Antec Earthwatts EA-500D Green test

Discussion in 'General Hardware' started by Makalu, May 30, 2010.

  1. Makalu

    Makalu Guest

    Welcome to my review and test of the Antec EA-500D Green. First, I get to give you a quick history lesson which I know you're all thrilled to pieces about. Fire up the WABAC machine Mr. Peabody.

    This is the third version in the line of 500W Antec Earthwatts models. The first and original model released in the fall of 2006 was a Seasonic made unit based around the popular Seasonic S12II platform. This was the first consumer level retail power supply to tout the new 80+ certification and really led the way with other brands quickly jumping on the 80+ bandwagon. I can either thank or blame the Antec Earthwatts for the emphasis on efficiency at all costs over the following years. Despite the dual rail labeling this was effectively a single rail design with no OCP in place. If you look up other Earthwatts 500W units reviewed on the net this is the model you will see since Antec hasn't sent the later models out for review.

    Next in the EA-500 history timeline was a Delta made design released in 2008 distinguished by the D in the model number. This was a true dual rail design and labeled as such. It carried on the line with 80+ standard certification. For clarification purposes I refer to this one now as the EA-500 Grey since it has an unpainted grey metal casing. Grey or gray depending on which spelling you prefer. Moving quickly into the present now we have the EA-500D Green which is at various sources also called the EA-500D Bronze due to it's bronze level 80+ test results. Now close your books and label a sheet of paper from 1 to 10 because I have a history test in store for you. Oh what the heck, nevermind, let's skip the history test for now and get on with the review shall we?

    The EA-500D arrived at my door in a box which appeared identical to the one in the photo below because it is.

    [​IMG]

    What can I say? It's a box with stuff written on it. Incredibly useful information like 500W ATX12V V2.3 and "DESIGNED BY ANTEC IN CALIFORNIA" seen in the upper right hand corner but you have to contort your neck sideways to read it so please don't hurt yourselves. I want you hearty and hale for the upcoming excitement. So we have a box designed by Antec in California, simply amazing. The actual power supply itself inside the box was designed by someone named An Thec in Khauliphoarnua. No it wasn't, I lied. So let's see what other tidbits we have on the box front. "Earthwatts...Cooler,Quieter,Greener" it proclaims. Well sir I do believe you are correct and this is without a doubt the greenest power supply I've ever laid eyes on. The In Win Commander PSU series is green too but it's a much browner shade of olive drab green to fit the military theme and not a bright vivid verdant forest green to fit the long hair tree hugging Californians who design boxes theme like we have here. I live in Eugene, Oregon (AKA Hippyville) so I'm allowed to say that...they ain't got nothin' on me or my state when it comes to hugging green trees.

    The box and most of the other written info you will find on this unit both warns on one hand that there is no power cord included and brags on the other hand that by getting no power cord and instead reusing your existing cord you can reduce waste and help protect the environment. Fair enough, albeit I suspect the increased profit margin made the decision much easier. I happen to have more unused power cords than God Herself but not everyone is as blessed as I. An unpainted unit has fewer harmful effects on the environment then spraying green paint and releasing VOC's into the atmosphere does, plus the petroleum to make the paint itself. Hey well I'm just saying if you want to get picky about buying eco friendly products. The paint itself here is a rugged powder coating with a slight texture and not going to be easily scratched or scraped off.

    Inside the box we see said green painted power supply with no power cord wrapped in a clear plastic bag enshrouded by a couple of egg crate style shipping protectors. I can use those as kneepads when working in the garden with my green thumb up my...oops let's keep this PG-13 in case there are any children about. Along with no power cord we get a bag of mounting screws not pictured here in my review because the PPSRRB (Puritanical Power Supply Review Rating Board) sent me a nasty letter about my usage of the word 'screw'. Not really, I just had already tossed the bag into my bag-o-screws container that my heirs get to fight over after I die. Along with no power cord we get a pamphlet in no less then 8 languages, five of which are identifiable to me, and another three Venutian dialects or possibly Vulcan or some other green planet inhabited by people with green blood. If you stick a babel fish in your ear you can understand them all but then you'd have not only no power cord but also no tea. The pamphlet shows which connectors the unit has and provides the information seen below.

