by Oleg Artamonov
06/25/2009 | 01:09 PM
The products to be discussed in this review are the senior models of power supplies from three different manufacturers. They are senior not only in terms of wattage rating but also in terms of technologies employed. The three makers claim that these PSUs are cutting-edge products with innovative and even unique circuit design solutions. I can’t help seeing into the matter with my own eyes because today’s peak of progress is tomorrow’s mainstream.
The promised innovations are not merely cosmetic like a new coloring of the cooling fan or something. They lie deeper, on the level of circuit design and switching power supply fundamentals. Therefore I will expand on the most interesting things as best I can. I will not just quote promo materials (for example, those of the Seasonic model list the honeycomb structure of the vent grid among its key features although 95% or something of mainstream and top-range PSUs have it) nor will I dismantle the PSU into its constituents and enumerate the transistors and diodes employed (this is a laborious but rather useless job because the brands of transistors can only be interesting for people who repair PSUs). I will instead focus on the features of the PSUs’ circuit design that differentiate them from yesterday’s products and will illustrate them with circuit diagrams (somewhat simplified in comparison with the real PSU). This will help you understand not what transistors are in the PSU but rather why they are in there and why the PSU manufacturer calls this an advantage.
Some basic knowledge of electronics, at least about the purpose of electronic schematics and the meaning of individual components in them, is necessary to understand the appropriate sections (called Circuit Design in the PSUs’ descriptions).
But if you are just looking for a good PSU and do not want to bother about its in-depth peculiarities, you can just skip over the electronics-related sections because the rest of the review is written according to our traditional easy-to-comprehend style.
Click the following link for a description of our testing methodology and equipment and a brief explanation of what the specified and tested parameters of power supplies mean: X-bit Labs Presents: Power Supply Units Testing Methodology In-Depth. If you feel overwhelmed with the numbers and terms this review abounds in, refer to an appropriate section of the mentioned article for explanation.
As we have promised, from now on we will mark the actual power consumption of several full systems on the cross-load diagrams, so that you could better estimate if the given power supply unit suits better for any of these configurations. We will mark only the maximum recorded power consumption during simultaneous launch of Prime’95 and FurMark programs.
There will be up to three marks on each diagram corresponding to power consumption of three system configurations as discussed in our article called PC Power Consumption: How Many Watts Do We Need?:
You can read more about the testing methodology and systems configuration in the above mentioned article. If the system power consumption is higher than the PSU capacity, it is not marked on the diagram.
You can also go to our Cooling/PSU section to check out reviews of other PSU models we have tested in our labs.
Although there are PSUs with higher wattage ratings in Antec’s product line-up, it is the Signature models that are officially positioned as the top of the range. The manufacturer promises stability, high wattage, silence and highest efficiency because the Signature PSUs comply with the 80 Plus Bronze standard (i.e. they are 82% or more efficient at loads of 20% through 100%).
The actual maker of the PSU is Delta Electronics.
The product box has an original design, being made from thick black cardboard with a bright yellow band in the middle. There are golden letters “Antec” on the top panel while the side panel has a barely visible pressed-out text that reads “Signature 850 watt power supply”. There is no other information on the box.
Besides the PSU, the box contains an installation guide, a set of detachable cables, a power cord, and four screws.
The PSU case is painted a black matte paint and has a length of 180 millimeters. The golden letters “Antec” are not painted – it is a separate plate glued into the depression in the top panel.
There are four connectors on the back panel: two for graphics cards and two for peripherals. The +12V line the connector is attached to is marked nearby.
The Signature has a somewhat nonstandard circuit design. Its electronics are placed on two full-size cards facing each other and located at the opposite sides of the case.
The right card carries an input filter (in the top left of the photo above), a standby source (in the bottom left), and active PFC with a rectifier and high-voltage smoothing capacitors (in the right part of the card). Since these components are all placed on a large individual card, the mounting density is not as high as in most other PSUs of similar wattage.
In the left part of the card you can see an electromagnetic relay (the rectangular thing in a brown case), which has become popular in top-end PSUs as a means to increase efficiency. The purpose of the relay is to completely cut high voltage off the active PFC device’s input when the PSU is shut down. This increases the PSU’s reliability (no voltage is applied to its components when not necessary) and somewhat lowers its power consumption in sleep mode (when only the standby source is active).
