OCZ ProXStream 1000 W Power Supply Review
By Gabriel Torres on February 9, 2008
With systems with multiple video cards requiring more power everyday, 1,000 W power supplies are becoming very common. OCZ has released a new 1,000 W product on the market, ProXStream 1000 W, targeted to power-hungry consumers that are willing to put three or four high-end video cards (this unit has four six-pin VGA power connectors) and several hard disk drives inside their systems. What is really different about this power supply compared to other 1,000-watt units around is it uses a small form factor, being at the same size of any standard ATX power supply, due to its interesting internal design using two printed circuit boards stacked. We completely disassembled this new unit from OCZ to see how it looks inside and what design and components were used, plus we put it on our load tester to see whether it can deliver its rate 1,000 W or not.
As we mentioned, the first thing that caught our attention was the physical size of this power supply, using the standard ATX size, i.e., being smaller than other 1,000 W power supplies available on the market. This was accomplished by using two printed circuit boards stacked inside the unit (you can have a quick glimpse of this design in Figure 3).
Because size was apparently one of the main concerns when designing this unit, the manufacturer used a standard 80 mm fan on the rear side of the power supply, just like a traditional ATX power supply (see Figure 2). This was done because there is not enough room for a fan on the bottom side of the product, since there is no available space inside the unit for anything else. Also, since this power supply uses two printed circuit boards, a fan located on the bottom of the unit wouldn’t cool the lower printed circuit board, and a fan located at the rear can cool down both boards.
Because of this very compact design will all circuitry squeezed in a very small form factor using a small fan we wondered if this unit wouldn’t suffer of any thermal issues. This is something we will play close attention during our tests.
On the front side of the unit we have a big mesh allowing air to enter the unit. You can see that there are two printed circuit boards inside the unit looking through this mesh.
This unit features active PFC (Power Factor Correction), which provides a better usage of the power grid and allows this power supply to comply with the European law, making OCZ able to sell it in that continent (you can read more about PFC on our Power Supply Tutorial). In Figure 2, you can see that this power supply doesn’t have an 110V/220V switch, feature available on power supplies with active PFC.
We assume that this power supply has a high efficiency (at least 80%) but what is strange is that precise information about efficiency isn’t available on OCZ’s website, on the product manual or on the product box. During our tests we will measure efficiency and we will see if there is any particular reason that OCZ isn’t talking about efficiency.
High efficiency power supplies consume less power from the power grid – an 80% efficiency means that 80% of the power pulled from the power grid will be converted in power on the power supply outputs and only 20% will be wasted, meaning a lower power bill – compare to below 70% on regular power supplies.
This power supply comes with eight peripheral power cables: four PCI Express auxiliary power cables for feeding up to four video cards, two peripheral power cables containing three standard peripheral power connectors and one floppy disk drive power connector each and two Serial ATA power cables containing three SATA power connectors each.
This power supply also comes with a 20/24-pin motherboard cable, one ATX12V cable and one EPS12V cable.
All wires are 18 AWG and we think OCZ should have used 16 AWG wires, which are a little bit thicker.
The finishing of the cables coming out of the power supply – something we always criticize – is just perfect, with the use of plastic sleeves coming from inside the unit.
Now let’s take a look at how OCZ ProXStream 1000 W looks like inside.
We decided to disassemble this power supply to see what it looks like inside, how it is designed, and what components are used. Please read our Anatomy of Switching Power Supplies tutorial to understand how a power supply works inside and to compare this power supply to others.
This page will be an overview, and then in the following pages we will discuss in detail the quality and ratings of the components used.
We can point out several differences between this power supply and a low-end (a.k.a. “generic”) one: the construction quality of the printed circuit board (PCB); the use of more components on the transient filtering stage; the active PFC circuitry; the power rating of all components; the design; etcetera.
As we mentioned previously, this power supply uses two printed circuit boards stacked, as you can see in Figure 5.
On the lower board we have the primary section of the power supply plus the +5VSB power supply – this output, also known as standby, is always independent from the rest of the power supply, being produced by a separated circuit, because this output is turned on all the time, contrary to the rest of the power supply, which is turned on only when you turn on your PC. The primary section includes the transient filtering stage, the active PFC and the switching transistors.
On the upper board we have the transformer and the secondary stage, which includes rectifying, filtering and protection stages.
On this page we will take an in-depth look at the primary stage of this power supply.
