Articles: Cases/PSU

Bookmark and Share

Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 ]

Circuit Design

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.

Group stabilization

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.

Magnetic amplifier

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.

DC-DC converter

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.

Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 ]


Comments currently: 13
Discussion started: 12/15/15 07:35:22 PM
Latest comment: 07/20/16 12:09:43 PM

View comments

Add your Comment