    [​IMG]

    Or is it provides the lack of information? Regardless, there is no manual, detailed or otherwise available at the Antec website. But hey it's just a dumb power supply and not something complicated to assemble like say maybe one of those surprise toys in a box of CrackerJacks. Now if they were to provide a datasheet and a Chroma report that would be something worthwhile to me. Phew, finally I'm through a cursory yet obligatory retail box discussion sprinkled with sardonic commentary and onto the power supply itself. Here she is up close and personal.

    [​IMG]

    There's an Antec sticker on the fan hub which you can barely read due to the fan grill. Here she is in all her birds-eye view glory.

    [​IMG]

    As you can see only the main 20+4 pin connector cabling gets sheathed and the other connectors get the multi-colored tangle treatment. Maybe if we had the colors of the spectrum arranged in rainbow fashion this would work better with the ecological theme but as it is it's just a cheap ass mess...PG-13 rating be damned.

    [​IMG]

    The chart above shows the cable lengths and a connector count. The cables are not short short but are 1.5" shorter than the NeoECO offers. The distance between the two daisy chained PCIe connectors is longer though which will work better in a dual single-connector cards setup. The one cable with two SATA's and a single molex on the end is a bit unusual but would be handy in a setup with a molex powered fan mounted near the drives. The connectors are good quality with all of the pins well formed and not loose in their sockets like some flimsy low end units have. They all have an identification label on them, "PCIe" on the 6+2 connector and P1-P13 on the rest. The 4+4 CPU connector is labeled P2 on both sides though despite different key shapes between the two. Most of the wires are 18AWG, exceptions are the -12V power, PW_OK and PS_ON control and 3.3V sensor wires which are thinner 20AWG and the last link on the molex chain for the floppy drive connector gets thinner still 22AWG.

    [​IMG]

    The more pertinent information on the label gives rail ratings of 24A on each of the 5V and 3.3V rails and a 130W combined rating for them. Two 12V rails rated at 22A each and a combined rating of 444W (37A). Worldwide safety certifications, RoHS compliance logo, Antec's UL number and the prominent 80 Plus Bronze medallion. Along the side we are informed that the unit has 500W Continuous Power, meaning that the 500W is not a peak rating. I would expect no less from Antec or any other major PSU brand but sometimes marketers struggle to find something they regard as a potential selling point.

    [​IMG]

    Looking at the front of the unit you can see that what sleeving there is ends well short of the unit itself. There's no plastic grommet around the hole to cover a sharp edge like many units have but the metal is bent completely back against itself so there's no sharp edges there and no danger of frayed insulation. That's good. There's a clear acrylic shroud that goes almost completely across the bottom of the air vents. I assume this is to concentrate some airflow up amongst the more crowded components and wiring on the secondary side of things. Someday I'll add colored smoke and an acrylic case to my PSU reviews and figure that out. Ha, well I did actually think about it at one time and even priced some colored smoky thingies but PSU testing methods are easier to think about then they are to actually find the time to do. Let's take a look at the interior now.

    [​IMG]

    First impressions are pretty good. The mainboard is full length, another PCB across the back of the unit houses the transient filtering and rectifier components and there's two small daughterboards on the mainboard too. One on the primary and one on the secondary. The heatsinks are pretty beefy compared to some other high efficiency 500W unit's these days. Overall it simply appears to be a relative lot of 500W power supplyness there. On the other hand this is a traditional double forward design with group voltage regulation which isn't any thing to write home to mother about.

    Breaking down some of the more important and critical components there are two Y capacitors on the AC plug connector and four more on the transient PCB/s seen below.