The second card carries a power transformer with its switch (the transistors on the small heatsink), rectifiers (the diode packs on the long heatsink stretching through the entire card), output LC filters, an output voltage & current control circuit, and two DC-DC converters whose job is to transform +12V into +3.3V and +5V.
Such converters are going to become widespread in power supplies and advertising materials, so I will dwell on them a little here.
Let’s start with the basics. A simple switching converter looks like that:
The high-voltage section (to the left of the transformer T1) is just a sketch. The input voltage of 400V is shown for PSUs with active PFC. PFC-less power supplies have a lower input voltage, about 310V. The high-voltage section represents a forward converter design which is currently highly popular among PSU developers.
The PWM controller controls the transistor Q1, switching it with a frequency of a few tens of kilohertz. The transistor is connected to the transformer T1 which lowers the voltage and isolates the PSU’s high-voltage circuitry from low-voltage one. The current impulses through the left diode of the pack D1 are charging the capacitors C1-C3 of the output filter and the choke L1 (the capacitors accumulate energy as an electric field whereas the choke, as a magnetic field), and the current passes through the load connected to the PSU. Between impulses the choke is discharged through the right diode of the pack D1 and the current passes through the load again. The choke L2 has some inductance and is only necessary to suppress high-frequency interference.
Thanks to the capacitors the voltage in the load varies in a very small range, rising at the moments the impulses come and falling in between them. But if the impulses get shorter, the voltage average gets lower and vice versa. Thus, there is an opportunity to control the PSU’s output voltage by changing the duration of the on-state of the transistor Q1 for each impulse. And connecting feedback from the PSU’s output to the PWM-controller we can not only control the output voltage but also make the controller keep that voltage constant.
NB: You can refer to Switching Power Supply Topology Review (a 1.09MB PDF file) and to Power Supply Topology Poster (a 143KB PDF file) for a brief introduction to the different types of switching power supplies.
However, there are not one but several voltages in a computer PSU. Which of them should be controlled? Suppose you launch a game. Your graphics card begins to work at its full capacity, the load on the +12V rail grows up, the voltage on the PSU’s +12V output sinks, and the PWM controller tries to lift it up to the previous level and at the same time increases the voltage on the +5V output.
Originally, computer PSUs used to have common voltage regulation to obtain several more or less stable output voltages from a single transformer.
To balance the different outputs a joint regulation choke L1 is introduced into the circuit: a single core with multiple windings, one winding for each output voltage. When the current in some winding increases, a negative voltage is applied to the other windings in order to make up for the above-mentioned increase in the output voltages on the corresponding power rails.
Thus, we have a PSU with multiple outputs which, notwithstanding only one regulating element (the PWM controller and the transistor Q1 it controls), keeps all of the output voltages at a more or less constant level. However, the voltages can deflect from their nominal values under greatly misbalanced loads.
In order to have more stable output voltages mainstream and top-end PSUs began to use additional regulators based on a magnetic amplifier circuit (also known as a saturated core design). Strictly speaking, such regulators had been long used on the +3.3V rail and then came to the +5V rail. As a result, the three main output voltages now have dedicated regulation.
In a magnetic amplifier circuit the joint regulation choke is replaced with two independent chokes L2 and L3 that have nothing to do with voltage regulation. Before one of them there is a special L1 choke whose behavior is controlled with the magamp control, which is an ordinary low-power linear voltage regulator. The choke’s job is to shorten the impulses from the transformer T1. The value of the shortening can be varied in real time.
Before and after L1 choke
The shorter the impulses, the lower the output voltage is. Thus, the secondary winding of the transformer T1 must have a reserve number of turns. The extra voltage can be removed by means of the magnetic amplifier choke L1.
As a result, we have two independent regulators: the main PWM controller is responsible for the +12V output and keeps its voltage stable, without bothering about the other outputs. And the additional magnetic amplifier regulates the +5V voltage. The circuit is not only simple but also very effective. The loss on the magnetic amplifier is close to zero.
