Zap
Elite Member
- Oct 13, 1999
- 22,377
- 2
- 81
My PSU in question is an Enermax SFX unit. Really heavy for something that small (using the "weight as quality" factor for PSUs).
According to Jonnyguru the Ultra 400W V series will really put out the 20A it claims. I'd say these PSUs are worth the $25, let alone FAR.
Likewise with the high-pitch noise. This is an older system so it should not be under intense load, but the noise kicks in moments after turning on and can be heard all over the second floor, but tends to go away after a minute or so. Maybe only a couple units are affected..but heck, it was 'free'.
"According to Jonnyguru the Ultra 400W V series will really put out the 20A it claims. I'd say these PSUs are worth the $25, let alone FAR."
Thanks a bunch for that link. Been looking for a PS stress test for a long time.
mindless1
Diamond Member
- Aug 11, 2001
- 8,334
- 1,550
- 126
Jonnyguru's testing still ignores a critical part of PSU current capability testing- how long it'll run like that. It's not at all unheard of to POP a generic PSU after a few months, but how many does he test that long at rated current? It's a known fact that running at higher temp (or lower margin, nearer parts peak ratings) wears out parts faster- for example typical ratings derate life by 50% for each 10C temp increase. Even a larger fan cannot cool the part dies (significantly) more if the heatsink isn't doing enough to sink that heat away.
This is not built as well as most PSU spec'd for 20A on 12V rail, excepting other generics. If you only care whether it can deliver 20A short-term, then still we'd have to take a closer look at what the other PSU we'd contrast, can deliver short-term, as it'll tend to be higher than the labeled value if we're considering peak rather than sustained. Sustained ratings are not just a few hours or days on a bench- the PSU must meet it's MTBF rating, as that's the whole point of an (honest) MTBF rating.
It also has a sleeve bearing fan, while other manufacturers that had used similar (if not same) sleeve bearing fans had later found their lifespan too short in this role (and horizontal orientation too, a no-no for sleeve bearing fans) and thus have switched to ball bearing fans. I highly recommend relubing this fan periodically if not replacing it.
To make matters worse, these fans are not balanced! It appears the manufacturer decided to just SKIP this step to cut costs. It would be less of a problem on a fan with a high quality plastic, molding, but looking at my fan one of the blades has a deformity on the trailing edge (irregularly thicker) and could't possibly be naturally, near-balanced. Instead it's a prime example of a fan that needed balancing more than ever.
That doesn't mean this PSU isn't a great deal for FAR, "IF" your system isn't so power hungry. It's just not a 400W PSU for the purposes of comparison to any real 400W PSU you'd pay more for, or might be needed for some particular system configurations.
... and yes, I have one of these, open in front of me right now. Free is free!
making my first attempt at a complete build soon... still deciding whether to run it with an e6300 or e6400... but at least i've got the psu now! thanks OP!
My final cost for 2 of the 400W supplies was $5.74
Are they perfect? Nope
Will they burn up if ran at rated load for an extended period? Yes
Will most other reasonably priced supplies burn up also? Yes
Is the Ultra 100x better than the cheap ass things NewEgg sticks in budget cases? Yep
Will they work in my intended application (3400+, One HD, One DVD)? Yes
Am I happy with them? Well, I bought 2 and I would buy 10 more if the rebate would work
Dude, after working with Deer, Sparkle, Mustang, and other shithole power supplies over the years, these Ultras are amazing for the price of ZERO.
I look at it this way, I just built 2 new 3400+ machines for less than $400 total. I couldn't be more satisfied with these supplies.
Are they perfect? Nope
Will they burn up if ran at rated load for an extended period? Yes
Will most other reasonably priced supplies burn up also? Yes
Is the Ultra 100x better than the cheap ass things NewEgg sticks in budget cases? Yep
Will they work in my intended application (3400+, One HD, One DVD)? Yes
Am I happy with them? Well, I bought 2 and I would buy 10 more if the rebate would work
Dude, after working with Deer, Sparkle, Mustang, and other shithole power supplies over the years, these Ultras are amazing for the price of ZERO.
I look at it this way, I just built 2 new 3400+ machines for less than $400 total. I couldn't be more satisfied with these supplies.