    [​IMG]

    Two X capacitors (one less than the EA-500D Grey), two coils and one big MOV round out the transient filtering section. The wiring on the choking coils here and throughout the unit is tight and the mounting is supplemented with good amounts of caulking at the base and near the wire entry points to help eliminate the harmonic coil noises that seem to be more and more of a problem with PSU's these days as GPU loads increase. The absence of the additional X capacitor may be a purely cost saving move or it may have been part of the increase in efficiency that the EA-500D Green offers over the EA-500D Grey. I can't say for sure but bear in mind that every component in a PSU has some resistance and reducing that resistance by removing components is a tactic employed by some in the quest for 80+ medals. A small amount of resistance and it's wasted heat energy here or there can have a big impact on efficiency test results--assuming that you consider a 1-2% difference in efficiency is big. I had to get some sort of partial explanation about my "thank or blame" comment in the preamble in somewhere.

    The bridge rectifying diode is Japanese Shindengen LL15XB60 rated for 15A average forward current when mounted on a heatsink which this one is...and a fine little finned sink it is too and not just a flat plate like sometimes seen. This is a more energy efficient upgrade from the Shindengen DX15XB60 rectifier used on the EA-500D Grey offering a lower forward voltage at the low steady state currents after start-up. Primary capacitance is provided by two Taiwanese Ltec LP Series 450V 180uF 85°C capacitors with a rated load life of 2000hr at 85°C. The queen is not amused. The EA-500D Grey has two Samxon HX Series 420V 180uF 105°C's here---or at least the one I bought a year or so ago did---that may have changed too now. I wasn't able to get a good look at the markings on the primary transistors but there are four power transistors and a diode and one of the transistors is an Infineon 15N60C3.

    Sorry about the lack of information, but I don't get paid to do this and have to buy test equipment and test units out of my own pocket and then either sell or give the unit to a friend that needs one and I also don't have a solder sucker and the PCB soldering skills nor time to dismantle and reassemble just to identify what's mounted on the sinks. If Antec would send samples of these out for review then others could fill in my gaps. Blame Obama and/or Bush---I'm sure it would be spit in the ocean to them at this point. On the other hand by being freelance I get to review what I want, when I want without pressure and focus on units that haven't been reviewed by others to satisfy my own curiosity and hopefully provide some useful information to the community in the process. And I also get to test very low end junk which is more interesting to me. But enough about me---So Greenie, what's a nice power supply like you doing in a place like this? <wink wink>

    [​IMG]

    The primary side daughterboard has an Infineon ICE1PC802 PFC controller with AC low voltage brown-out protection and an STMicro UC3845B PWM controller. Nearby is a Power Integrations TNY278PN switcher. All of the transformers are Delta made.

    Moving on to the secondary I again wasn't able to read the markings on all eight transistors mounted on the secondary heatsink but there are two STMicroelectronics STPS61L60CW 30A Power Schottkys on the 12V outputs and two STPS30L30CT 15A Low Power ones on the 3.3V and probably two more of those handle the 5V and two more in the same package size but unknown amperage probably free-wheeling on the 12V outputs.

    The secondary capacitors are all rated at 105°C which is good since the mass of output wiring on that side blocks a lot of airflow and secondary caps are more prone to failure than primaries. The secondary cap brand is a mixture of Ltec, Taiwanese Taicon and two Japanese Chemi-Con KZH series on the 12V output. The Ltec's used are LTG Series which the Ltec (which stands for Luminous Town Electric Company for some added spice if you made it this far in the review) website doesn't seem to have any data on those. They're brown...which is another good earth tone like green. The blue secondary caps are Taicon LZG Series and the green ones are Taicon bearing a 0923A(M) marking. Neither of these seem to be listed at Taicons website.

    The secondary side daughterboard houses the fan controller and a BCD Semiconductors AS339AP-E1 quad voltage comparator implements the over current protection. The little blue potentiometer on the daughterboard I didn't mess with. But on the EA-500D Grey it adjusts the 5V output over a very small range and from examining the circuit traces on the EA-500D Green it would seem to have the same function here. Maybe it's for fine tuning the voltage comparator, I don't know. You can see two silver colored current shunts on the mainboard where the single 12V source is divided into the two 12V rails in the photo below. The 8-pin (4+4) CPU connector is on one 12V rail and everything else on the other.

