Meet Mr. Zippy

Power supply test – Phil Roth and Ross Lapkoff

Pic

If “Zippy” sounds like a name that belongs on a generic power supply, prepare to be enlightened! Quite contrary to the sound of the name, Zippy Technology Company was founded in 1983 and has been producing the high quality server and redundant power supplies since 1993. Given that standards for server and workstation power supplies are higher than those for consumer PCs, Zippy power supplies are higher in quality and reliability than those targeted for a non-professional market. To some extent, they are also higher in price; however, as high-end enthusiast and gaming power supplies continue increase in price, the gap is being narrowed.

Overclocked gaming or benching rigs with a pair of high-end video cards in Crossfire or SLi and a dual-core processor can greatly benefit from high-power, high-quality server supplies that produce clean power under the harshest conditions. In a lot of cases, even the strongest name brand “gaming” power supplies just can’t keep up with power demands of overclocked high-end systems. That type of situation is exactly what led to obtaining this 700W Zippy PSL-6701P from Servercase.com.

After a short three months of heavy use, a popular name brand 600W PSU failed. Compared to the failed PSU, the intent of the Zippy is clear: all work and no play. Instead of LED fans, shiny or painted cases and sleeved cables, you get a power supply that’s sole purpose is to power big systems reliably and the last time we checked, power supplies were a critical system component, not an accessory to make cases look better.

Pic

700W Zippy PSL-6701P with cover removed.

The brawn of the Zippy PLS-6701P is immediately apparent when comparing it to the previous PSU and if you could hold both, you would feel that beefiness with the Zippy weighing twice as much (10 pounds) as the old PSU (5 pounds). Clearly, the Zippy is quite a bit longer than the average PC power supply, which may be an issue if you plan to stick one in a mid-tower case at home.

ATX power supply design reference size is 140mm x 150mm x 86mm. Some manufacturers already take liberties with the length, stretching them out to 175mm, but the Zippy is hardly even close, measuring a very long 220mm x 150mm x 86mm, so make sure you have room for before purchasing one.

There are enough cables to power up anything you will ever need and alternate cable configurations for dual XEON processor platforms are available. Standard cabling supports ATX 2.03, ATX 12v(P4), EPS12V, WTX, AMD-GES:

  • 1 – 24 pin ATX 2.1 mobo connector
  • 1 – 8 Pin CPU power connector for Dual Core CPUs mobos
  • 1 – 4 Pin CPU power connector for Single Core CPUs mobos
  • 2 – PCI-E power connectors
  • 5 – Traditional Dual 1×4 power connectors

{mospagebreak}

Specs and Feature of the Zippy PSL-6701P

Ever wonder how a $30 power supply has the similar specifications as a $300 supply? There are many tricks in arriving at specification numbers, but rest assured that good looking specifications based on testing done with easy parameters does not make a power supply anywhere near as good as a PSU producing similar numbers under much more harsh conditions.

In the following specification chart you can see the massive current available on the 12V rail. Few enthusiast PSUs are over 36A for single 12V rails, so 45A is very impressive and useful, since a high-current 12V rail is crucial for overclocking today’s processors and top-end video cards.

You should also note that ATX specifications require no more than 120mV ripple for +12V. Ripple represents how “clean” the output power is and, as you can see, the Zippy’s maximum specs are 17% lower than the required maximum, indicating their commitment to and delivery of high-quality power. As you will see later when testing under real world conditions, measured ripple is substantially lower.

Pic

TOTAL OUTPUT POWER SHOULD NOT EXCEED 700W for PSL-6701P.

  • Operating Temperature: Operating 0°C – 40°C
  • Humidity: 10 ~ 90 % RH
  • Hold-up time: 16 ms minimum at full load & 90 VAC input voltage
  • Dielectric withstand:
    • Input / Output 1500 VAC for 1 minute
    • Input to frame ground 1500 VAC for 1 minute
  • Efficiency: 70% typical, at full load
  • Power good signal: On delay 100ms to 500ms, Off delay 1 ms
  • Overload Protection: 110 – 160% max
  • Over-volt Protection:
    • +5V: 5.7V – 6.5V
    • 3.3V: 3.9 – 4.3V
    • 12V: 13.6 ~ 15V
  • EMI: FCC Class B, CISPR22 Class B
  • Safety: UL 1950, CSA 22.2 NO/ 950, TÜV IEC 950
  • Remote On / Off control
  • Short Circuit Protection: Shutdown and latch
  • Remote sensing on 3.3V
  • Built-in Active PFC
  • I2C features (optional)

Powering up the unit revealed one very loud fan that is easily louder than an x1900 fan running at 100%. Keep in mind this is a server supply and not designed for the enthusiast market, so noise is not a consideration in the design. Why a single fan for a power supply this large? Multiple quiet fans to do the same job with a reduction in noise require more space. In most cases, making room for extra fans requires a compromise in design and construction of important internals, such as smaller heat sinks, caps and coils. As you will see, even with such a large size Zippy has still made full use of every square centimeter gained by using extremely hefty components in place of extra fans.