Zap
Elite Member
- Oct 13, 1999
- 22,377
- 2
- 81
Jonnyguru's new testing procedures add in a "hot" test.
As for testing it for a few months, nobody does that, probably not even PSU manufacturers. Who wants their expensive test equipment tied up for months? If a part has a MTBF measured in years, who's going to run one for that many years to test the MTBF and then publish results... by then you won't be able to buy those parts anymore and they may even be outdated and no longer in use by most people.
How do you know?
Post pics at Badcaps forums and list all the capacitor brands it uses.
The thing is that it IS designed to put out rated power. That's something which some expensive PSUs can't even do. Regarding longetivity, I don't think it would be designed to have a limited lifespan, especially since Ultra is willing to give it a lifetime warranty. As for "real" lifespan, that will have to do mostly with capacitors and even then it is just a guesstimate based on known long-term performance for that brand of capacitor.
The other thing is that even assuming it can't sustain 20A long term, computers don't draw the same amount of power continuously, and will only peak in intervals.
Even ignoring the rebate, if you can find another PSU for $25 shipped that you can recommend above this one, please let us know! Otherwise, let's agree to disagree.
mindless1
Diamond Member
- Aug 11, 2001
- 8,334
- 1,550
- 126
That's nice, and yet still non-applicable to a lifespan test.
It's not that I expect reviewers to have all their tested units running for years, that is not a reasonable expectation.
However, the fact remains that short testing periods can only DISqualify a PSU, they cannot qualify it, not even for whether it meets it's rated current.
Consider PSU built beefier, but rated the same. What's the point of pretending they made it better for no reason? Clearly they had reason, more expensive parts aren't spec'd just for the heck of it.
PSU manufactures do test continually, not just months but for life of the unit.
Someone whose core business is, err, building PSU. Also someone who wants to try to qualify PSU for a use instead of disqualifying.
If a PSU works for 1 week in a test, and given a large enough sample size, the psu has been qualifed as "good", "appropriate", or your choice of adjective for the length of time tested, NOT for a system intended to run for orders of magnitude longer.
There are industry standard testing procedures to find MTBF without running that many years. Quality manufacturers do strive to build better units, we can see it in the actual construciton of the units with same or similar ratings!
I'm sorry it's so confusing to you but this is not a difficult topic, switching PSU are not new technology, only reinterations of existing tech with application to a computer's various rails and current per.
Because I have experience with electronics, including design, repair, down at the component level, and I don't jump to false conclusions instead of tests. We can compare the parts as it is not a particularly unique design. Citing something can sustain a current without the context- the life of the system at a bare minimum, is a false conclusion. I don't just have stacks of dead PSU, I find faults, because that is the only way to avoid them in a cut-throat market where you are a system integrator. Then there's also that part about the market not being a vacuum, there are other whole PSU to compare and even more discrete components available.
They are yellow jacketed "JEE" branded, and a few Canicon in smaller sizes in the areas not needing low impedance. It's not my burden to take pics and post on some other 'site based on your prompting. If you'd like to pay for this service, fine, but otherwise there is a large difference between what I volunteer to do and what someone else wants but won't do themselves- as anyone who has this unit can do it if they want it done.
It might be worth mentioning that the -V 400W JonnyGuru tested appears to have some topology, but obviously does not have these yellow caps, and has a different transformer too. If the pictures had been better, possibly other parts differences might have been revealed too. Parts can be changed for legitimate manufacturer supply sourcing or cost reasons (but in the latter, could be a problem), but also less legitimate purposes like cherry picking units for various reasons.
The thing is, that no, it is not. Any PSU can be overdriven up to the point of saturated cores (it has smaller transformer and inductors than decent 400W) to get more current but the other requirement is that it still has the necessary lifespan doing so (fan, heatsinking, caps, transistors and diodes). It is NOT designed to put out rated power, unless you are using a different rating system which is the crux of the problem in the first place, that rating systems are in place to comparatively choose between different products.
Did I ever claim expensive automatically meant quality or accurate ratings? No. On the other hand, it is true that there is a certain inherant cost to build a decent 400W PSU, and while that cost isn't direclty related to retail selling price, in the end any PSU that can't meet it's labeled rating for it's rated MTBF, is fradulently marketed.