    [​IMG]

    Notice the black fan control thermal sensor mounted in the middle of the heatsink there behind the 12V/5V coil. Any readers in the quiet computing crowd may be interested to know that I was able to loosen that by inserting a screwdriver through the coil so it could possibly just be swiveled up 90° where it would sit against the lower part of a fin or can be moved to another cooler location entirely to lessen the fan noise. Or if you're a cooling fanatic like me and want to see capacitors last as long as possible then swivel that sucker down between the 12V and 5V Schottky's and let her rip!

    [​IMG]

    That's an ADDA ADO812HS-A70GL 80mm sleeve bearing jobbie rated at 38.6CFM, .160" pressure, 34.4dB/A as measured 1m from the intake side, 3200RPM ±10% and a life expectancy of 31000hrs at 25°C. Turning the main PCB over reveals impeccable soldering with no globs or evidence of hand soldering and no solder residue. All the leads have been trimmed short unlike what I saw in the Seasonic made Antec NeoECO 520C. Overall my impression of the internals is a bit of a mixed bag. The build quality is good with very good construction and a generous design but the double forward design is not cutting edge and the capacitor brand choice is stingy yet adequate.

    [​IMG]

    Before I move on to the testing section I want to show you this photo of the EA-500D Grey I have.

    [​IMG]

    Quickly now, I'll point out the biggest differences seen between the two. One more X capacitor and a less efficient bridge rectifier as mentioned above. It has Samxon 105°C primary caps and a mixture of Ltec and CapXon secondary caps all 105°C. The primary heatsink employs two Infineon 15N60C3's, two Toshiba K2698's and a diode. The secondary employs three Vishay VTS40100CT for the 12V output and two each STPS30L40C handle the 5V and 3.3V. The quad comparator used here is an STLM339AN. You can see the primary heatsink is a more expensive milled affair and more surface area than the simpler fingered and bent style seen on the EA-500D Green. The location of the fan control thermal sensor is different and that's about all the evident differences between the two.

    Now let's move on to the part where I do what I do best. Try to blow things up in the name of science.
     
    Last edited by a moderator: Jun 4, 2010
  2. Makalu

    Makalu Guest

    Load Tests:

    [​IMG]

    I'll mount the unit in the load tester at ambient room which is pretty cool this time of year and just let the PSU and some heat from the 12V load banks increase the temps a bit as I increase the load in roughly 100W increments up to a full ~500W load on the PSU. Since this is a 500W unit and most likely to be used in a single GPU system I'm going to keep the loads on the 5V and 3.3V rails in line with what a single GPU system would put on those rails (≤6A). Be aware that I've measured AC watts and calculated efficiencies at 220VAC input but all other measurements are done at 110VAC input. I've not seen any differences in DC output voltages under the two conditions so only perhaps the fan and temp measurements would be a bit lower at 220VAC. Let's see how it does.

    [​IMG]

    12V output begins not too far above nominal 12.00V and still just above that at typical system loads around 300W and doesn't begin to sag much until we hit full load. These are acceptable drops for a 500W group regulated unit. The 5V and 3.3V are well behaved as expected under their light loads. The 5VSB is rock solid as expected since I didn't raise the load on that rail at all during the test this time. Efficiency is very good at 110V and 1-2% higher across the board at 220V with levels topping the 90% mark in the peak efficiency sweet spot around 300W. The fan spins significantly faster then other units I've tested and may be objectionable to purists but is likely to be acceptable to most when still under 7V. You can see the fan and heatsink cooling solution is doing a good job keeping the temperature delta low at all load levels.

    Now let's crank up the hotbox to 50°C (122°F) and run through the same load pattern as above. I've also let each of the five loadouts run for one hour not counting the time it takes me to measure and set things so this is a total test run of about 6 hours at elevated temps and half of it at elevated loads. Let's see if she burns in or burns out.

    [​IMG]

    The 12V output suffers just a tick under the increased temps and the efficiency varies a touch but for the most part the unit shrugs off the heat like water off the back of a duck. That's what we like to see. Not surprisingly the temperature controlled fan starts out pretty fast and gets only faster but it does a very good job keeping the exhaust temps in hand.

