Reviews - NorthQ Black Magic Flex 650W PSU
Sample Provided by: NorthQ (By OklahomaWolf on Sun, Oct-26-2008)

Page 2 -

Before I get into the business at hand, today is special. And I don't mean Ralph Wiggum "special" either, though that could get interesting. I'm trying out a brand new test on this review, and you're about to see it in action. To my usual room temperature and hot box progressive load tests, I'm adding what can be called overshoot testing. Basically, I'm going to take this here unit and see if the output voltages show spikes at power up that go out of spec. I'll describe that a bit more further down the page.

Meantime, let's get the cold tests done. This will go as it always does, with me hooking the unit to my SunMoon SM-268. The 268 will load down the unit first in five progressive loads ranging from 20%-100% of full power, and then unbalance things first towards 3.3V and 5V, and then 12V. This will tell us how good the stability of the unit is, how efficient it is, and how well it likes being surprised by Grandpa's old dual Pentium III Xeon rig. Usually I choose my loads based on the combined ratings from the label, but there really aren't any of significance on this unit. So, I'll just guess.

Once the cold tests are done, I'll get into the new overshoot tests, and following them will be the usual hot box testing wherein I will roast the Black Magic alive.

Results from NorthQ BM-650 COLD load tests
Test # +3.3V +5V +12V1 +12V2 DC Watts/
AC Watts
Eff. Intake/
Exhaust
Simulated system load tests
Test
1
3A 3A 4A 4A 139W/
176W
79% 22°C/
30°C
3.33V 5.21V 12.40V 12.40V
Test
2
7A 7A 8A 8A 271W/
345W
79% 22°C/
36°C
3.33V 5.18V 12.36V 12.35V
Test
3
10A 10A 12A 12A 392W/
517W
76% 23°C/
44°C
3.32V 5.15V 12.30V 12.30V
Test
4
14A 14A 16A 16A 520W/
719W
72% 23°C/
45°C
3.30V 5.09V 12.26V 12.25V
Test
5
18A 18A 20A 20A 653W/
972W
67% 23°C/
60°C
3.27V 5.03V 12.22V 12.21V
Test
CL1
18A 18A 1A 1A 184W/
284W
65% 23°C/
39°C
3.31V 4.91V 12.59V 12.57V
Test
CL2
1A 1A 18A 18A 514W/
672W
76% 23°C/
43°C
3.33V 5.39V 12.10V 12.09V

And we're off like a herd of turtles running through a tar pit. One thing is very, very clear to me right now - this ain't no 650W power supply here. Look at the efficiency in test five... 67% is terrible! As if that weren't enough, the exhaust temperature is a toasty sixty degrees. The unit is clearly running beyond its intended capabilities here.

That aside, the voltage readings actually don't look too awful, unless... you look at test one. Doh! The 5V is very nearly out of ATX spec in that test at 5.21V. It's way out of spec in test CL2. The 12V regulation is awful as well, flirting with ATX spec again in test CL1 and only really behaving itself in test CL2 and the five progressive tests. This means one thing - this is a group regulated design. Every rail but the 3.3V starts off too high for my liking, and it's pretty clear to me that this was an effort on Kingnod's part to keep the rails from getting below mean value.

There's a big problem with trying to overvolt the 12V and 5V of a group design to maintain readings above mean value, however. That problem is the 5V reading in test CL2. These are loads that could actually come up in an everyday situation these days with modern hardware, and that rail is so far out of spec (almost 8% from mean value!!!) that I'm not sure I'd want it in or even near my system. I see what you tried to do there, Kingnod, but it doesn't work. Stop it. Or switch the design to independent regulation... I know you can do those designs well, I've seen you do it. Otherwise, just let a group design be a group design and turn down the 5V and 12V rails. I'd rather see the 12V readings in test five get below 12V than look at the 5V rail heading for outer space when crossloaded.

Speaking of efficiency, do you see 80% up there? I don't. While it is likely that this unit could pull down the 80% number at 230V input voltage, which would make sense for the market this unit is intended for, I dislike claims like this that lead you to think it could be as efficient at 120V. Or, if not that, that it could be that efficient over its entire operating range. Going by these numbers, at 230V, you would only likely see 80% once or twice, and certainly not at full power where this unit falls flat on its face.

And the fan? Loud. Just... loud. Every bit as loud as big brother. On to the new tests, and then I'll see how this thing does in the hot box. If it survives the hot box.

Ok, here's how the overshoot tests are going to go. First, I'll test the 5VSB rail for voltage overshoot when power is first applied to the unit. Then, I'll have the SM-268 power on the unit from standby using test five's load levels (650W) and measure the 12V rail for spikes. Finally, I'll power the Black Magic on using the power switch on the back using test five's load levels and measure 12V again. The ATX spec calls for any spikes to be within 10% of mean value, and no inverted polarity (that would be real bad). What I really should be doing is measuring all active rails for data on overshoot tests two and three, but I'm a lazy man. So, y'all will have to settle for 12V only.