These are not parameters that are subject to further thought on your part, they are set-in-stone requirements. Some expensive PSU are also over-rated and they too should be avoided, and those doing the ratings, in prison.
Apparently you don't see how design decisions can cost or save money, and that there is a difference between designing to last a long time and designing to minimize cost per unit such that if/when a few units get returned for replacement, it was still more profitable to replace a few of those rather than making every unit at a higher cost. The fact remains- these are not built as well as many other 400W PSU. It is quite clear what their strategy was with the guarantee.
Actually, the primary failure points include fans, caps, diodes, and transistors. A few failures might be defective parts (making some brands more desirable from a QC perspective), but when so many corners are cut, the average failure rate climbs from the multitude of causes.
While trying to predict what day or month a cap may fail will be a guess (unless one had instruments measuring moisture content, pressure buildup, electrolyte composition with a specially prepared cap), there is still clear correlation between quality and lifespan and those badcap forum examples are poof enough of this.
You're playing the shell game, moving the target to make it seem ok. By allowing products that don't meet specs to florish, you support that market segment and it persists.
Let's imagine for a moment if all these falsely rated units didn't exist anymore, if you could actually buy a PSU based on it's ratings-THAT is how it was supposed to be, it's the core purpose of ratings. If Dell et al had bought these, there'd be lawsuits over this factor, because ratings are an objective parameter, not subjective to whether a given system could run anyway.
Even ignoring the rebate, if you can find another PSU for $25 shipped that you can recommend above this one, please let us know! Otherwise, let's agree to disagree.
Nobody claimed it wasn't a great deal at FAR. I'd still rather a market where only appropriately rated PSU exist and those overrating were in prison. If they'd rated it for lower wattage and current, it'd still be the same PSU and could still be FAR, but they went over the line instead.
mindless1
Diamond Member
- Aug 11, 2001
- 8,334
- 1,550
- 126
A few more observations about this PSU.
It's very rare for a manufacturer to be so caught up on saving a few cents that instead of adding a riser board for the control circuit, they just ignore the main circuit board keep-out zone along the edges. This Wintec model does that, some holes were so close to the edges they broke through the side of the board, and the copper traces are right AT the edges, no margin at all.
The board itself is very low quality, is already bowed from the weight of the parts- and these aren't particularly heavy parts on it compared to many PSU. The copper is thin enough they had to add extra jumpers on top for the ground away from the transformer and flood the area with solder..
Several wires in the wiring harness have no place on the PCB- two wires enter the PSU case then are joined into only one soldered to the PCB. Several wires were not inserted far enough into the board to where the insulation starts, have at least 3mm exposed bare wire. The same is true for jumpers and capacitors, this thing looks like it was hand assembled, and in a hurry at that.
The heatsinks are only 2mm thick. The crude stamped out 'sinks on ~250W PSU from yesteryear were usually 3mm thick. This thickness is evermore important on a short sink (to fit the 120mm fan inside), or of course for higher current small package parts (like the STPR1620, dual 8A ultrafast diode - this isn't even a Schottky so there goes more heat/efficiency, it's always nice to see hotter parts on thinner 'sinks).
Now look at your label and wonder as I did, are they using a 16A rated part for the 30A rail, the 18A rail, or the 20A rail? If you just guessed the 18A rail, try again.
The so-called 20A, 12V rail is already limited to 16A by the very first part spec I checked - and even 16A requires good 'sinking which may depend on your ambient temp. They saved mere pennies on this downgraded part when it's the most loaded rail on a modern system making use of the connectors it supports. The 5V rectifier (itself, not necessarily the transformer or inductor) does support 30A though, I'd thought about just switching these two diodes because their choice is bizarre, almost as though they have retrofitted an old design not suited for ATX2.0 systems but only so far as the wiring harness and connectors are concerned.
It's very rare for a manufacturer to be so caught up on saving a few cents that instead of adding a riser board for the control circuit, they just ignore the main circuit board keep-out zone along the edges. This Wintec model does that, some holes were so close to the edges they broke through the side of the board, and the copper traces are right AT the edges, no margin at all.