    Crossload tests:

    In the crossload tests I want to see how the unit handles heavily unbalanced crossloads with one major voltage at it's maximum rated current and the other two at a minimal 1A and see if it can keep all three rails within the ±5% tolerance ranges.

    [​IMG]

    Well she holds up very good for a group regulated unit under these extreme crossloads except for the 5V heavy load where the 12V output jumps way high out of spec and the 5V is flirting with dropping below spec. It doesn't like the 5V heavy load much at all. Neither do I since I can't imagine any real life system that would load a PSU this way. And the same holds true for all of the crossload tests I do. This is more of a FYI type test. Perhaps I'll change the methodology to reflect a real life potential crossload. As it is a typical system load with just a few amps on the 5V and 3.3V and a 12V load in the 20A's range could very well be called a heavy 12V crossload---since it is. My regular load test pattern does that.

    You can see in these test results the energy losses from 12V power to 5V power through the windings on the transformer and again a loss with the mag-amp implemented 3.3V power. This is why you see a lot of units going to the DC-DC conversions design to improve efficiency.

    Efficiency:

    The PSU guides (Intel ATX and SSI EPS) require two efficiency measurements. One for measuring the 5VSB rail efficiency during standby and one for total efficiency. In addition as of 2010 units sold in the EU are required to consume 1W or less under a no load off mode condition. This unit claims to comply with the new European ERP (Energy-Related Products) directive. For total efficiency the guides both say to use the 80plus.org loadout for efficiency measurements. 80+ uses the label ratings and a formula to arrive at balanced loads that won't exceed any of the combined output ratings. The guides also say to carry out the efficiency test under their environmental conditions which means 50°C and not the 23°C benchtop temp that 80+ tests at. I'll do the no load off mode and low load standby mode testing first:

    [​IMG]

    The no load consumption of just .35W at the EU type power grid of 220V input is remarkable considering we've seen some units consume as much as 5-7W here in the past. Also the EU power directive for no load consumption will drop down to just .5W effective January, 2013 so the EA-500D is ahead of the game and good to go.

    [​IMG]

    Above are my results using the 80+ loadouts---or as near as I can come with the switched load banks built into my load tester---I could dial it in with some extra resistors I have but I didn't bother with it. The difference in my loads and theirs would have a negligible effect on the efficiency calculation though. The unit shows bronze level results in the hotbox as well. The 80plus.org tests results are here.

    Vmax tests:

    The last three of the load tests aren't part of the guides "per se" but they're used to obtain results in some later tests I'll be doing. Under these three loads each of the voltage rails will be at it's maximum output per label rating as was done in the crossload tests and the other rails will be loaded up too so that a full ~500W total load is achieved. Notice that I also keep the 5V and 3V combined loads under the rated 130W for the 5V and 3.3V combined per the label. The load amounts and results are in the chart below for reference or amusement or test data overkill depending on your perspective.

    [​IMG]

    Before I move on to the next test regimen I'll mention that I tested for short circuit protection on the three major rails which the unit passed. I also did an OCP (over current protection) test on both 12V rails and measured the unit latching off at 27-28A on both. That's what I like to see here on a dual rail 500W unit. It's going to allow you to take advantage of the full potential of this unit if you are inclined to push it that far, yet the levels are still well under the potential for fire under a fault condition which is around 35A in my experience and opinion. I also did an over load test where I put ~9A each on the 5V and 3.3V rails and then ramped up the 12V loads while monitoring 12V voltage and 12V ripple until either the voltage or ripple goes out of spec or the unit shuts down---whichever happens first. For this unit the 12V ripple went over spec at around a 12V/43A load and ~600W total load. I also did something I don't normally do which is just try and load the unit until it shuts down with no regard for keeping the outputs in spec. The unit shut down with a load out of 12V/54A, 5V/23.5A and 3.3V/13A which works out to ~800W total load. Didn't burn anything up and it restarted just fine. 12V ripple was way off the charts around 400mV and rising the last time I looked though. Don't ask me why I even did this test. It's not really relevant to anything that matters and wasn't conducted as a proper OPP or UVP or OTP shut down test and there are a number of possible reasons for the unit shutting down...the 12V rail was maxed out as much as the OCP there allows and the 5V and 3.3V combined were way over the rating and namely it's a 500W unit being pushed way beyond reason. ;)