As mentioned the primary stage is located on the lower printed circuit board, together with the +5VSB power supply. This stage includes the transient filtering stage, which have some components soldered directly on the power cord plug.
As we mentioned on other articles, the first place we look when opening a power supply for a hint about its quality, is its filtering stage. The recommended components for this stage are two ferrite coils, two ceramic capacitors (Y capacitors, usually blue), one metalized polyester capacitor (X capacitor), and one MOV (Metal-Oxide Varistor). Very low-end power supplies use fewer components than that, usually removing the MOV, which is essential for cutting spikes coming from the power grid, and the first coil.
This power supply has far more components than necessary for this stage, as you can see on Figures 8 and 9. For a clearer picture we have removed the rectifying bridge, the PFC diodes and the four transistors (two for the switching section and two for the active PFC) in Figure 9. Also in Figure 9 we are describing some components that are not part of the transient filtering stage just for your learning experience.
In Figure 10 we are showing a complete view of the bottom printed circuit board also with the abovementioned components removed to show you the other components found there. On the upper right corner you can see the +5VSB power supply (a.k.a. standby power), which is independent from the rest of the power supply, as it keeps supplying its voltage even with the power supply shut down. You can easily recognize its own transformer and its own voltage regulator integrated circuit (the component attached to a heatsink). And on the lower right corner you can see the integrated circuit that controls both the PWM circuit and the active PFC circuit: an ML4800CP.
All the big electrolytic capacitors found on this printed circuit board are from the active PFC circuit and they are connected in parallel – which means that they add up their capacitance: since there are six 100 µF capacitor in parallel, the total capacitance is of 600 µF x 450 V. Other power supplies usually use just one or two capacitors with the same total capacitance and voltage, thus the choice of using six smaller capacitors than just one or two bigger capacitors has more to do with the space available inside the power supply – the use of a single but bigger capacitor probably wouldn’t allow the use of two printed circuit boards stacked. By the way, all these capacitors are rated at 105° C and are Japanese, from Rubycon. Unfortunately on the secondary this power supply doesn’t use capacitors from Rubycon, but from Taiwanese OST.
This power supply uses a D15XB60 rectifying bridge, which can deliver up to 15 A, rated at 100° C. This is more than adequate rating for a 1,000 W power supply. The reason why is that at 115 V this unit would be able to pull up to 1,725 W from the power grid; assuming 80% efficiency, the bridge would allow this unit to deliver up to 1,380 W without burning this component. Of course we are only talking about this component and the real limit will depend on all other components from the power supply.
It also uses two power diodes connected in parallel on its active PFC circuit, where the normal is using just one. The rectifying bridge and the two PFC diodes are located on the same heatsink, which you can see in Figure 11.
The active PFC circuit uses two 20N60C3 power MOSFET transistors and the switching section uses two more of them. Several other power supplies from several different power ranges, like Cooler Master iGreen Power 430 W, Antec Neo 550 HE, OCZ StealthXStream 600 W, Zalman ZM600-HP, Corsair HX620W, OCZ GameXstream 700 W and Thermaltake Toughpower 750 W, just to name a few, use these same transistors on their active PFC circuit. On the switching section, however, only OCZ ProXStream 1000 W uses them; these other power supplies use other transistors, some with higher specs.
These 20N60C3 can drive up to 20 A at 25° C or up to 12.5 A at 100° C in continuous mode (see how the maximum current drops a lot depending on the temperature; since power is directly proportional to current, drop in maximum available current also means drop in maximum available power) or 60 A in pulse mode, which is the one used.
The two switching transistors are using the traditional two-transistor forward configuration. A detailed description of this configuration including schematics can be found on our Anatomy of Power Supplies tutorial.
The four 20N60C3 are found on the same heatsink, as shown in Figure 12.
Now let’s take a closer look at the secondary stage from this power supply.
As we have already explained, the main transformer and the entire secondary section from this power supply are located on its upper printed circuit board. We have already posted a picture of this board in Figure 7, but in Figure 13 you can take another look of this board with all its heatsinks removed.
This power implements a synchronous topology on its secondary. On this topology the diodes are replaced by power MOSFET transistors. The idea, at least in theory, is to increase efficiency. This is achieved by reducing the waste produced by the rectifying components: while Schottky diodes have a typical voltage drop of 0.5 V (i.e., voltage wasted by the component), power MOSFET transistors have a typical voltage drop of 0.1 V or even less. Unfortunately, however, this unit provided a not so good efficiency, as we will discuss in more details later.