NB: You can learn more about magnetic amplifiers from the article Magnetic Amplifier Control for Simple, Low-Cost, Secondary Regulation (a 1.5MB PDF file).
Although the magnetic amplifier proper is the choke L1, it is easier to identify such PSUs by the large and conspicuous L2 and L3 chokes. L1 is much smaller and is usually located near the power transformer.
Despite the ability of magnetic amplifiers to keep a PSU’s output voltages within ±3% of the nominal value at any load, they have a number of drawbacks. First, the additional chokes (L2 and L3) are rather large but cannot be got rid of. They play an important part in forward converter by accumulating power transferred through the transformer and yielding it into the load. Second, each output voltage requires a dedicated winding on the transformer T1, which makes the latter more difficult to design and manufacture, especially considering what wattages have to be fitted into the standard PSU housing today.
DC-DC converters are a replacement to magnetic amplifiers.
Here, a DC-DC converter is based on the transistors Q2 and Q3 and on the choke L2. In fact, it is a fully independent forward switching converter that has a dedicated PWM controller and can lower the +12V voltage to any desired level, be it +5V or +3.3V. As opposed to the PSU’s main converter, it does not have a transformer because it is already isolated from the high-voltage section.
This design has a number of highs. First, a DC-DC converter is powered by the direct +12V voltage and does not require an individual transformer winding. Thus, the design of the transformer T1 is greatly simplified – it only has one secondary winding. Second, a DC-DC converter can work at much higher frequencies than the PSU’s main converter and thus can use a smaller choke L2 and lower-capacity filtering capacitors at the output. This helps save some space inside the PSU case. Third, a DC-DC converter has a dedicated controller and, like with magnetic amplifiers, the PSU’s output voltages are regulated independently from each other and are stable as the result.
Why are DC-DC converters used only in top-end PSUs? The reason is simple. They are expensive consisting of a PWM controller chip and a few transistors. But semiconductor components are steadily getting cheaper and the above-mentioned advantages (the simplification of the power transformer T1 and the smaller size) help save a little, so now DC-DC converters are economically justifiable at least in premium-class PSUs. In a couple of years they are going to come to mainstream PSUs just as magnetic amplifiers did earlier.
Does the end-user benefit from DC-DC converters? No. It is rather hard to learn that a particular PSU has such converters unless you look inside it. You will need a good oscilloscope for that. These converters are interesting and expedient for developers and have only begun to be used because their pricing has become reasonable.
Is it a new invention? No. Every electronics engineer who has ever dealt with switching power sources can draw you a couple of basic circuits without thinking twice. Moreover, I have seen PSUs with such converters before, from Silverstone’s products to the 1500W models from Xigmatek and Thermaltake.
In the Antec Signature there are two cards with DC-DC converters between two heatsinks. One card yields +5V while the other, +3.3V. Both are powered by the main +12V source. The photograph shows the converters’ chokes clearly and you can see how small they are.
There are United Chemi-Con’s KZE and KZH series capacitors at the PSU’s output.
The quality of manufacture is very high: tidy soldering, secure fastening of every large component, and neatly laid cables. There is nothing I can find fault with.
The PSU is equipped with the following cables and connectors:
Included with the PSU are:
The selection of cables is just sufficient. If you don’t use adapters, you can connect a couple of graphics cards and about six hard disk drives (there are nine SATA power connectors in total, but one cable is going to be occupied by your optical drive and will not reach to the HDD cage in most system cases). Antec shows a good example with the peripheral power cables, by the way. There are four cables for the two connectors offered by the PSU. Two cables are PATA and two are SATA. Thus, the user can choose what is appropriate for his system configuration.
The Signature SG-850 is rated for a continuous output power up to 829W and can yield 780 watts across the +12V rail which is split into four “virtual” output lines. These are perfectly normal specifications.
Together with an APC SmartUPS SC 620 this power supply worked at loads up to 380W when powered from the mains but I had to reduce the load to 350W for the UPS to switch to the batteries normally. The PSU would occasionally produce gurgling sounds at that, so I can’t call this pair absolutely stable.
