How a PSU works

[Tom41]

Just thought I'd post this, from what I remember when I followed the Open College electronics course. This is almost like a FAQ page, explaining how exactly a PSU works - for those new to the subject, and those wanting to learn about electronics.

As you know, the electricity coming in from the mains is high-voltage AC (alternating current), moving first one way and then the other. Computer motherboards, CPUs and other peripherals only work with low-voltage DC (direct current), moving one way all the time. The role of the PSU is to convert the 240v 50hz (or 120v 60hz) AC mains into a stable DC supply.

The AC first goes into a transformer, which transforms the 240v/120v AC into a lower voltage AC supply, such as 12v. Transformers rely on moving magnetic fields to do their job, so they will ONLY work with an AC supply.

Now we have low-voltage AC, we need to make it into DC. The simplest way of converting AC to DC is to use a diode, effectively a one-way gate for electricity. The electricity can flow through the diode one way, but is blocked when it tries to move the other way. Using a single diode in this way, you get the electricity only going in one direction - but you are only using half of the AC waveform, so it's called half-wave rectification.

You need to add a second diode pointing the other way, but connected to the same DC line, to use the other half of the AC waveform. Doing this, you are using the full power provided by the AC waveform, so it's called full-wave rectification.
In practice, 4 diodes are usually used - two to allow the electricity in and out for the positive half of the AC cycle, and two for the negative half of the AC cycle.

We now have DC, but if you look at it on an oscilloscope you'll see that it's a series of positive voltage pulses. Computer components need a smooth DC supply (straight line on the scope), not a series of pulses.

This is where capacitors come in. A capacitor in a PSU will charge and discharge rapidly to smooth out the DC current. In effect, it's storing up charge during the voltage peaks, and putting it back into the circuit during the troughs. The result is a smooth, DC supply that can be used by most electrical appliances.

However, the output of the capacitor segment is still not smooth enough to be used by certain sensitive devices, such as computer CPUs. If you look at the trace on an oscilloscope, you will see that the line still has 'wrinkles' in it, since the capacitors can't fully smooth out the supply.

The next stage is to use Zener diodes. These are just the same as normal diodes in that they allow current to flow in one direction, but not another. However, that's only true up to a certain critical value. Once the voltage reaches that value, the Zener will start allowing a small current to flow in the 'wrong' direction. The bigger the reverse push, the more current the Zener passes through.

The practical upshot of this is that if you place a Zener across the positive and negative lines coming out of the capacitor segment, in reverse polarity, the voltage across the Zener will remain constant - no matter how the power supply output varies. Therefore, you take the power output not from the capacitor segment, but from across the Zener diode itself. This should, in theory, give you stabilized, constant, DC power for sensitive components.

Zener diodes are available in a wide range of voltage ratings, and are commonly found in PSUs in the form of a voltage regulator. This is a little black box containing a Zener diode, and also some internal circuitry that protects it against short circuits and overloads.

There is one disadvantage of this though - the output voltage is only stabilized as long as the voltage rating of the Zener is greater than the output of the power supply itself. If the load gets too great, and the power supply output dips BELOW the voltage rating of the Zener, then the Zener will go back to being a normal diode (stopping all current), and the output voltage is no longer stabilized.

Experience has shown that the PSU output voltage is rarely a clear, stable trace on the oscilloscope - even the best PSUs have what's called 'ripple current'. Also, if the PSU capacitors fail, the Zener will no longer be able to stabilize the fluctuating voltage, and you'll get 'bad cap' symptoms even if your board capacitors are fine.

That's in a nutshell how a normal PSU works. If you know where they are, you can measure the outputs of the various stages of the PSU on your own oscilloscope.

[PeteS in CA]

This is a good intro to power. It describes a basic shunt regulator, which is good for low currents, but above a couple hundred milliamps, it gets impractical due to inefficiency. The series resistor and the zener have to be very high power devices.

The next step is a simple series regulator, in which the series resistor and zener are in the base circuit of a bipolar transistor or darlington. This keeps the zener power low, lets a small current control a much larger output current and simplifies thermal management. Depending on the input voltage range, it can still be inefficient. The next step from there is to add an error amplifier sensing the output voltage to improve regulation and permit functions such as current limit, OVP, etc.. Efficiency still isn't helped by this.

[willawake]

thanks guys you are most welcome to write FAQs. you can see by the viewcount on some of the FAQs that many people read them and appreciate your advice.

[999999999]

Tom41, while you are describing "A" power supply, it is not the typical switching type used in computers. My comments below pertain to how a typical computer switching PSU deviates.

Originally Posted by Tom41

The AC first goes

Through a rectification stage into bulk capacitors. It reaches a peak DC voltage around 335V (after a basic voltage doubler circuit on 110VAC supplies). This assumes a passive PFC PSU, otherwise we have a boost circuit that maintains cap charge above peak line level, that tracks voltage to cause a near sinusoidal, corresponding current to charge the caps. Then through switching transistors at around 50KHz, maybe up to 100KHz is getting more common that it once was, AC and we pick up where you left off "into a transformer".

