But wounding isolation is aging too at high Temp levels - so definitely it is better to avoid them by right design.

edit
Along with time people will looking for more powerfull and platinium certified at least PSU-s. Progress in power semiconductors will allow for introduction even higher rank(s) than titanium. So demanding and aware users will be even more demanding. Spending 10 or 20 bucks more to slower aging of 200 or 300 bucks expensive PSU is not a bad option imho. Greedy users or unaware will still be looking for PSU at lowest possible price.

 

[Pizdarenkowitch]

Try to contact this man:

AllThingsOnePlace

Videos of reviews, vintage electronics, teardowns, failure analysis, acoustics tutorials, project builds, and more. All Things One Place

www.youtube.com

 

[jonnyGURU]

I thought the main high frequency transformer could run at over 100°C

I would be more concerned with the main transformer running over 100°C than the BD.

Here is an example of a unit we just failed because the T2 Tj/TjMax is at 73.2% and the L2 is at 73.8%. This test was run at 50°C because it's a higher end product with higher expectations:

To keep this on topic: 99% of reviewers would pass this. Why? Because, did it fail on the ATE? No. Will it fail over time? Probably not. But you don't want to knowingly put out something that can potentially fail.

I'm not saying it is bad design or extremly cheap but even you admitted that it could be better at bit higher BOM.

Sadly, desktop computer power supplies have been a commodity market for a while. You can build a better product that will have .001% failure rate versus 1% failure rate, but if your elevator pitch doesn't have enough marketing buzzwords to make it sound worth the extra money, people aren't going to buy it. And if the "cheaper solution" results in 99% of units not failing within warranty period, who is complaining? Or rather, who is listening to the 1% that is complaining?

 

[pk67]

Sadly, desktop computer power supplies have been a commodity market for a while. You can build a better product that will have .001% failure rate versus 1% failure rate, but if your elevator pitch doesn't have enough marketing buzzwords to make it sound worth the extra money, people aren't going to buy it. And if the "cheaper solution" results in 99% of units not failing within warranty period, who is complaining? Or rather, who is listening to the 1% that is complaining?

I think in platinium rated PSU division it is way lower probability to find out hot spots than in gold certified.
Even hotspots of higher power rated PSU-s are meaningless at 100% load level cos for practical reasons real load will be higher than 50% in very rare cases .
But if I had an option to pick up in similiar price the same efficiency rated models but one with more evident hotspots than in other one - i would pick one with lower hotspots.
If I'm buying 200 bucks expensive PSU gold or platinium I dont care how much of them failing within warranty period. For me it is investition for way longer period(10-15years). Even my old low tier cheap PSU worked fine almost whole 20 years until I changed my habits and switched for laptop.
It worked fine cos I was lucky enough and aware enough to not abuse it by high load). So even cheap design can live much longer than warranty period at relatively low load. But for high power modern PSU i would expect something extra for sure.

So avoiding hotspots is priority for me. Unaware users would pick tier 1 brands PSU in similiar case (if they are looking for long living product) I guess.
For most people without technical background top tier brand name would mean much more than technical reports, detailed reviews or recomendations.

So I think real detailed reviews must be for real geeks. For mainstream users marketing poetry would be enough. So it is a job more for skilled typewiter than for hardcore reviewer. Cos most readers are not patient enough to read detailed reports. But devil is in details when we are dealing with PSU-s.

 

[darkbreeze]

I don't understand why sites don't simply pay Aris what he's worth, to use his reviews and findings. It would make sense, cost less, help Aris, help users, help sites and just be far simpler until such time as he's no longer regularly doing reviews of them. Just sayin'

 

[jonnyGURU]

I think

It doesn't matter what "real load" is. If you put out a product that is advertised to put out 1000W, for example, at 90Vin, at 50°C, it needs to meet those specs.

What you're saying is you'll accept whatever as long as it operates at your requirement... which is perfectly fine. But if you're going to advertise a 1000W PSU that can do 1000W at 50°C with 90V mains, it sure as shit better be able to do that. Otherwise, it's a fail.

Yes.... anyone can make a 600W PSU that is bullet proof that costs as much as a typical 1000W. But what's the point when the cost of the 600W is the same as the 1000W? This essentially defines the failed fanless PSU market. Fanless PSUs are nothing more than higher wattage PSUs derated to lower specs. But a 1000W PSU that is rated for 600W because there's no active cooling is going to sell almost zero units. So what's the point? Companies are in the business to move units and make a profit. Not sell 10 units a month out of their garage and hang on to their jobs at the local McDonald's to make ends meet.

I don't understand why sites don't simply pay Aris what he's worth, to use his reviews and findings. It would make sense, cost less, help Aris, help users, help sites and just be far simpler until such time as he's no longer regularly doing reviews of them. Just sayin'

Because websites are supposed to be profitable. If you pay Aris what he's worth, there's no money left for the profitability of the website. ;-)

 

[taka]

Most PSU reviewers are not providing full data and if you cant find your next model reviewed you can check the data from Cynernetics like i did. Those reports are worth to me more than a Youtube review from some random guy. Example this unit https://www.cybenetics.com/evaluations/psus/616/ a CWT rebranded Thermaltake that for my usecase was the best fit.

And yeah the PSU performance is not just JAP CAP and fan used, it's the entire list of components used from metal case design to the last SMD. The PCB quality and the solder used down to the design of the PSU. All components used and the schematic used and the build quality also external cable AWG and connectors used.

Put on top of that the electrical parameters and protections used load tests and there you have it.

I realy wish you luck in finding a PSU guy here, there are not too many left out there.

 

[Macro Device]

The design is and always will be best for keeping the PSU cool. Sadly, low noise won that battle.

True that. Fortunately idgaf about the noise so my fans roar and temps stay sane or even low for, like, 99.9% of the time. That's why my PSUs and other electronics live for much longer than they are relevant performance/feature wise.

 

[chrcoluk]

True that. Fortunately idgaf about the noise so my fans roar and temps stay sane or even low for, like, 99.9% of the time. That's why my PSUs and other electronics live for much longer than they are relevant performance/feature wise.

Ironically the silent mode on my PSU made a weird noise, I guess is a defect on it somewhere so I enabled normal mode to shut it up, and its never made any kind of noise that I can hear over the rest of the PC.