    Ripple & Noise tests:

    Ripple (fluctuations) and noise (spikes) are unwanted AC currents remaining in your power supplies DC outputs. If the world was perfect and the DC outputs from your power supply were a perfect DC current like that produced by a battery then capacitors would get no hotter than their ambient temperatures. Ripple causes capacitors to self-heat, raising their temperature and decreasing their lifespan (and ability to filter out ripple) by half for every 5C increase.. Ripple & noise is also unwanted because it lowers transistor voltage thresholds and raises gate delay times which leads to errors. In todays world as transistors get faster and faster and chip voltages get lower the effects of ripple are more important than ever. Ripple is also a concern for people who use overclocking methods like raising voltage or lowering Vdroop because these methods lower thresholds too. Here's a diagram of idealized ripple and noise waveforms.

    [​IMG]

    Four major content types can be readily identified:

    • time domain low frequency ripple related to the AC mains frequency
    • high frequency PWM switching ripple
    • noise induced by the switching ripple
    • random aperiodic noise not related to any of the above

    The ripple & noise testing was done at two load levels---low and high. For low load results I measured at the first ~100W loads used in the full load tests above and in the interest of time I'll just mention what those results were. Then for high load results I measured using the Vmax loads and provide the oscilloscope shots at that load level. Testing was done at 110VAC. First let me show you the common background noise measurement on the 12V rail. I use this to see how clean my AC is and insure that I'm not picking up any extreme interference from any other devices in my home or nearby.

    [​IMG]

    It's pretty clean and down close to the minimum amounts my oscilloscope can measure accurately which is ~2mV.

    In the following screen shots the left side window zooms in on the switching frequency level content and the right side steps back and views the whole picture so to speak. The right hand side waveform and it's P-P measurement in the little window at the lower right hand side is the one that must be below specifications which are 120mV for the 12V and 50mV for the others. In the past I have done these using a 10mV per vertical division of the grid shown on the oscilloscope scan. That's the lowest setting and allows you to get the best view of the waveforms. For this unit I had to use the 20mV setting in order to display the 12V results and I've kept that same setting for all the rails so that the results are visually consistent. Let's take a look at the 3.3V rail first.

    [​IMG]

    The low load 3.3V ripple was 22mV and rose to 35mV at high load. That's decent but not exceptional.

    [​IMG]

    The low load 5V ripple was 15mV and rose to 40mV as seen in the image above. That's acceptable and still under the 50mV limit but again it's far from impressive. Now the 12V.

    [​IMG]

    12V ripple at low load was 32mV and rose to ~94mV under full load. That's a lot of ripple but still actually a smaller percentage of the allowable tolerance than the 5V rail was. The main thing here is that it's still well under the 120mV tolerance but it's also a bit of a disappointment compared to the original Earthwatts 500 based on the Seasonic S12II platform which had exceptionally clean outputs.

    [​IMG]

    Above is the 5Vsb ripple which I measured using the 5Vmax loadout but took a couple amps off the 5V rail and put the full rated 2.5A on the 5Vsb rail instead. I didn't do a low load test of it. The high load test results are ~27mV and perfectly acceptable for this rail.
     
    Last edited by a moderator: May 30, 2010
  3. Makalu

    Makalu Guest

    Timing Tests:

    The timing tests and requirements differ quite a bit between the ATX and EPS guides. I've chosen the harder of the two when there's a choice but this unit only claims ATX v2.3 compliance and EPS12V (which I take to simply mean it has an 8-pin CPU connector). No EPS guide version number is given. Have a look at the the EPS power supply timing diagrams anyway.

