Every day I learn more about power supplies and I finally finished studying some points that were still a little bit obscure to me. Now everything is clear and I revised ALL power supply reviews I wrote. See below a complete list of all corrections I made and why.
- 1. Math for calculating the maximum theoretical current supported by the rectifiers
Computer power supplies are based on “step-down buck topology”. On this topology the minimum rating for the rectifying diode can be calculated thru the formula I = Io x (1 – D), where Io is the output current and D is the duty cycle from the waveform applied on this diode. Thus if we know the maximum current supported by the diode, we can find the maximum output current using Io = I / (1 – D).
It is important to notice that power supplies usually use two diodes inside a single component, and this component is usually rated with the maximum current it supports, which is the addition of the maximum capacity of its two internal diodes. Thus we have to consider the current for only one of the diodes (i.e. half the maximum capacity of the rectifier). Of course if the power supply has rectifiers installed in parallel we will have to multiply the current by the number of diodes installed in parallel.
For the correct calculation we need to know the duty cycle. Our reader Travis Chen pointed out that we could consider use a typical duty cycle of 30% for power supplies based on two-transistor forward configuration (the most common design for PC power supplies with active PFC) and 50% for power supplies based on half-bridge configuration (the most common design for PC power supplies without active PFC).
On the first case since we will be using a typical value – and not the real value that the reviewed power supply was using when delivering its labeled power – we may end up calculating the wrong value, but since this is just an exercise to help our readers to have an idea of the potential of the rectifiers being used.
Just a hypothetical example: a power supply with two-transistor forward configuration using a 30 A rectifier for its +12 V output. What is the maximum theoretical current and power the +12 V output from this power supply can deliver? Answer: 21 A [15 A / (1 – 0.30)] or 257 W (12 V x 21 A).
For power supplies without active PFC based on the half-bridge configuration, this math is easier. Just use the maximum total current each rectifier can deliver. Consider the above example but with a power supply with half-bridge configuration. The maximum current will be 30 A [15 A / (1 – 0.50)], so actually we don’t need to make any math, since the maximum current will be always double the maximum current from the rectifying diode, just use the total current supported by the rectifier (30 A on this example).
We were making the mistake of using this same logic for power supplies with active PFC (forward design), but as we demonstrated the values are completely different.
So I went back to all reviews I published and corrected all the maximum theoretical numbers for the rectifiers.
- 2. Removed acid comments regarding power supplies based on regular NPN transistors
All reviews from power supplies without active PFC got very acid comments here, where I said things like “old and ridiculous design” and worse. The matter of fact is that power supplies without active PFC typically use the half-bridge design, which is the same design used by older power supplies. It took me a couple of years to understand that there is nothing “wrong” with that and if a user wants a power supply without active PFC chances are that he (or she) will get a power supply with half-bridge design. And we have to live with that.
- 3. Removed criticism regarding using regular NPN transistors instead of MOSFETs
Power supplies with half-bridge design typically use traditional bipolar (BJT) transistors instead of MOSFETs and I heavily criticized this, because in theory MOSFETs can deliver a higher efficiency – but the story is not quite that. BJT transistors generate less noise and ripple than MOSFETs, so there is an advantage of using BJTs instead of MOSFETs. And they also present lower “switch dissipation” compared to MOSFETs. Switch dissipation, also called “crossover loss”, is the amount of power the transistor wastes when switching from on to off and vice-versa. So, the truth is that when comparing a BJT to a MOSFET working at the same frequency in theory the BJT will offer a lower loss, which translates in a higher efficiency.
On the other hand, BJT transistors are slower than MOSFET, i.e. with MOSFETs the switching frequency can be higher. At higher frequency the time spent between transitions (i.e. between the transistor switching from off to on) is also lower, which then reduces the switching loss (because less time is spent during the transitions). So if you have a MOSFET switching at a higher frequency than a BJT then the MOSFET loss will be lower than BJT, thus leading to a higher efficiency.
In summary MOSFET transistors will provide a higher efficiency only if they are being switched at a higher frequency (which is usually the case). Because of this “if” and because almost always power supplies based on half-bridge configuration use BJT transistors (i.e. usually there is no other option – the other option would be going to active PFC with a forward configuration and then the reviewed power supply would have to be a completely different animal) I decided to remove this criticism.
- 4. Removed schematics that followed the two criticisms above
To emphasize that the power supply used an old design, I would add a schematics of a very old AT power supply and say that the reviewed power supply used the same project. Since I decided to re-evaluate the points already exposed above, I also decided to remove this schematics and any comments about it. Of course a power supply with passive PFC will have a project resembling old power supplies: they are all based on the same topology (half-bridge design).
- 5. Corrected analysis of power supplies based on synchronous design
On synchronous design the diodes from the secondary are replaced by MOSFET transistors. The idea is to increase efficiency, as MOSFETs provide a lower voltage drop compared to diodes (translation: less power is wasted to operate them). At least in theory, because we’ve seen some power supplies with this project reaching relatively low efficiency.
I failed to detect and analyze this kind of project correctly before, simply writing “unusual design” or “since we couldn’t understand the design used”. This is now completely fixed.
- 6. Function of the thermal sensor
In some older reviews I wrote that the thermal sensor located on the secondary heatsink is in charge of controlling the speed of the fan and also to shut down the power supply in case of an overheating situation. This second part only happens if the power supply implements over temperature protection (OTP) and almost all power supplies don’t have this protection. When they do have, usually they use a separated thermal sensor for this feature, and not the same one used for the fan control. Fixed.
- 7. Which components play a role on limiting the maximum current
All components play a role in the maximum current (and thus power) the power supply can deliver, but capacitors. I posted several phrases in the past like "especially the transformer, the coil, the capacitor and the wire gauge used, as mentioned before". Edited to remove references to the capacitor.
- 8. Added comments on the rectifying bridge
Since I was revising all reviews I decided to post a small analysis regarding the rectifying bridge used on the primary. I added this to all reviews. Now the only main semiconductors that are missing a better analysis are the switching transistors. I plan to add that in the future.
I edited older reviews to make them follow the same format we are using on our latest reviews, like separating the component analysis into Transient Filtering Analysis, Primary Analysis and Secondary Analysis; add/fix the maximum temperature rating for the semiconductors; etc.
Summary
In summary, I had a lot of work in the past two days but I believe it was worthwhile. I decided to write this somewhat long text to give a detailed explanation on why I edited all the power supply reviews so if you follow them you will also “fix” some ideas I was getting wrong. As you know, learning is an on-going process and we all learn new things and change our point of view every day.
If you find any error on any of our reviews or articles, don’t hesitate in telling us what is wrong so we can fix.
BTW, I will be updating our Anatomy of Switching Power Supplies very soon with lots of more detailed information.