I'm doing these tests between the cold and hot tests for one reason, and one reason only. I don't think this unit will survive the hot box. And it certainly wouldn't be the first or last to fail in the hot box either. Best to do these tests while the unit can still handle them, since I will be going from no load to full load several times in just a few seconds.

Overshoot Transient Testing

VSB On

VSB to Full

Off to Full

There's some good news here to balance out the bad news from the cold tests. That good news is, nothing goes out of spec. To put it childishly, the unit refuses to color outside the lines. To put it even more childishly, goo ga ga goo goo ga goo ga ga goo. Giggidy.

The 5VSB rail gets a wee bit close at 4.7V, but the 12V rail barely gets any higher than 6V. Now, keep in mind that these shots aren't testing for rise time, so we're not looking to see how fast they go on to hit the voltage readings for test five themselves, we're just looking at that initial turn on spike that all units have and all designers seek to control.

While the ATX spec does allow for spikes 10% above mean value on these tests, I confess I'm going to be meaner than that in the future. Just being in spec isn't good enough for yours truly on this test. I want to see it stay below mean value. Points come off if a unit can't do that. Oh yeah... I'm hardcore baby. Pay no attention to that Pound Puppy over there. That's... uh... someone else's stuffed toy.

Let's get the hot box in gear, and heat this puppy up. Not the Pound Puppy, that would be just wrong, sick, and cruel. Shame on you for thinking I could do such a thing.

Results from NorthQ BM-650 HOT load tests
Test # +3.3V +5V +12V1 +12V2 DC Watts/
AC Watts
Eff. Intake/
Exhaust
Simulated system load tests
Test
1
3A 3A 4A 4A 139W/
177W
79% 26°C/
30°C
3.33V 5.22V 12.40V 12.41V
Test
2
7A 7A 8A 8A 271W/
347W
79% 31°C/
38°C
3.34V 5.19V 12.36V 12.36V
Test
3
10A 10A 12A 12A 393W/
523W
75% 32°C/
39°C
3.32V 5.15V 12.32V 12.31V
Test
4
14A 14A 16A 16A 522W/
729W
72% 36°C/
52°C
3.30V 5.09V 12.28V 12.27V
Test
5
18A 18A 20A 20A N/A N/A N/A
FAIL FAIL FAIL FAIL
Test
CL1
18A 18A 1A 1A N/A N/A N/A
FAIL N/A FAIL FAIL
Test
CL2
1A 1A 18A 18A N/A N/A N/A
FAIL FAIL FAIL FAIL

And there it goes! About a minute into test five, I heard the suspicious sounds of the AVR compensating for the line voltage taking a puke, and suddenly the Black Magic 650W was no more. You remember me saying I had two samples, right? Well, they both failed in similar fashion. Although, my first one actually failed at room temperature. Only this one waited for the hot box. Craptastic. And look, the efficiency plummeted by test four again, reinforcing the overrated idea yet again.

You know, it's just too bad. Kingnod seems to be good at what they do, even if they had this thing tweaked for too high voltages. I mean, look at test four... it's not doing too bad there at all. The 3.3V rail is dead on target, the efficiency is... well ok, the efficiency isn't good. But the steady decline in the efficiency numbers does tell me that this would make a decent budget 500W unit. And indeed, the specs on this unit are more in line with Kingnod's own 500W units.

Too bad it's... you know. Overrated. Because I have to score it as a 650W. Let's see what the scope shots from the load testing look like.

Oscilloscope Results - NorthQ Black Magic 650W

Test #

+3.3V

+5V

+12V

Test
1

Test
2

Test
3

Test
4

Test
5

Test
CL1

Test
CL2

Now, this really is a sad sight, because Kingnod really is good at keeping ripple under control. How does 30mV or less on all rails grab you? Heck, look at that - even at 650W, running 150W over what I'm thinking the unit should be rated for, it's still well under control. This is excellent performance, even if we have some funky waveforms up there.

Just a quick note before I go on to the autopsy of this unit. I've had some criticism lately on my scope shots, and how I don't take them at the full 20MHz spec. My response to this is, all scope readings were and always have been confirmed at ATX spec. I just don't take my screen shots at the spec, because often the results don't look any different and the shots just turn out better at 25 kS/s. That said, sometimes I do see some oddness in the waveform that requires me to fool around with the scope shots to show you. But, I didn't get any of that this time. So, I'm moving on now.



JonnyGURU Feeds
Add our RSS feeds to your favorite RSS Reader or homepage.

JonnyGURU Reviews
AddThis Feed Button

JonnyGURU News
AddThis Feed Button

JonnyGURU Articles
AddThis Feed Button

JonnyGURU Forums
AddThis Feed Button

Follow us onFollow Us!
 
Site-Info | Affiliates | Contact Us | Site-Map
Copyright © 2012 - JonnyGURU.com