    [​IMG]

    There's like 20 measurements there not counting the fact that some have multiple parameters to check and there's also 3 Vouts, 12V, 5V and 3.3V. The 5VSB has it's own timing requirements. Here's a simplified version showing the timing tests I'll do. Except I completely spaced out doing Test 4, the PW_OK delay. Well it wouldn't be an official Makalu PSU test if I didn't forget to do a few things.

    [​IMG]

    T1: 5VSB_On_Time

    [​IMG]

    This is the delay time from AC being applied to 5Vsb coming into regulation with the PSU off. EPS requires 1500ms max (that's 1.5 seconds...a long time). This is the first unit I've tested that didn't have some significant delay here as the 5VSB powers up almost immediately. I don't quite know what to think about this but I don't see any reason off the top of my head why this would be a problem so maybe it's the slower units that have a "problem" getting the 5Vsb in order. There's just a tiny bit of overshoot here but not near enough to fail the 10% tolerances for that.

    T2: Power_on_Time

    The power on time is defined as the time from when PS_ON (the green wire) is pulled low to when the 12V, 5V and 3.3V outputs pass within their regulation ranges. The Vmax loadouts are used in this test. Voltages must power on within 400ms and they must all come into regulation range within 20ms of each other.

    [​IMG]

    [​IMG]

    [​IMG]

    Top to bottom the 12V,5V and 3.3V rails come into tolerance range right around 200ms and within just a few ms of each other so that's a definite pass. Almost four times the amount of delay as I saw with the Antec NeoECO 520C though. Regardless it's still around half the required delay time.

    T3: Rise_Time

    [​IMG]

    Rise time is the time it takes for voltages to rise from 10% of nominal to within regulation (95% of nominal). The guides go into some detail about rise time characteristics:

    • Rise time must be no faster than .2 ms and no slower than 20 ms.
    • There must be a smooth and continuous ramp of each voltage from 10% to 90% of its final set point within the 5% tolerance range.
    • During the rise time the slope must be positive (always rising left to right like this "/" with no drops like so "\" yet not straight up and down...no slope.)
    • Slew rates (the ratio between voltage rise and time) are defined for the entire rise time and also for any 5ms segment of it. Rise time slew rates must be greater than zero and ≤.33V/ms for the 3.3V rail, ≤.5V/ms on the 5V and ≤1.2V/ms for the 12V rail. 5ms segment slew rates I ain't doin'...gives me a headache.
    • Finally the voltage overshoot must be less than 10% above nominal.

    I added settling time to the diagram which is the time between 10% voltage and when voltages enter and remain within regulation, but it's not part of the specs. Just an important attribute to consider when a unit has problems getting and keeping it's outputs within tolerance.

    If this thing was having issues with it's rise times I'd zoom in on them and capture two at a time but they're well behaved enough to just look at the Power_On_Time shots in the previous test. Rise times are roughly 15ms for 5V and 12V and 10ms for 3.3V...all of them well under the 20ms requirement. Slew rates are roughly .3v/ms for the 3.3V and 5V and 1.2V/ms with the 12V. You can see all three voltages rise smoothly and come into regulation with no overshoot or stability issues. Here's a good example of the kind of crap we don't want to see.

    [​IMG]


    T5: Hold-Up Time

    Hold-up time is the amount of time that the PSU can continue to operate in spec after the loss of mains AC power. This allows the unit to "ride-through" microcuts on the line. AC power can have very fast imperceptible outages (microcuts) caused by things like heavy motors switching on, utility workers doing livewire work or switching on the power grid. Hold-up time also gives stand-by type UPS's time to switch to battery power before the PSU shuts down after a power outage. The ATX v2.3 hold-up time is 16ms at 100% load. The 12Vmax loadout is used here.