This power supply uses nine FDP047AN08A0 power MOSFET transistors for the synchronous rectification (two for each main positive voltage), each one being able to drive up to 15 A at 25° C in continuous mode or 80 A in pulsating mode (which is the mode used), also rated at 25° C. This equals to 73 A at 50° C, 62 A at 85° C or 56.5 A at 100° C, calculated using the formula present on the datasheet from this transistor.
As you can see, the maximum current a semiconductor can deliver varies with its working temperature. This is why it is so important to know at which temperature the manufacturer labeled their power supplies. When not specified usually the power supply is rated at 25° C, a temperature the power supply will never work under (a typical working temperature is around 40° C), meaning that the maximum labeled power will only be reached on the manufacturer’s lab with the PSU internal temperature put at 25° C but never at your home, where your PSU will be running hotter.
By the way, we couldn’t find any mention to the temperature used to label this power supply on OCZ website, on the product box or on the product manual. We will discuss more about this in the next page.
The maximum theoretical current each line can deliver is given by the formula I / (1 - D), where D is the duty cycle used and I is the maximum current supported by the rectifying diode (which in this case is made by one transistor with a 56.5 A limit at 100° C - we will have to use the value at 100° C for a fair comparison because usually rectifiers are rated at this or greater temperature). Just as an exercise, we can assume a typical duty cycle of 30%. This would give us a maximum theoretical current of 81 A for each output, i.e. 969 W for the +12 V output, 404 W for the +5 V output and 266 W for the +3.3 V output. The maximum current each line can really deliver will depend on other components, in particular the coil used.
Other three transistors are used for other functions, so a total of nine power MOSFET transistors can be found on the secondary from this power supply.
On this power supply the +5V and +3.3 V outputs use independent transformer outputs. Usually on high-end power supplies these two outputs use separated rectifiers but connected to the same transformer output, which limits the maximum current each output may reach.
Now let’s talk a little bit more about the actual power specs from this power supply.
In Figure 15, you can see the label from this power supply containing all its power specs.
The +3.3 V output is labeled at 28 A or 92.4 W and the +5 V output is labeled at 30 A or 150 W. Funny enough the label says that the maximum combined power for these two outputs is of 150 W – which is strange, as on this power supply these two outputs don’t share any components. Maybe the unit was labeled this way just to fulfill ATX standard.
As you can see, this power supply has four +12 V rails, each one labeled at 20 A or 240 W. This would give a total +12 V power of 960 W but the label says that the maximum combined +12 V current is of 70 A, which equals to 840 W.
Adding this 840 W to the 150 W combined power of the +3.3 V and + 5 V we have 990 W, but the label says that the maximum combined power for all these outputs is of 976.5 W. Adding this to the +5VSB maximum power (17.5 W) and to the -12 V maximum power (6 W) we have exactly 1,000 W.
As we mentioned OCZ didn’t post at what temperature they labeled their power supply but don’t worry, we will test this during our load tests.
As for power distribution, as mentioned this unit has four +12 V virtual rails, +12V1 (yellow wire with black stripe), +12V2 (solid yellow line), +12V3 (yellow wire with blue stripe) and +12V4 (yellow wire with green stripe). On OCZ ProXStream 1000 W the power distribution is the following:
We think that there are two problems here. The EPS12V is the only connector on +12V1. So if your motherboard has an EPS12V connector, use it instead of the ATX12V connector. This will provide a better power distribution. For the regular user with a motherboard with just one CPU a better power distribution would be with the ATX12V connector on +12V1 together with the EPS12V connector, but probably OCZ did this way thinking on users with two CPU systems.
The second problem is with the naming of the PCIE connectors. If you have only two video cards on your system using SLI or Crossfire, you will probably connect PCIE1 plug to the first card and the PCIE2 plug to the second card. After all, that is the way it is labeled. You will get a better power distribution if you connect PCIE1 cable to the first card and PCIE3 cable (and not PCIE2) to the second card. We think that OCZ/FSP should have labeled these connectors differently (i.e., labeling PCIE2 as “PCIE3” and PCIE3 as “PCIE2”).
Why is this so important? The maximum labeled current for each rail is the level when the OCP (Over Current Protection) kicks in (actually it enters in action at a level a little bit above of what is labeled). If you have two very high-end video cards connected to the same rail and if they together consume more than 240 W, the overcurrent protection will shut down your power supply, even with your computer theoretically running inside its specs. Putting each card on a separated rail prevents this from happening, as you will now have a limit of 240 W for each card (480 W total).