The cross-load diagram is typical for a PSU with dedicated voltage regulation: the +12V voltage is ideal at any load distribution. The +3.3V voltage deflects by less than 3%. The +5V voltage exceeds a 3% deflection only at extreme loads. I can remind you that the industry standard allows a 5% deflection, so the Signature shows an excellent result in this test.
It is all right on the +12V rail but the highest peaks of the pulsation on the +5V and +3.3V rails are higher than the allowable limit of 50 millivolts. The oscillogram does not show serious problems, though.
Take note of the difference in the pulsations on the low-voltage and +12V rails: it is because the latter is provided by the PSU’s main converter while the low-voltage rails have dedicated switching regulators working at a high frequency.
The PSU is cooled by a D08A-12PS3-06AH1 fan from Nidec Beta SL. Nidec’s website does not mention that model, though. Despite the high wattage of the PSU, the fan measures 80x80x25mm only. It uses a 4-pin connection with PWM-based speed control which should provide a wide range of speeds.
Indeed, the fan speed varies more than threefold depending on the PSU load. At loads up to 400W it is rotating at about 700rpm, making the PSU absolutely silent. Then the speed begins to grow up in a linear manner, but the fan only becomes noisy at loads above 650W. Thus, this PSU is one of the quietest models available, especially at low loads.
It also means that a large fan is not a necessity for a PSU to be quiet. A clever approach to cooling system design is more important than size here.
I should note, however, that the PSU may be hotter in a system case with a top position of the PSU and poor ventilation. It then produces an audible hiss. Therefore you may prefer to use a system case where the power supply is installed at the bottom. This piece of advice refers to any PSU, though.
The PSU has a good efficiency which is higher than 88% at loads of 300W through 600W and only lowers to 86% at full load. The power factor good, too. It is about 0.99 in about half the diagram.
The standby source of the Signature PSU is rated for a current up to 3A. Its voltage is only 0.1V below the rated value even at full load, which is perfectly acceptable.
The Antec Signature leaves a very nice impression with its excellent quality of manufacture, splendid electrical parameters, sufficient selection of cables and connectors, and with such a quiet operation at low loads that could hardly be expected from a PSU with only one 80mm fan. No wonder that Antec considers the Signature series the best of its product range even though the company offers PSUs with higher wattage ratings.
Perhaps the only downside of this PSU is its high price. But you won’t be disappointed if you pay it.
A representative of Enermax told about this PSU that they tried to put as many innovations into it as its housing allowed. This explains the name Revolution. The PSU is certified to comply with the 80 Plus Silver standard (an efficiency of 85% or higher at loads from 20% through 100%). Moreover, it may be over 90% efficient in a 220V power grid. The PSU can work under any load, including zero load, in order to avoid problems with systems that have advanced power-saving technologies for idle mode. Besides, Enermax talks about specific technologies employed in the PSU. I will discuss the most interesting of them below.
Judging by this brief introduction, the Revolution 85+ is a worthy opponent to the above-discussed Antec Signature.
The PSU comes in a large cardboard box on which the series name, wattage rating and key features are indicated.
Besides the PSU, the box contains an installation manual, a set of detachable cables and a pouch for storing them, a power cord and four screws.
Enermax’s designers have done a good job here. The PSU is painted a coarse (not just matte) gray paint and there is a separate red plate under the fan. This combination of colors and the very fact that the PSU is not just painted different colors but assembled out of differently colored parts leaves a nice impression. I mean the first impression you get, not the quality of metal, assembly, etc (which are high class, too).
There is a record number of connectors for detachable cables on the back panel – as many as ten! Six of them are for peripherals and four for additional power cables of graphics cards and CPUs. By the way, the latter have a large reserve of pins, so you can not only use one cable to power two end connectors but can enjoy the support for future graphics cards with new types of power plugs and high load currents. Of course, when such graphics cards do come out, you will have to purchase appropriate cables separately.
The connectors are similar to Molex Mini-Fit Jr. in form-factor but have side locks for fixing the cable (a standard Mini-Fit Jr. has a lock at the middle of the longer side, as you can make sure by looking at a mainboard power connector, for example). This ensures secure, but not quite handy fastening because it is hard to unfasten the locks between the rows of connectors when the cables are all plugged in.
The PSU case is 190 millimeters long.