    If you've ever taken on a task and then asked yourself over and over why you bothered then you know how I felt doing the hold-up time tests on this unit. The first unit I did hold-up time was the OCZ Fatal1ty 400 and it passed handily and essentially the same results every time I ran the test which was many many times as I was getting a handle on how to even do the test to my satisfaction. Then I tested the NeoECO 520C hold-up times and it wasn't a hassle either, it simply and consistently failed over and over with results in the 13ms range. This one gave me a headache, it did fail over and over but the results from one test to another were inconsistent with hold-up times ranging from a low of ~12ms to just under the required 16ms. I did everything I could think of that might be causing this on my end of things to no avail. And I do really try to give units the benefit of the doubt in my tests, especially since oscilloscopes are known to not be the most accurate DC voltage measuring tools and the results depend on me placing the cursors in the right position. But finally after several hours I decided this is just the way this unit is and moved on.

    [​IMG]

    Above you can see the best result I could get measuring hold up time between loss of AC and the 12V rail dropping out of spec which was 15.69msec and still just under the 16ms requirement. Most of the results were in the 12-14ms range. But in my efforts to get good results I also measured the time between loss of AC and the PW_OK warning signal asserted low which should happen ~1ms before the voltages drop out of spec---and has with the other units I've tested. But this one the PW_OK is running about 4ms behind the voltage dropping out of spec. As you can see in these two screen shots, one where I measure the time between voltage out of spec and PW_OK deasserted and one where I measure the time between loss of AC and PW_OK going low.

    [​IMG]

    These test results were at least consistent. Consistently fail that is with PW_OK going low later than any of my voltage dropout results. This is a concern because what this means is that the power supply will continue operating after voltages are out of spec and could cause the system to crash instead of shutting off. That could result in data corruption or other errors. Ok it's a minor concern I admit...but it's not right IMO and I've followed the hold-up time testing done at canardpc.com and he saw the same issue with one unit too. I forget which one. If they would just quit skimping on primary capacitor size and give these things ample hold-up times it would make my life much easier! If the $50 bling-bling Sirtec/OCZ unit can use good sized caps then why can't a spartan Antec?

    The EPS hold-up time requirement is 17ms between loss of AC and PW_OK warning but only at a 75% load level which is easier. In my frustrated state I messed up here and measured 12V voltage drop instead of PW_Ok. But the results for that are 22ms and well above the 17ms requirement so that's all well fine and dandy.

    [​IMG]

    Sometimes ya just gotta say enough is enough and I did my best and time to move on...so let's move on.

    T6: 5VSB Hold-Up Time

    [​IMG]

    ATX 5VSB requirements are the same as the other outputs (16ms) but EPS requires 70ms hold-up time for the 5VSB. It's not a problem though with over 200ms hold-up time under the full 2.5A load.

    Conclusions:

    To wrap up I'll characterize this one as a solid build for those who want high energy efficiency and not looking to spend over US$75 which is the average retail price on it. I bought mine through ATXpowersupplies.com though which has it for $55, which I feel makes it a much more attractive option then it does when comparing it against units in the $75 range. The Delta made Antec Earthwatts 650W reviewed here sells for around $75 and offers better capacitor brands, independent voltage regulation, better power distribution across three rails, a quieter 120mm cooling solution and of course 150 more total watts---100W of which come from an extra 8A on the 12V combined rating---all at the sacrifice of a few percentage points of efficiency though. Personally, I'd spring for the EA650 between these two choices in a heartbeat because if I wanted to save a few dollars on my electricity bill I'd keep my bathroom light off at night instead of on like I do just so I can see down the hallway when I get up to pee in the middle of the night.

    What I like:

    • Solid well built Delta construction
    • High energy efficiency


    What I don't like:

    • Only one cable sleeved
    • Mediocre ripple and noise suppression
    • Failure to pass ATX hold-up time due to inadequate primary capacitor sizing
     
    Last edited by a moderator: May 30, 2010
  4. dcx_badass

    dcx_badass Guest

    Nice post, read the first two but the second two went way over my head.
     

  5. Makalu

    Makalu Guest

    mission accomplished then ;)
     
  6. TearTheRoofOff

    TearTheRoofOff New Member

    Messages:
    1
    Likes Received:
    0
    GPU:
    3060 ti
    Apologies for the necro, but I just had my one of these go pop. 16 years is not too shabby.
     

Share This Page