Videocard: EVGA 8800 Ultra
Processor: Intel C2D E8500
Mainboard: Asus P5Q
Memory: G.Skill 2x1gb PC28500
Soundcard: Audigy LS/ Logitech THX 5
PSU: Gigabyte Odin GT 550W
Antec Neo ECO 520C Test -
| posts: 4,197 | Location: Eugene, Oregon
This is a review and test of Antec's new model Neo ECO PSU which came out a few months ago. Antec doesn't appear to be sending out any Neo ECO review samples and the retail photos showed through the grill that this wasn't the same as any Seasonic platform we've seen in the past and finally my curiosity got the better of me and I bought one to see what we have here. First though a quick history of the Antec Neo series. It was the summer of 2004 when Antec introduced the Neo line of power supplies. The name and packaging were inspired by the movie "The Matrix" and the marketing focus was the modular cabling which was the "us too" feature in mainstream power supplies in 2004. This original NeoPower was a Channel Well Tech design similar to the CWT made Antec Truepower 2.0. It featured independent regulation and 3% voltage regulation and came in 380W and 480W flavors. However, this model had quality issues, both short term and long term from the use of the notorious Fuhjyyu brand capacitors coupled with poor cooling and this design was shelved after just over a year on the market.
The replacement made by Seasonic and dubbed NeoHE dropped the Matrix themed marketing in favor of energy efficiency (HE stands for High Efficiency). This was a new and custom design seen only in the Antec NeoHE's. The unit featured both modular cabling and independent 3% voltage regulation like it's predecessor. There were some motherboard compatibility issues addressed over several revisions and it also had some quality issues, namely poor soldering and general low quality control but if you got one that wasn't DOA or died prematurely and worked with your board then you did ok and no long term concerns like the bad caps in the original NeoPower. The wattage range was 380/430/500/550. The 650W NeoHE was the same platform but used a different HSF solution making it basically a modular version of the Antec TruePower Triple 650W.
Antec never could seem to make up their mind what they wanted to call this one. NeoHE, NeoPower and NeoPower HE were used in various places at various times. The 430W and up models have been discontinued (most of the world) and Antec is currently calling those NeoPower whilst the one model still in production is called NeoHE. In the middle of last year there was a short lived version of this indy regulated Seasonic selling in China called the NeoHE ECO. The NeoHE's have those huge sinks for 80mm fan flow like Seasonic used on the PCP&C Silencer 750.
Enter late 2009 and we see the Neo ECO appear on shelves without any fanfare. The marketing theme is once again energy efficiency in an environmentally friendly package. The Asian (and possibly Australian) markets get a modular version without the "C" in the model name in a glossy retail package as seen here.
Notice it has a PCIe only connector on the modular interface instead of the "one-plug-fits-all" scheme on the NeoHE which makes it so hard to find replacement cables for those. The ECO comes in 400W, 450W, 520W and 620W versions. You can view the retail box and external photos here.
The cables are good length and nicely sleeved from end to end. My only gripe here is the 8 and 6-pin PCIe connectors on the same cable. For most cards that draw 50-60W through each connector this isn't going to be a problem. I did a quick experiment and put the max specification of 150W on the ECO's 8-pin and 75W on the 6-pin and measured .3V drop at the pin side of the connectors from no-load to load. That's probably not going to be an issue but I didn't run it for hours in a heated case and measure either...something to consider when comparing is all.
Connector count and cable lengths:
The label rating shows a single 12V rail rated for 40A and worldwide safety logos. Antec says this is an ATX v2.3 and "EPS12V" unit...I presume they just mean it has an EPS 8-pin CPU connector since no EPS version number is given. Let's take a look at the guts before we put it to test and find out if it has any.
Opening the unit reveals a traditional double-forward design with group regulation and more in common to the Seasonic S12II then the NeoHE indy design. Two "big" heatsinks here and the bridge rectifier gets it's own little sink. The S12II is a three "big" sink layout with the rectifier and PFC stage on one, PWM got it's own sink and secondary another. The PCB is single layer paper/resin lowish-grade but typical for most budget units. It's dated 2009 and good for components up to 620W and single 12V rail design only. There's no glaring defects or quality issues that I can spot. The overall impression is rather anemic for a 520W unit but I'm spoiled and it looks beefier than the S12's which perform pretty good so I'll reserve judgement for later.