The PSU follows a somewhat nonstandard variety of the dual-transformer design. It is a circuit with synchronized transformers. There is nothing new in the idea of using two transformers, though. With a high-wattage PSU, a single transformer may not fit within the required dimensions, so it is logical to split it in two transformers of half its capacity. The problem is to distribute the load in such a way as to avoid situations when one transformer is overloaded while the other is idle. This could be observed with Enermax Galaxy DXX power supplies. For them to be stable, the load had to be connected in such a way that each of the two transformer worked at a load of a few dozen watts.
So, this is where the transformer synchronization fits in.
The design shows the circuit in a simplified way, both in the high-voltage (in fact, two transistors control each transformer, allowing Enermax to talk about a “quad converter”) and low-voltage (the Revolution 85+ uses transistors instead of diodes as I will discuss shortly) sections, but that’s unimportant for understanding the point.
So, we’ve got one PWM controller that controls two forward converters Q1-T1 and Q2-T2 in such a way that the transistors Q1 and Q2 are opening up alternately.
Each transformer has a dedicated rectifier and a choke in which power is accumulated, but after the chokes the two circuits join into one circuit that has ordinary smoothing capacitors. Since the transistors Q1 and Q2 open and close alternately, the impulses arrive to the chokes L1 and L2 in antiphase.
As a result, the circuit’s operation recorded in an oscillogram looks like that:
The impulses that come alternately from both converters sum up but do not cross each other in time. As a result, the oscillogram at point 3 (i.e. at the PSU output) looks exactly as if we had one converter working at a double frequency. Thus, the problem of load balancing between the transformers is solved: the circuit is built in such a way that each of them delivers half the total output power and, from the load’s point of view, the PSU is no different from a single-transformer one. This solution also doubles the frequency at the output filter. When the frequency is higher, smaller capacitors and chokes can be used to additionally smooth out the voltage ripple.
Increasing the frequency of a single converter is not easy: it requires expensive high-frequency transistors and expensive materials for the transformer’s core. Here, Enermax’s engineers have solved a few problems with one solution: the transformers fit into the PSU housing, the load balancing between them is ideal, the job of the output filter is easier now.
NB: You can learn more about synchronous transformers in the article Interleaving power stages – not just for buck converters any more.
Is it a new technology? Yes, it is new in power supplies but it was developed not by Enermax and not today. This is indicated by the mentioned article which is dated the year 2004.
Two identical chokes can be seen at the PSU’s output: one choke for each transformer.
A UCC28220 chip located on a small daughter card is the controller of the synchronous converter.
This is just the beginning of the circuit design peculiarities of the Revolution 85+, though. Taking a look at the heatsink that usually carries diode packs of the output rectifier (marked as D1 and D2 in the schematic above), you can find no diode packs! Instead, there are IRFB3307 field-effect transistors:
The fact is the Revolution 85+ employs so-called synchronous rectifiers in which diodes are replaced with transistors. Why?
Let’s take a look at the specifications of a typical Shottky diode which is often used in power supplies: STMicro S60L40C (a 55KB PDF file). Look at Figure 1 on the second page that shows the correlation between the power dissipated on the diode and the current. At a direct current of 20A there will be more than 8 watts of power wasted – dissipated as heat – on the diode. This is due to the fact that there is a small voltage drop occurring when the current passes through the diode – about a few tenths of a volt. If you multiply that by tens of amperes, you get a few watts.
What does the diode do in the rectifier? It opens in one direction of current flow and closes in the other. It can be replaced with a transistor that is controlled in such a way as to imitate the diode’s operation. It can be the above-mentioned IRFB3307 (a 357KB PDF). In the open state the resistance of its channel is a mere 5 milliohms. So, at a current of 20A, it will dissipate P=I²R = 20²×0.005 = 2 watts. This is only one fourth of the dissipation of an ordinary diode! Of course, that’s an ideal example, but it provides a general notion of how much power can be saved.
It is no problem to make the transistors switch at necessary moments. In the simplest case their gates are connected right to the transformer’s windings.
If a higher efficiency of control over the transistors is necessary to minimize power loss, a synchronous rectifier controller is introduced into the circuit.