We conducted several tests with this power supply, as described in the article Hardware Secrets Power Supply Test Methodology. All the tests described below were taken with a room temperature between 45° and 49° C. During our tests the power supply temperature was between 50° and 68° C (we will talk more about this very high temperature level below).
First we tested this power supply with five different load patterns, trying to pull around 20%, 40%, 60%, 80%, and 100% of its labeled maximum capacity (actual percentage used listed under “% Max Load”), watching how the reviewed unit behaved under each load.
For the fifth test, where we wanted to pull 1,000 W from this power supply, we ran into a problem not related to the power supply, but with our load tester. Our tester (a SunMoon SM-268) has two +12 V inputs, each one with a limit of 400 W or 33 A. For this particular test we wanted to pull more than 33 A, but we couldn’t, as we were limited by our equipment. So in order to achieve 1,000 W we had to increase the amount of current (and thus power) we pulled from +5 V and +3.3 V outputs. This wasn’t the ideal scenario (we wanted to put the +12 V outputs at 36 A and the +5 V and +3.3 V outputs at 14 A instead of 22 A) because nowadays a very high-end PC will concentrate its power on +12 V outputs and not on +5 V and +3.3 V. This happens because the video card auxiliary power cables have only +12 V wires, and the reason high wattage power supplies exist is to feed computers with two or more very high-end video cards.
On the other hand we were pulling a total of 66 A (800 W) from the +12 V outputs and the power supply label says the +12 V limit is 70 A (840 W), so we were very close.
In the table below we list the load patterns we used and the results for each load. +12V2 is the second +12V input of our load tester and on this test it was connected to the power supply EPS12V connector, which is the only connector on the power supply +12V1 virtual rail. Keep in mind that power supply uses a multiple rail design with four virtual rails and during our tests the first rail (+12V1) was connected alone to the +12V2 input of our machine, while +12V2 and +12V3 rails were connected together on the +12V1 input from our tester. The power supply +12V4 rail wasn’t connected to our tester, as it has only one video card power connector input. See previous page for more details on how the four rails are distributed on this power supply.
8 A (96 W)
14 A (168 W)
22 A (264 W)
30 A (360 W)
33 A (396 W)
8 A (96 W)
14 A (168 W)
22 A (264 W)
28 A (336 W)
33 A (396 W)
2 A (10 W)
6 A (30 W)
8 A (40 W)
10 A (50 W)
22.5 A (112.5 W)
2 A (6.6 W)
6 A (19.8 W)
8 A (26.4 W)
10 A (33 W)
22 A (72.6 W)
1 A (5 W)
2 A (10 W)
2 A (10 W)
3 A (15 W)
3.5 A (17.5 W)
0.5 A (6 W)
0.5 A (6 W)
0.5 A (6 W)
0.5 A (6 W)
0.5 A (6 W)
% Max Load
Ripple and Noise
Even though this power supply passed on our load tests, we have some important comments. First our full load test wasn’t done as we liked, as already explained.
Second, this power supply efficiency dropped below the 80% mark on tests four (800 W) and five (1,000 W) and that is the reason OCZ isn’t advertising this power supply efficiency. While when pulling 800 W efficiency was at almost 80% (79.6%) at full load efficiency dropped to 74.5% and we were pulling 1,335 W from the power grid to produce 1,000 W, completely wasting 335 W of power just to have our system running.
Also during our fourth and fifth tests we started hearing a high pitch noise coming from the power supply. We conducted some tests to discover when exactly this noise starts and we found out that whenever you pull more than 22 A from 12 V you will start hearing this noise. It isn’t loud and since at this time the power supply fan will be spinning at its full speed anyway – making a lot of noise –you won’t hear it.
But the greatest problem with this power supply was its temperature. We suspect that because of its reduced size and the use of an 80 mm fan – which is too small for a 1,000 W power supply – this unit would overheat. Our suspicion proved to be right. With other power supplies we tested the power supply temperature was only one to three degrees Celsius above room temperature – for example, 47° C when the room temperature was 45° C. But with this power supply this didn’t happen. On the first test the difference was of five degrees, on the second test the difference was of four degrees, on the third test the difference was of seven degrees and on the fourth test the difference was of nine degrees. But the real problem was when pulled the full 1,000 W from the power supply: the unit reached 68° C while the room temperature was of 49° C.