The next stage is to use Zener diodes. These are just the same as normal diodes in that they allow current to flow in one direction, but not another. However, that's only true up to a certain critical value. Once the voltage reaches that value, the Zener will start allowing a small current to flow in the 'wrong' direction. The bigger the reverse push, the more current the Zener passes through.

The practical upshot of this is that if you place a Zener across the positive and negative lines coming out of the capacitor segment, in reverse polarity, the voltage across the Zener will remain constant - no matter how the power supply output varies. Therefore, you take the power output not from the capacitor segment, but from across the Zener diode itself. This should, in theory, give you stabilized, constant, DC power for sensitive components.

I'm not quite sure what you're suggesting here, that PSU use zener shunts to regulate? They don't. After the LC filter(s) following the transformer the feedback through an error amp is put through the controller chip's comparitor subcircuit (or indirectly through another comparitor) to flip flop the switching transistors just before the transformer. There are zeners involved in the shutdown protection circuitry but not to directly regulate (in most cases, though it would be possible to make some kind of zener/pass-transistor type of arrangement using the zener as a reference but efficiency would go down the tubes). The controller tends to have it's own reference voltage, that is not dependant on zeners across supply lines either.

[bgavin]

I certainly appreciate this thread. Keep it up.

Anybody have a tutorial or other visual aid that home gamers can use to follow the dialog?

[starfury1]

Just to try and put a bit of form around the content so far.

Tom41 has described the operation and basic components used for a some what crude very low amperage
LINEAR Power Supply.

(I think he may have missed the series resistor for the zener, he did say from what I remember )

Please bear in mind I'll bend the truth to convey a concept

For us to be able to get regulated DC voltage from our mains systems

We first have to convert the mains Voltage to a more usable level

This is achieved By the transformer (page 2 of link below)

The Next Step is to convert this smaller AC voltage to DC

This is a two step process

Step one is to convert the small AC voltage to a pulsating DC voltage

This is known as Rectification and achieved by use of a component know as a Diode

(This can vary depending on if we want utilize the full cycle of the AC voltage)

The pulsating DC is still fairly useless to us so we now need to smooth it.
(EG an example of pulsating , think of turning your volume control up and down in a smooth way...not much good to listen to

Step 2
is Done By filtering or storage and the Component used for the job
is a Capacitor (generally)

Which will now gives us DC voltage
(think of the capacitor, in this case as a bucket and the pulsating DC as small cups filling the bucket)

At this point we now have what is know as an Unregulated DC Power Supply

basic power supplies circuits ( Linear)

The above link has the circuit diagram (fig 5.6 page 3) tom41 has described the operation of


If I now attach a load say a light bulb to the capacitor and and start drawing current from it
depending on how much, this will cause the voltage to drop on the capacitor.
(think of the load as a hole in the bucket)

With most circuits you want the voltage to stay close to what it was designed to operate at.....say 5 volts

We then Need to keep the voltage at that level regardless of the current being pulled.

So next We Need a Voltage Regulator circuit of some kind

In Tom41 Example it is made from a Resistor and Zener diode
in series across the Capacitor.

The output is taken across the Zener Diode

to put it simply

The Zener Diode will try to maintain the zener voltage at a constant across it and will adjust its internal resistance to compensate for changing values of current drawn by the load.
(the resistor would have flowing through it the sum of the zener current and the Load current)

here is Zener Regulator link


PeteS expended on the idea by adding requirements for more current and better regulation, although he didn't mention it in a word.
It requires one key concept not really there in tom41 example and that is FEEDBACK use to control the circuit
(although you could think of the zener as possibly having internal feedback loop.)

999999999 has drawn comparisons to what has now become the almost universal type of power supplies used these days for most everything and how it differers.

The switch mode power Supply

These while having more complexity and component count have some great advantages.

some of which are better efficiency no need for bulky expensive mains transformers.

Here are a few link for those that wish look into supplies further.

The link below is a bit hard to follow as you need to download the pdf's for the diagrams so you got them to look at...beter I would have thought to have them inlie
at the bottom is a link to a page about an SMPSU...SCS

Practical power supply theroy

This PDF may be off use although not directly about psu design HERE

On Switch mode psu here

see if I can get some better links on switcher

HTH Cheers

[tazwegion]

Originally Posted by starfury1

This is known as Rectification and achieved by use of a component know as a Diode

(This can vary depending on if we want utilize the full cycle of the AC voltage)

I'm not really conversive in all this technical terminology... but don't PSU circuits use 2 groups of 2 zener diodes (opposite polarity) to capture both ends/extremes of the AC wave?

[PeteS in CA]

The first post in this thread is a description of a switching P/S.

[starfury1]

just to somwhat correct my post above

I did some how miss the Regulator bit at the bottom.
As in "Black box" and I would think he was referring to a 3 Terminal regulator
Here

Zener diodes are available in a wide range of voltage ratings, and are
commonly found in PSUs in the form of a voltage regulator. This is a little black
box containing a Zener diode, and also some internal circuitry that protects it
against short circuits and overloads.