NB: You can learn more about the use of synchronous rectifiers in the article The Implication of Synchronous Rectifiers to the Design of Isolated Single-Ended Forward Converters (a 433KB PDF file).
Is the synchronous rectifier circuit new for computer PSUs? Yes. We have not tested such models before in our labs. Is it Enermax’s invention? I can quote myself: “I can predict, for example, that synchronous rectifiers will be sooner or later used in the secondary circuits of PC power supplies. There’s nothing new in that technology, but it is too expensive as yet and its advantages don’t cover its cost”. That was written in 2006.
Taking yet another look at the internals of the Enermax Revolution 85+, you can note a number of small components on the card with the output connectors.
These are DC-DC converters you have already seen in the Antec Signature section above. They are used to transform +12V into +5V and +3.3V. Enermax’s engineers have used these converters to the full, moving them from the PSU’s main card to near the connectors in order to make use of the space at the back panel.
Anpec APW7073 chips are used as the converters’ controllers. Next to them there are power transistors which are cool and do not require a heatsink. The function of a heatsink is performed by the copper foil of the card the transistors are soldered to.
On the reverse side of the card there are chokes (one for each converter) and smoothing capacitors. The connectors for detachable cables are nearby: it is to them that the voltages produced by the converters are applied, besides everything else.
There are other tricks you can find in the Enermax Revolution 85+. For example, the bridge in the photo connects two points of the same interconnect and reduces its overall resistance and power loss. These things are not fundamental and are far less interesting, though.
The PSU is equipped with the following cables and connectors:
Included with the PSU are:
So, even though quite a lot of the PSU’s connectors are unused, you have as many as six 6+2-pin graphics card connectors, six power connectors for PATA drives, and 12 power connectors for SATA drives.
The PSU has a max output power of 850 watts and the manufacturer claims it can work at such load continuously at an air temperature of 50°C. The PSU can yield up to 840 watts across its +12V power rail which is split into six lines. These lines are “virtual” because, notwithstanding the two transformers, the Revolution 85+ is no different from ordinary single-transformer PSUs for the load connected to it and imposes no specific restrictions.
Together with an APC BackUPS SC 620 this PSU worked at loads up to 385W (from the mains) and 350W (from the batteries). The pair switched to the batteries normally. The UPS was absolutely stable, producing just a soft buzz.
The dedicated voltage regulation brings the predictable effect: the +12V and +3.3V voltages stay within a 3% deflection. The +5V deflects more than 3% but only at extreme loads.
You can note that the diagram begins at zero on both axes: the PSU is indeed stable at zero load.
This diagram resembles the one of the Antec Signature: there is considerable high-frequency pulsation with occasional tall spikes. There are no serious problems here, though.
The PSU uses a 135x135x25mm Globe Fan RL45 fan. It is an ordinary 3-pin fan, unlike the 4-pin fans of Enermax’s MODU82+ and PRO82+. The PSU has a tachometer output you can connect to the mainboard and control the fan speed from BIOS or with Windows-based tools.
The fan speed is constant at about 700rpm until a load of 550W when it begins to grow up linearly. However, even at the highest load the speed is only 1120rpm, making the Revolution 85+ a very quiet power supply.
Besides, the manufacturer claims the design of the PSU case with the rolled-in edging of the fan hole reduces the noise by 1-2dB more. I could not check this out as I didn’t have a sufficiently accurate noise-measuring tool.
When the PSU is turned off, the fan keeps on running at a low speed for 45 seconds.
The Enermax Revolution 85+ sets a new record as it is the first power supply tested in our labs to exceed 90% efficiency! I guess this is largely due to the above-described synchronous rectifier. The Revolution 85+ is the first PSU with a synchronous rectifier in the main transformer that I have ever tested.
As opposed to most competitor products, the standby source of the Revolution 85+ has a load capacity of 5A. It does its job well: the output voltage is 4.87V at full load, the allowable minimum being 4.75V.
I won’t say if it is revolution or evolution, but Enermax’s engineers have indeed come up with something new and original from a technical standpoint. This PSU is original not only with its exterior but also with its interior design. The capabilities of the Revolution 85+ may seem even redundant for today – the number of unused connectors for detachable cables is a clear indication of that. However, people at Enermax stress the point that they have tried to create a platform not only for today but also for near future, stuffing it with everything best that modern electronics can offer. And they seem to have succeeded.