Electrical noise was within specs, reaching 68 mV peak-to-peak on +12 V outputs during the fifth test. Some people would prefer to see something below 60 mV here, but we think noise was fine, especially because we were pulling the full 1,000 W from this unit. On lower loads noise was also lower (45 mV with 60% load, for example).
After these tests we tried to pull even more power from OCZ ProXStream 1000 W. The problem was that the +12 V outputs were already maxed out because of the limit of our machinery. So the only option we had was to increase current/power on +5 V and +3.3 V, which would allow us to pull more power from this unit, but wasn’t reflecting the reality, as a high-end gaming PC wouldn’t pull that much from these two outputs. We did this only as an exercise – as we are explaining, these results are flawed as we should had pulled more current from the +12 V outputs.
Below you can see the maximum amount of power we could extract from this unit keeping it working with its voltages and electrical noise level within the proper working range. When we tried to pull more than 24 amps from +5 V and +3.3 V voltages were still within specs, but electrical noise increased to levels we didn’t want. At 25 A noise at +5 V output was at 48 mV, touching the 50 mV limit, so we can’t consider this result as valid since it isn’t safe. And above 25 A noise was above 50 mV. At 24 A noise level at +5 V output was at 22.8 mV peak-to-peak and at +12 V output noise was at 70 mV.
33 A (396 W)
33 A (396 W)
24 A (120 W)
24 A (79.2 W)
3.5 A (17.5 W)
0.5 A (6 W)
% Max Load
Under this condition the power supply temperature was at 68° C, room temperature was at 52.7° C and we were pulling 1,363 W from the power grid.
Short-circuit protection for both +5 V and +12 V outputs worked just fine, but it seems, however, that this power supply doesn’t have over current protection (OCP), or it is set way over 33 A – while according to the power supply label the limit for each +12 V rail is of 20 A. As you saw we above pulled 33 A from the EPS12V connector which was connected to the +12V2 input of our load tester and the power supply kept working just fine instead of shutting down.
As for the over power protection (OPP) we couldn’t test it because of the 33 A limit of our load tester.A feature we could see in action was the fan speed changing depending on the power supply temperature. But since this unit works at very high temperatures during our tests its fan was spinning at full speed all the time, producing a very loud and unpleasant noise.
OCZ ProXStream 1000 W power supply specs include:
* Researched at Shopping.com on the day we published this review.
Like we have already said several times, our preferred architecture for power supplies starting at 1,000 W is the design using two transformers, making the power supply to have two real and completely independent +12 V rails. This provides higher current limits and also makes sure that the transformer isn’t the component limiting the maximum current the system can draw from the power supply.
Even though we still think that, we were really surprised by the architecture used on OCZ ProXStream 1000 W - synchronous -, which uses high-current power MOSFET transistors to make the rectification instead of using power Schottky rectifiers. In theory this provides a higher efficiency, but the reality was very different.
This architecture allowed OCZ ProXStream 1000 W to truly deliver 1,000 W under a room temperature of 50° C, which is outstanding – but, unfortunately, this isn’t translated into an outstanding product.
At a first sigh creating a 1,000 W power supply using the same standard ATX size seems to be a great idea, but the problem is that power supplies on this range simply don’t fit such small form factor. OCZ opted to use an intricate design using two stacked printed circuit boards and one small 80 mm fan on the rear, leading this product to face serious thermal issues: when pulling the full 1,000 W the temperature of the power supply skyrocketed to 68° C (this is so hot that you can actually burn your fingers by touching it).
Based on that we think that for 1,000 W power supplies the best solution is still use a bigger form factor with a 120- or 140 mm fan installed on the bottom of the unit.
The 80 mm fan wasn't capable of correctly dissipating the amount of heat produced by the power supply and also generated a lot of noise, as it was always spinning at its full speed during our tests.
While this power supply could achieve an efficiency on the same level of other high-efficient power supplies (83-84%) when it was loaded up to 60% of its capacity (i.e., up to 600W) when we loaded it with 800 W its efficiency dropped to 79.6% and at full load its efficiency dropped to 74.5%.
Even though it is cheaper than other 1,000 W power supplies available on the market and it can truly deliver 1,000 W at 50° C we simply can’t recommend this power supply mainly because of its working temperature, since it will overheat your system.
As a final reminder, if you buy this unit and will use it with just two video cards under SLI or CrossFire, make sure to use PCIE1 and PCIE3 connectors (and not PCIE1 and PCIE2) for a better power distribution.