Tom41 then goes on to state that once the unregulated DC voltage drops below the "Zenering" voltage it drops out of Regulation.

Originally Posted by starfury1
This is known as Rectification and achieved by use of a component known as a Diode
(This can vary depending on if we want utilize the full cycle of the AC voltage)


I'm not really conversive in all this technical terminology... but don't PSU circuits use 2 groups of 2 zener diodes (opposite polarity) to capture both ends/extremes of the AC wave?

What your are referring to is a circuit of 4 diodes know as a bridge rectifier
these are power diodes not zener diodes
They can be build from 4 diodes or come as package.

bit further reading if anyone wants to or needs to

Ok with regard to the Rectification section.....
they are normal power type diodes (in the case of linear supply)

which act as a one way valve.

with the right polarity voltage on the Anode and cathode (A K)
current will be allowed to flow.
(open valve (or almost short circuit)
when it is reversed it acts as a closed valve ( or almost open circuit)
(there is in both cases a finite resistance in the real world) Ideally none.

Diodes will have a point when they have a reversed voltage
on them that they will breakdown
and start conduction of current.

normally referred to as "Peak Inverse Voltage" (PIV)

to put it simply
Zener diodes are specially design to exploit aspects of this
reverse break down region.

So as above its being used as either the regulator or part of the regulator


See This page for a bit more detail


Having said that the rectification process can use 1 2 or 4 power diodes
basically really depends on what your objectives are

Halve wave rectification (1 diode)
full wave rectification (2 or 4 depending on transformer)

if you have a center tap transformer or not (2 diodes)

the above with regard to linear supplies and there are as normal
advantages and disadvantages depending on how you want to go

PIV More so with a Switchmode PSU there is no stepdown transformer
and the rectification process is done directly on the mains voltage!
normally an RFI filter and some protection components before it

Here is were you really want to know what the PIV is

bit more wiki on diodes

bit more but heavier going here

Thanks PeteS for including that link its a GREAT intro for SMPSU
(I had forgotten it )

hope it helps and doesn't confuse you all...

cheers .....any errors or corrections in this or the links please post

[starfury1]

I should point out that by the PIV voltage with regards to power type diodes bridge rectifiers you can be talking 50v~1000v and higher depending on the diode type or bridge package.

peak inverse voltage

(electronics) Maximum instantaneous anode-to-cathode voltage in the reverse direction which is actually applied to the diode in an operating circuit.

in this Document its referred to as "peak repetitive reverse voltage Vrrm"

data sheet PDF on 1N4001~7 HERE

Another link on diodes HERE

And lastly I like the way the bridge rectifier is represented in this circuit Here

if you put your thumbs diagonally on each of 2 diodes (left lower right upper)
then move up and down to the opposite ones above or below (two thumbs at the same time)

you will get the effect at each half cycle,
only two are conducting
one is suppling the other as returning
(the load exits between them)
to complete the closed loop...circle if you like

When the AC voltage changes polarity those two act as open circuits and the other two start conducting then it repeats

transformer V+ Diode ----load----diode V- Transformer

"Load" in this case is all of the circuit between the 2 conducting diodes

Hope that makes some sense

The circuit in its self is interesting and I guess there must be some draw backs to it, other wise we wouldn't need large filter caps

here is the main power supply index of the site HERE

(I wouldn't build number No. 13 thought, you are playing with mains voltages)

There are a few regulator "black box" circuits and has a few variations on how you can use them to get more current etc

Mainly(if not all) linear type supplies from what I can see.

Here the main page of the website HERE

covers other stuff

the site is a bit light on good info thought

HTH cheers

[starfury1]

wiki on DIODES

BTW I use the term PIV little loosely so just be aware of terms....

here a little active type page for the bridge rectifier that you click on and it show 1 or the other of the 2 pair of diodes working.

bridge rectifier operation

Just click left or right of the line in the Vin circle
The transformer and secondary winding is not shown but it would be connected to the 2 tiny circles with the plus or minus (changing) so just imagine it there with your mind

Had this been done with the diagram in the above post it may have been a little clearer.

BTW diode bridges were sometimes used by people to idiot proof mobile car equipment (hard wired in place)
because of this "Steering" ability.
in other words it didn't matter which wire on the input (AC) went were cause the output positive would always be positive....ditto the negative

Bridge rectifies will have 2 terminals marked with AC a + and -
(but not in all cases)

These days mostly IC regulators are used so finding info on discreet build ones with transistors and op amps is getting a little harder.

Anyway this page has a bit of what I was looking for
whats inside a 3 terminal regulator plus some other circuits
HERE

here's a PDF on Regulator Fundamentals from national HERE

Bit techie thought


HTH Cheers

[PeteS in CA]

Common usage is to refer to signal types and zeners as diodes and higher current diodes as rectifiers. The 78XX/79XX and LM340/LM320 series 3-Terminal regulators are 30+ year old designs - Fairchild and National did a good job. The devices offer good, stable regulation, current limit and thermal protection. Not bad for a TO-220 sized device (plus, maybe, a small heatsink).