Besides the moral satisfaction of owning one of the most technically sophisticated power supplies, the Enermax Revolution 85+ offers you good quality of manufacture, excellent electric parameters, a rich selection of cables and connectors, and quiet operation throughout the entire load range, up to the maximum 850W.
As you may have guessed, the biggest downside of this product is its price. The recommended price is $309 or €229 taxes excluded.
Although Seasonic does not make such ambitious claims as Enermax, its M12D power supply can challenge the Revolution 85+ in a number of parameters. Particularly, it complies with the 80 Plus Silver standard and has an efficiency of 90%. By the way, Seasonic has recently introduced a series of PSUs with lower wattage that are even 80 Plus Gold compliant.
The M12D also features DC-DC converters that I described above in the Antec Signature section.
This PSU comes in a small box painted orange and black. Its key features are listed on the back panel of the box. Included into it are detachable cables, a pouch to store the cables, an installation guide, a power cord, screws, and a sticker for the system case.
The PSU has a standard Seasonic case painted a matte black against which the silver-and-blue sticker on the fan grid stands out.
There are six Molex Mini-Fit Jr. connectors on the back panel of the case for detachable cables: four for peripherals and two for graphics cards. It is inconvenient that the connectors are all the same color. However, you cannot plug anything wrong because the connectors have different number of pins.
This PSU has an ordinary appearance, differing from Seasonic’ previous products with the shape of the heatsinks only. Two of them have acquired very wide petals while the third, with the power elements of the active PFC device, has become smaller and is inconspicuous among other components.
The PSU has one power transformer. Its active PFC and main regulator are based on the Champion Micro CM6802 controller (a 338KB PDF).
The most interesting thing is hidden beneath the cables: it is a narrow aluminum bar standing near the big heatsink. This bar is a dedicated heatsink of a DC-DC converter that delivers +3.3V and +5V voltages.
Seasonic was kind to provide this converter to me and I did not have to dismantle the PSU to get to it.
There are two chokes on the face side of it – one for each output voltage – and filtering capacitors. Take note of the size: no wonder that the manufacturers prefer to use such compact cards instead of additional chokes of magnetic amplifiers.
The back side of the module is covered with a heatsink. Without it the converter, squeezed between the cables and the large heatsink of the PSU, would overheat. It is easy to take the heatsink off, though.
You can see two Anpec APW7073 PWM-controllers and seven transistors. This is the control and power sections of the DC-DC converter. It is fully autonomous. You can take this card apart from the PSU and connect +12V to it – and it will yield +5V and +3.3V at the output.
KZE series capacitors from United Chemi-Con are used at the PSU’s output (in the 12V rail, to be exact).
The PSU has the following cables and connectors:
Included with the PSU are:
The list is long and the M12D may seem to offer the richest selection of connectors among the PSUs included into this review, but this is not so. It is only the handiest. Instead of a heap of identical cables the manufacturer has included a set in which each cable has a specific length, so you can choose what suits you and your system case best.
The number of connectors is more than sufficient, though. There are six graphics card connectors, 12 SATA and eight PATA power connectors. It is hard to imagine a system that would require more.
The PSU is rated for a continuous output power up to 850W. It can deliver 840W across the +12V rail split into two “virtual” output lines.
Together with an APC BackUPS SC 620 this PSU worked at loads up to 360W (from both the mains and the batteries). The pair switched to the batteries normally. The UPS was stable.
Despite the dedicated voltage regulation, the M12D is the only PSU in this review to be not ideal in this test. Its +5V voltage varied greatly depending on load and eventually exceeded the allowable limits.
On the other hand, everything is going to be normal in a real computer. This problem is only observed when there is a high load on the +5V rail, which cannot occur in today’s systems whose components are mostly powered by the +12V rail.
The PSU is excellent in this test. The voltage ripple is very low on every rail, excepting the occasional spikes on the +12V one.