Edit: After this review was done Hardwaresecrets.com reviewed the Seasonic S12II Bronze model which is the same design as the Antec Neo ECO
The fan is a 120mm ADDA AD1212MB-A70GL medium-speed ball-bearing affair rated for .33A, 2050rpm, 80.5cfm, .117 inches pressure and 38db noise. The transient filtering section is good with two coils, two X capacitors, 4 Y capacitors and a nicely sized MOV.
A single GBU806 rectifer rectifer is used. Two Fairchild FDP1850N MOSFETs and a diode handle the PFC and two more FDP1850N's are used for PWM. There's an Infineon ICE1CS02 combination PFC/PWM controller on the daughterboard next to the primary capacitor. That's a Japanese Chemi-Con 85°C 400V 330uf MSQ series and seems awful small for a 520W unit to me but we'll find out later. Two SBR30A50CT Schottkys in parallel handle the 12V output and an HY-510N supervisor chip on the other daughterboard is charged with housekeeping and safety protections. The secondary capacitors are all 120°C Su'scon SD series from Taiwan.
The soldering is a little sloppy but it's better than my photography. You get what you pay for.
So these are all common components used by Seasonic in the S12II 500's except the Infineon Combi chip replaces the old popular Champion-Micro CM8600 and I don't recall seeing an S12II based unit with Su'scon secondary caps, most of them used OST. The heatsink arrangement, 620W platform limit vs. 500W and no dual-12V rail option are the other evident differences.
Correction: Three 30A Schottky's produce the 12V output, not two as I wrote at the time. That's another change over the original S12II which was two in parallel there. I need an optical toy to read these without dismantling it.
Finally the part where I get to try and blow stuff up in the name of science. We'll mount the unit in the load tester/hotbox 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 we increase the load in roughly 100W increments up to a full ~520W load on the PSU. Since this is a 520W 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). Let's see how it does.
A quick note regarding my efficiency measurements/calculations. They're not accurate. I've known for awhile that my efficiency numbers are lower than others results but wasn't completely sure where the problem was (could have been in either the input or output measurements or both) and just how low they are. At this time, a couple of months after this test, I now have upgraded my equipment from a $20 power meter to a $200 one and determined that this was the source of the error (input measurements). I've not retested the Neo Eco but preliminary results with another unit indicate that my efficiency numbers are about 2% low. Sorry 'bout dat. I do still recommend that if you want to compare efficiency numbers between units then go to the 80+ page
Everything looks pretty good here. The 5V rail is steady, the 3.3V rail can't seem to make up it's mind if it wants to go up or down but stays well within the acceptable tolerance range. The 12V drops .21V which isn't bad for a group regulated design but it's not the best I've seen in some newer designs either. It's merely acceptable and to be expected and it's always above nominal 12.00 which is good. Efficiency runs around 82% at the real world load levels of 150-300W. The fan was exceptionally slow and quiet at "2d" load levels before the controller gives it a more serious 7V at 300W/"3d" load levels. Each loadout was ran until both the temperatures and fan voltage stabilize which was about 30 minutes on average for this unit today. All in all a good showing here.
Now let's crank up the hotbox to 50C (122F) and run thru 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 was a total run of over 6 hours at elevated temps and half of it at elevated loads. Let's see if it'll burn-in or burn-out.
Surprisingly the unit does a little better in a couple of areas than it did in the "coolbox". The amount of 12V drop is down a tick to .14V and efficiency is up a percentage across the board. As you can see I forgot to get a fan RPM reading after the first load...new laser toy wasn't worked it into my "routine". Anyway, you can tell from the voltage measurement that it was slow and quiet. That didn't last long though as it rose quickly with power output under these high ambient temperatures. In fact this units fan controller responds to load increases quicker than any unit I've tested before. Other than that it's like the cool run...another good showing.