What’s Inside

Real power and reliability can only be obtained with massive, heavy components and the Zippy has the guts to back up its massive size and weight. Cracking the case revealed massive aluminum heat sinks and even for its size, a very compact layout of power components and circuit boards. As one would expect, the PLS-6701P appears to use high quality Nichicon capacitors and Intersil Power FETS along with beefy foil traces, unlike some supply manufacturers that solder buss wire to the traces to increase current capacity.

The Zippy uses a 12V 3600RPM Sanyo Denki/San Ace 80mm fan. The fan utilizes a standard 2-pin molex, making it a simple swap for a more reasonable noise level fan with a 2-pin connector. On this unit, a more reasonable sounding PABST 80mm fan was swapped in which required soldering due to the PABST’s flying leads.

Pic

Inside view of the PSL-6701P

Pic

Original Sanyo Denki fan

Performance

Checking the voltage with a DVM shows the 12v rail is at 12.15 during idle conditions. Since the 12v rail is the predominantly used power source for heavily overclocked and graphics systems, the plan was to load the PSU and find out how beefy the 12v rail was and what it would take to drag it under 12.0v. A load test connecting ballast resistors was done to check stability of the rail under various loads and find out when the overcurrent limits of the supply were reached and how well voltage regulation was held.

Pic

120 Watt (10AMP) ballast packs

Pic

300 Watt (25 AMP) ballast pack

The supply was run through its paces starting with an initial load of 10 AMPS, followed by incremental loading of 5 AMPs, until the overcurrent limit was reached. Voltage was logged after the load was applied and the supply was running for 3 minutes.

Voltage Reading

Power Draw

12.15

10A

12.14

20A

12.13

25A

12.13

30A

12.13

35A

12.12

40A

12.12

45A

12.11

50A

12.10

52A

NOTE: Overcurrent limit exceeded

At 52 Amps, the supply overcurrent limit tripped and the supply shutdown after 15 seconds. It’s possible this current may have been maintained for a longer period with the original fan, but we believe the supply held up exceptionally well above the nameplate current maximum of 45 AMPS. It should also be noted that the full load amperage rating was maintained with only 50 millivolts voltage drop from the minimum load draw.
{mospagebreak}

Next was an extended burn-in period of 8 hours at 45 AMPS which passed without failure.

Pic

Stress testing with ballast loads don’t produce the dynamic loading conditions on the supply as the rigors of a heavy workout in a highly overclocked, graphics bench environment. Although we did not have dual x19000s to simulate maximum draw that a crossfire setup demands, we ran an extreme benchmark with the 300W ballast load connected to check performance under real conditions.

Test Bed:

  • Intel 955 Extreme Edition
  • ASUS P5WD2-E, Vcore modded, Vdimm modded
  • Corsair TWIN2X1024-8000ul
  • ATI x1900xtx
  • Modded EXOS II with Alphacool 1510 pump
  • Danger Den RBX CPU block
  • Danger Den modded X18 block

First up was a 5.00 GHz OC and a SPi 32M run which resulted in a minimum reading of 12.12 on the DVM. Next was adding a 760 GPU/860 MEM OC on the x1900xtx and a quick bench with 3DMark05 which resulted in fluctuations between 12.10 to 12.12V. A try for Quad SPi32M running on 2 physical, 2 logical cores while benching 3DMark05, but the test rig cannot handle the load with only water cooling. Until some more extreme cooling allows for it, the Zippy’s beefy 12v rail has easily held up to a substantial 500W total load on it.

Speedfan Plot of Estimated 300+W Load During a Highly Loaded 2 Hour Workout

Pic

NOTE differences in ADC circuitry of the mobo measuring 12.04v – 12.1v whereas actual DVM readings were 12.10v – 12.12v

.

Clean power – we’ve mentioned it several times. In addition to a lower electric bill, another benefit of the Zippy was the lower temperature of the MOSFETS supplying power to the CPU. The Fluke 81/8020 DVM recorded an 8-12ºC decrease in temperature over the previous gaming PSU, as measured on the case of the MOSFETS. A quick scope check on the rail shows the likely reason for this.