The PSU is equipped with a 120x120x25mm Sanyo Denki San Ace 120 fan. Besides the somewhat unusual shape of the impeller, you can see one more interesting thing here:
I don’t mean the open bearing – it is still impossible to lubricate it. And that’s useless because ball bearings are lubricated on the inside and the lubricant does not trickle out. I mean the grooves two of which are partially filled with brown lacquer. Can it be that the fan manufacturer balances each sample of the fan by adding a drop of lacquer on what side of the impeller proves to be lighter due to inaccuracies of manufacturing process? At least, it looks like that.
The fan speed is constant at about 800rpm at loads up to 500W and then begins to grow up. The PSU is very quiet at low and medium loads but the fan becomes audible at 650W and downright loud at near-maximum loads. On the other hand, it is difficult to assemble a computer whose components would be quiet at such a load, so the M12D can be considered a quiet power supply.
Well, this PSU stops just short of notching a record. It has an efficiency of 90% (at a load of 400W), which is 1% lower than the efficiency of the Enermax Revolution 85+. Anyway, this is the second best result ever shown by PSUs in our labs.
The standby source is rated for a current up to 3A. It does its job well. The voltage is always within the norm.
The Seasonic product perhaps does not look as impressive as the Revolution 85+ but in practical terms it is not inferior to its opponents from Enermax and Antec. Yes, one of them is technically perfect, but what’s the purpose of that perfection if the output parameters are almost the same?
When it comes to the effective parameters, the M12D can only be criticized that its output voltages are not very stable. There is no other fault I can find with it. It features low voltage ripple, a very high efficiency, a gorgeous selection of cables, and quiet operation at low and medium loads. Thus, it will make a perfect power supply for a home computer.
The recommended price of this PSU is $299.
Power supplies that are touted as the best-of-the-range models by renowned manufacturers cannot have bad results in tests. Indeed, the Antec Signature, Enermax Revolution 85+ and Seasonic M12D are free from any technical defects. These are high-wattage, well-made products with good electric parameters and quiet operation, suitable for top-end computers including configurations with two or three graphics cards. You won’t be disappointed with any of them. The only downside is that they come at an expectedly high price.
The Enermax Revolution 85+ stands out in terms of technologies. This is the first PSU tested in our labs that delivers an efficiency higher than 90%. It features an ideally balanced dual-transformer circuit capable of working at any load from zero watts, a synchronous rectifier on the +12V rail (it is the first time I see such a solution!), and dedicated DC-DC converters. Enermax’s engineers have done a good job on this product. If you are interested in power electronics and want to know how computer PSUs will develop in near future, the Revolution 85+ is a good example to study.
The two other models, Antec Signature and Seasonic M12D, are designed in a more traditional way. Instead of revolutionary innovations their developers have preferred to elaborate on well-known and long-used technologies (I have seen DC-DC converters in serial products over two years ago, for example). They cannot match the Enermax in their effective parameters but the difference is small. They are 1-3% less efficient, produce somewhat more noise under load, and do not differ at all from other aspects.
By talking so much about the circuit design of the PSUs, I wanted to emphasize two points. First, computer PSUs do not stay as they are. They evolve and improve and this development is not limited to the shape of the holes of the vent grid or the color of the fan highlighting. There emerge new controllers, some design solutions are replaced with others. There is little in common between two PSUs manufactured ten years apart, although they seem to have components of similar colors and shapes. That said, you should be reasonably skeptical about the manufacturers’ announcements of newest, recently invented and wholly patented technologies. New components only come to serially produced PSUs when they are commercially justifiable. You can take magnetic amplifiers for example. They are long used as regulators of the +3.3V rail and you can find such a regulator in every good 250W ATX power supply manufactured at the end of the last century but it is only in recent years, when the load capacity of the +12V rail has grown up greatly, that the use of two magnetic amplifiers – what is called dedicated voltage regulation – began to make sense. It’s the same with other technologies. They exist but may not be commercially reasonable until some point.
What can we expect in the near future? I guess there will be digital programmable PWM controllers whose algorithm will be adapting “on the fly” to various types of load. They exist already but are far from becoming widespread because the technology is not yet polished off and is expensive. Of course, this is just an example. There are other interesting technologies waiting to be implemented in serial products.