The board itself is very low quality, is already bowed from the weight of the parts- and these aren't particularly heavy parts on it compared to many PSU. The copper is thin enough they had to add extra jumpers on top for the ground away from the transformer and flood the area with solder..
Several wires in the wiring harness have no place on the PCB- two wires enter the PSU case then are joined into only one soldered to the PCB. Several wires were not inserted far enough into the board to where the insulation starts, have at least 3mm exposed bare wire. The same is true for jumpers and capacitors, this thing looks like it was hand assembled, and in a hurry at that.
The heatsinks are only 2mm thick. The crude stamped out 'sinks on ~250W PSU from yesteryear were usually 3mm thick. This thickness is evermore important on a short sink (to fit the 120mm fan inside), or of course for higher current small package parts (like the STPR1620, dual 8A ultrafast diode - this isn't even a Schottky so there goes more heat/efficiency, it's always nice to see hotter parts on thinner 'sinks).
Now look at your label and wonder as I did, are they using a 16A rated part for the 30A rail, the 18A rail, or the 20A rail? If you just guessed the 18A rail, try again.
The so-called 20A, 12V rail is already limited to 16A by the very first part spec I checked - and even 16A requires good 'sinking which may depend on your ambient temp. They saved mere pennies on this downgraded part when it's the most loaded rail on a modern system making use of the connectors it supports. The 5V rectifier (itself, not necessarily the transformer or inductor) does support 30A though, I'd thought about just switching these two diodes because their choice is bizarre, almost as though they have retrofitted an old design not suited for ATX2.0 systems but only so far as the wiring harness and connectors are concerned.
mindless1, thanks for doing all this work! It's really nice to get some actual facts on the part before I ordered it (free is free!), and it sounds like this PS will still work out OK to keep as a spare for the older computers. It's good to know it's probably not suitable for a new/higher end build
i didnt buy this product cause i dont relally know what it do but the discussion here is interesting. I look up this diode spec and it says 16A is at 125 degree C. On one of the curve if u extrapolate it kind of shows that 20A is possible at 100 degree C.
i saw that plot to it show that 20A requires just over 1V. what do u think would be the typical operating temperature on the diode?
oh wait wait, i am new to this voltage current stuff just start reading on web today and my roommates books. seems like current is what creates heat and it is linear to friction so therefor linear to the amount of heat generated? a 25% increase in current would make a 25% increase in heat generated?
the Ultra 400W V-series works great powering my backup system (A64 3400+ Clawhammer, 7600GS, 1GB DDR, Theatre 550 TV tuner, CL Audigy)
mindless1
Diamond Member
- Aug 11, 2001
- 8,334
- 1,550
- 126
While you can use typical forward voltage values to get a ballpark, for failure prevention purposes you have to calculate worst case (max and/or min. values). That's 1.2V, forward voltage at 125C @ 16A. I'm going to avoid extrapolating (as you put it), because graphs aren't necessarily able to be extrapolated, they are often meant to be the spec'd operational range but another more complex expression than the simple single digits they follow.
[ V(f) @ I(f) ] * I(f) = W
1.2V * 20A = 24W
Too many variables remain to resolve this to a case temp, such as that the worst case V(f) would be higher than 1.2V @ 20A, ambient temp, active airflow rate, current->heat from other actives on the same 'sink, plus it's using an unknown quality silicone pad.
The PCB had accomodations for a TO247 part instead of TO220. IMO, they should have used one given the minor cost difference in quantity. This alone would have made the thin heatsink less significant. I may end up doing that to mine, but upgrading parts on a FAR PSU that already has a transformer bottleneck too isn't exactly a high priority.
FWIW, "IF" you're still considering one of these it might be as well to buy the 500W version for the cost of shipping. I wouldn't try to load it till it's rails were maxed out either, but it'll come closer to meeting the needs of a more power hungry, modern system even though it may still have some of the QC issues mentioned above like poor assembly and PCB.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 20K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 12K
-
Discussion Speculation: Zen 4 (EPYC 4 "Genoa", Ryzen 7000, etc.)- Started by Vattila
- Replies: 13K
-
Facebook
Twitter