Recall we had before pointed out that the maximum 100mV ripple specified on the 6701P was substantially less than the ATX specification limit of 120mV. Since the PSU specs are measured during testing with tougher conditions that seen by us, the actual ripple measured only 40mV during max load. That is only 40% of maximum specs for the PSU and only 30% of what is deemed acceptable by current standards. Clean power.

Scope Trace of Ripple During High Load – Estimated 300+ Watt

Pic

NOTE: 10x probes are used. Vertical Division = 100 millivolts.

Summary

The PSL-6701P is one of the finest single rail PSUs on the market today and has the performance and reliability for the PC enthusiast needing to exceed the 5GHz barrier.

It’s a great feeling to know you have the power behind you when pushing systems to the edge of their capabilities. The guys at Servercase.com are great and this unit shipped the same day the order was placed. Three emails were received during the day confirming the status of the order from the time it was placed until being sent out for delivery. The only negative comment on this PSU is the fan noise. Size cannot be deemed a negative factor since it is not being used for environment for which it was intended, but size should definitely be taken into account before purchasing one for a standard PC case.

Phil Roth and

Ross Lapkoff

Loading new replies...

Avatar of greenmaji
greenmaji

Senior Spellcheck

7,465 messages 0 likes

LOL.. I read the artical but had no clue that it was done by proth and Ross DOE! (sorry guys I didnt know your offline names)
Agreed with Gautam.. nice artical, Its nice to know it exists, and I'll keep it in my dreams.. :D

Reply Like

R
Ross

Senior Intel-lectual

3,838 messages 0 likes

Thanks guys :D There will be more reviews coming and they will get more in-depth as things progress.

I want to test my Zeus single rail 45A and see how it stacks up :) We now have access to a lot larger resistors and all I can say is I hope they are sinked. I haven't seen them yet, but they are rated up to 105W each :eek: They might make some great load testers for fine tuning a *cascade*, LOL.

Reply Like

Avatar of Joe Camel
Joe Camel

Senior Camel Kicker

3,509 messages 0 likes

that was a kick-*** review!!

ive got one (zippy 700) on-the-way to my door :)

ill be sure to post some results here (too)

from the MONSTER x1900 thread in the ATi section:
(post 1808)

hummmm i wonder if an even MORE powerful PSU would get me...HIGHER ? (;))

cant find anything with >36A 12v rail...even the MONSTER 1000W only matches that (although thats 1 of 3x 12v rails...the other 2x are lower Amps)

maybe i should look @ todays front page... :rolleyes:

a 45Amp (12v rail) psu

http://www.overclockers.com/articles1313/

:eh?: :bang head :shrug:

EDIT:
ordered...should be here early next week

:thup: great work guys!! :thup:

Reply Like

click to expand...
Avatar of larva
larva

Inactive Moderator

5,535 messages 0 likes

Thanks guys :D There will be more reviews coming and they will get more in-depth as things progress.

I want to test my Zeus single rail 45A and see how it stacks up :) We now have access to a lot larger resistors and all I can say is I hope they are sinked. I haven't seen them yet, but they are rated up to 105W each :eek: They might make some great load testers for fine tuning a *cascade*, LOL.

Just bear in mind that purely resistive loading is not a particularly realistic load. The real loads the PC power supplies must drive have capacitance and reactance as well as variability in magnitude not present in purely resistive loads. It's a good rule of thumb, but it's not the finite word on load-driving ability.

One thing it does do very effectively is stress test, and therefore debunk inflated power claims easily. If your $17 '400W' power supply turns into a pumpkin when confronted by a 400W resistive load, that is what you are looking for. The final assessment of the upper end of a good supply's output range still is best accomplished with a actual PC for the load. And if you test resistively long enough and compare the figures to the real-world results, you can hopefully grasp a relationship between the two. This is the point at which the test-bench numbers start saving you time (as opposed to just hooking it up and observing the results).

However you put hands on a supply, doing so improves knowledge of its qualities. Using it on the bench or in a machine are different things, but they both contribute to the understanding of the nature of a particular unit. The pictures of the internals alone is about all I really need to see to feel confident choosing one unit or another, so I appreciate the glimpse at what is obvsiously heavy machinery.

Reply Like

click to expand...
Avatar of proth
proth

Member

972 messages 0 likes

Just bear in mind that purely resistive loading is not a particularly realistic load. The real loads the PC power supplies must drive have capacitance and reactance as well as variability in magnitude not present in purely resistive loads. It's a good rule of thumb, but it's not the finite word on load-driving ability.