In the crossload tests we 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 range.
The 12V heavy load manages to keep things in spec but not the 5V and 3.3V. On the 5V heavy crossload test the 12V jumps way high out of tolerance while the 5V and 3.3V both just barely manage to keep within tolerance. I had to put 22A on the 12V in order to pull it down into spec with the 5V rail maxed out. With the 3.3V heavy crossload the 12V was just slightly over tolerance with a 1A load but dropped into spec at 2A so I used that. The 3.3V rail however jumps high out of spec and wouldn't come around by just increasing loads on either the 12V or the 5V rail alone. Only by loading both the 12V and 5V worked to get the 3.3V in line but that's a balanced load not a crossload. However dropping the 3.3V load from 24A to 20A and everything came into spec with 1A loads so maybe the 3.3V rail should be rated for 20A not 24A.
I'm not going to knock the unit hard for failing these two extreme crossloads. It's asking an awful lot out of a group regulated design and besides nobody is going to be able to put any of these loads on a PSU with any kind of system configuration I can imagine. Sure you can load up the 5V with a bunch of drives but you can't do it without loading the 12V considerably too. 3.3V maxed out just isn't happening unless maybe you're running several motherboards off one PSU. Also the ATX guide says that 3.3V output must be less than the 12V and 5V combined output at all times so I'm violating that with the 3.3V heavy crossload test. Any degree of crossloading is sure to be 12V heavy...which the unit handles just fine.
Please see the note above regarding my efficiency results
The PSU guides (Intel ATX and SSI EPS) require two efficiency measurements. One for measuring the 5VSB rail efficiency and one for total efficiency. 5Vsb efficiency was ~71% at loads 1A-2.5A and the guide wants 70%...pass.
The guides both say to use the 80+ 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 first two loadouts are realistic but I think an energy efficiency requirement at 100% load on a PSU is about as useful as a gas mileage requirement at top speed on a car...unless your driving habits tend to include a lot of high speed police pursuits it's not pertinent. The guides also say to carry out the test under their environmental conditions which means 50C and not the 23ishC benchtop temp that 80+ tests at. EPS reguires testing at both 115VAC and 230VAC.
I had to do the 230VAC round (which is a loosely regulated 220V/59.9hz with my setup) later after I had removed the fan voltage wire and exhaust temp sensor and didn't want to take the time to open it back up.
Gained quite a little bit at the low load and none at medium load and the difference at high load probably isn't outside my margin of error. There's 13 different measurements and 10 calculations to make to get an efficiency number for each load and too much margin for error for me to claim my numbers are accurate to two decimal places and the 80+ accuracy requirements in their test protocol aren't either.
Anyway the ATX and EPS recommendation and Energy Star requirement is >80% at all three loads which the Neo Eco passes...we knew that from the other tests huh?
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 520W total load is achieved. The load amounts and results are in the chart below for reference.
Looks like I went a good deal over 20W on the 5Vmax one because the 12V was much higher than nominal. (I just calculate my loads in advance figuring nominal...close enough). And then I went over on the 3.3Vmax because I messed up somewhere apparently.
Before I move on to the next test section I want to see how the unit handles loads in excess of 520W. I'm not going to take the time to find some kind of "peak power" figure since that would require too much time trying various combinations of 12V,5V and 3.3V loads and any real world overload is going to be 12V heavy anyway. I'll start with the 12Vmax loadout and increase the 12V load until either the voltages or ripple go out of spec or the unit shuts down...whichever happens first. A unit should never be loaded over it's 100% max rating in practice. The guide says 100% maximum continous rated loads (peak in their vernacular) on any voltage should last no longer than 17 seconds with not more than one occurence per minute. The only reason I'm going to do this test is to give me an indication about how close to the platform design limits the unit is being pushed and thus it's going to be more telling with the top of the range 620W version. Anyway the 12V dropped out of spec below 11.4V at ~48A load...total output ~620W.
Last edited by Makalu; 04-20-2010 at 22:57.