One thing it does do very effectively is stress test, and therefore debunk inflated power claims easily. If your $17 '400W' power supply turns into a pumpkin when confronted by a 400W resistive load, that is what you are looking for. The final assessment of the upper end of a good supply's output range still is best accomplished with a actual PC for the load. And if you test resistively long enough and compare the figures to the real-world results, you can hopefully grasp a relationship between the two. This is the point at which the test-bench numbers start saving you time (as opposed to just hooking it up and observing the results).

However you put hands on a supply, doing so improves knowledge of its qualities. Using it on the bench or in a machine are different things, but they both contribute to the understanding of the nature of a particular unit. The pictures of the internals alone is about all I really need to see to feel confident choosing one unit or another, so I appreciate the glimpse at what is obvsiously heavy machinery.

Larva, Really good point about the simple resistive loading not simulating real world conditions. That's why we benched with the 300W load attached. Granted it would have been better to pull the extra 300 watts with another x1900 or something, but that's all we had at the time. Perhaps future reviews will have switches or relays that throw in variable loads to check the PSU response time and any droop on the rail as the PSU has to build current in a hurry.

Reply Like

click to expand...
Avatar of Joe Camel
Joe Camel

Senior Camel Kicker

3,509 messages 0 likes

larva,

as a (former?) Rockford Fosgate employee, i KNOW you KNOW your PSUs. :cool:

could you shed any light in our ATi X1900 thread as to how/why some seemingly ample (pun intended :)) PSU's, such as the OCZ 600w PowerStream seem to produce less than satisfactory OC results when clocking these 12V AMP HUNGRY cards...

(ie. i lost almost 100 MHz on (both) my core and mem OC (X1900 xtx) with the OCZ600 vs the Sparkle 550)

(sorry for the off-topic nature of my post)

Reply Like

Avatar of larva
larva

Inactive Moderator

5,535 messages 0 likes

Larva, Really good point about the simple resistive loading not simulating real world conditions. That's why we benched with the 300W load attached. Granted it would have been better to pull the extra 300 watts with another x1900 or something, but that's all we had at the time. Perhaps future reviews will have switches or relays that throw in variable loads to check the PSU response time and any droop on the rail as the PSU has to build current in a hurry.

Well, I don't even have the zippy so I can excuse the lack of hot and cold-running X1900s :) But it is true that between phase change'd dual core Intels and X1900s real power draw has risen to stupifying porportions. And and all attempts to better quantify the full implications of this factor contribute significantly to understanding exactly how system designs must evolve to assure good results.

Reply Like

click to expand...
Avatar of larva
larva

Inactive Moderator

5,535 messages 0 likes

larva,

as a (former?) Rockford Fosgate employee, i KNOW you KNOW your PSUs. :cool:

I worked for RF in 1989-1990. It was pretty wild, with load resistors and Astron 12V DC power supplies everywere. They burnt in all amplifiers right behind the area where I worked, and occasionally you would get a bad batch of caps that would create a rather 4th of July-esqe episode.

could you shed any light in our ATi X1900 thread as to how/why some seemingly ample (pun intended :)) PSU's, such as the OCZ 600w PowerStream seem to produce less than satisfactory OC results when clocking these 12V AMP HUNGRY cards...

(ie. i lost almost 100 MHz on (both) my core and mem OC (X1900 xtx) with the OCZ600 vs the Sparkle 550)

Fair question, simple answer. The OCZ 600 is junk, the Sparkle 550 (and OCZ 520) is not. Topower is a really cheap OEM, and they have only really hit one home run, in my estimation--the supply we know as the Powerstream 520. The rest of their stuff just doesn't have the real output to match its specifications and user expectations.

The Sparkle 550 is super stout. It's loud and runs hot, but it's got real guts. Pics tell the story there. The 510 PCPower&Cooling is gutsy too, but doesn't appear to have quite the 12V capacity of the Sparkle. The Zues 560 is a strong unit, heavy machinery like the zippy. Pictures really do tell the story with these units, they make power the old-fashioned way.

The outlier here is the Powerstream 520. It doesn't look like much, but drives monumentally difficult loads in reality. Pics don't tell the whole story with this unit, but if you try the other supplies and then look at the pics you will see the general trend.

Reply Like

click to expand...
Avatar of Joe Camel
Joe Camel

Senior Camel Kicker

3,509 messages 0 likes

:beer:thanks for the input :beer:

OCZ600,
huh... good marketing and name recognition (OCZ's) trumps true performance (once again) :(

Reply Like

Avatar of Sentential
Sentential

Contributing Member

14,206 messages 0 likes

Great review guys if I had the money Zippy would be my PSU of choice

Reply Like