shinji2k

This is getting pretty off topic but this is the first I've heard of this. Most PSUs I know of have a separate IC in charge of OCP, usually a PS223, that is completely independent from the PWM circuit. This OCP IC is found on the secondary after the voltage has been regulated, rectified and then filtered. This feedback you are talking about if the rails are recombined after the OCP, does not exist AFAIK. I could be wrong since you seem to know more about electronics than me, but this is the first I've heard of this. I can find at least one PSU reviewer using a breakout board to combine the rails into one, with no ill effects, since his Sunmoon load tester cannot handle more than 2 rails.

edit: And yes, you can think of the rails not existing when we are discussing the maximum 12V amperage of a PSU. There is only one set of rectifiers in a PSU for the 12V which corresponds to one 12V source with one maximum load. The rails do not matter when you are discussing the max current on the 12V.

__________________
Quis custodiet ipsos custodes?

waqasr

Back on Topic..To answer your question..No your psu is not limiting your overclock..Silverstone make very high quality psus and it should last you years..so you need not worry about your psu.

__________________
SEX
Now that ive caught your attention please scroll up and ACTUALLY read my post
939 RIG : DFI Ultra D, Opty 165@2.7ghz, OCZ BH-5 Ram 1.5-2-2-0, BIG Typhoon.
17718 3DMARK 06 score with sig righttp://service.futuremark.com/compare?3dm06=6793718
My Xbox Gamertag is LiL WaKy, i challenge you to Halo3 and Forza 2!

MagicBox

@sjinji2k

The rectifiers are usually found in the primary section, which is the high voltage/low amperage DC section. It feeds *all* the voltages coming out of the PSU. That explains why. Because it is high voltage, buffer caps can be relatively small; likewise for the isolation transformer. After this stage you'll find the switching mosfets. They transform the high voltage into the desired output voltages. A simple example: Let's say after rectifying the DC output of the primary is 24v (which it isn't but it's to simplify). With a duty cycle of 50% (50% of the time the mosfet conducts, 50% of the time it's off) the resulting average is 12v. Simply by varying the dutycycle the output voltage increases or decreases. The PWM controller measures the output voltage and current and adjusts the dutycycle accordingly. That's what feedback means. The voltage measure point and current measure points are feedback lines to the PWM controller so it knows how to adjust the dutycycle.

Basically this control loop is what defines a rail. If there's 1 control loop with 1 output voltage measure point and 4 current measure points, then you speak of virtual rails and then what you say is right. There are some PSUs with virtual rails. There is only one set of switching mosfets. If one rail gets overloaded, all rails go down. If each rail has its own control loop (thus own PWM controller, 1 voltage measure point and 1 current measure point, and own set of mosfets) then you speak of independant hardware rails. These PSUs are more expensive, offer better quality and have better protection. All rails are fed by the same primary obviously.

Silverstone PSUs have independant rails and no virtual rails. The reason why you generally shouldn't bother much about rails is because the rails are assigned to the modular connectors in such way that the rails are allocated across various hardware already.

Just some indepth info on PSUs

shinji2k

I am well aware of the design basics of a SMPS and how they are set up. There is also a set of rectifiers on the secondary after the step down transformer which is what I was discussing. On almost all PSUs there is only one transformer providing 12V (or two in parallel on a few high watt designs), meaning there is only one true rail. Pretty much the last step on any multiple rail PSU is where the OCP is added. The OCP does not have any effect on regulation since it is a completely separate circuit. It just seems like you are ignoring what I am saying and lecturing on PSU design which has nothing to do with this thread. We are incredibly off topic here and our discussion has not been helpful to the OP.

__________________
Quis custodiet ipsos custodes?

MagicBox

Err... the transformer doesn't provide 12v.. have any clue what the size would be if it has a 12v output given the amperage of a 750W supply? These are insulator transformers part of the control loop and usually pass through AC at an 1:1 ratio. But I disgress, I've helped the OP in my first post.
Open up a Silverstone and reverse engineer the schematic

shinji2k

Err, yes it does. After the mains goes through rectifiers to convert to DC, the voltage is then doubled (for us 110V folk) or sent through the PFC circuit (rectification would occur after this if PFC is present), it is then converted to high frequency AC using switching transistors. From there it goes to the large main transformer where it is stepped down to the various necessary voltages (12, 5, -12 and -5, usually the 3.3V shares the same winding as the 5V which is then regulated later on). The reason a switching mode power supply is used as opposed to linear, is for the reason you mentioned. The transformer would need to be huge if you didn't convert to high frequency AC.

edit: Accidentally deleted something

In addition to the main transformer there is the one to isolate the PWM circuit like you mentioned and one for the 5VSB since that needs to be on all the time.

__________________
Quis custodiet ipsos custodes?

MagicBox

Sorry, I stand corrected. I've checked out several ATX schematics proving you're right on the topology described. It's not the SMPS layout I'm familiar with. It's clear though that in this topology every 12V rail has its own winding as shown by one scematic I ran into. There's only one controller loop, if too much power is drawn, all rails, from 5 to 3.3 and -12, they all go down. This topology is efficient, but not extremely accurate since basically every rail is dictated by one PWM controller, meaning rails drop in voltage when more power is drawn.

I work with SMPS where the output isn't stepped down by transformers, but by the switching devices themselves, using a different topology

shinji2k

I still have a lot to learn so no apologies are necessary . You had me questioning what I knew so I learned some new things in the process. And you are exactly right about increases in loads causing voltage to drop. That is a side effect of these designs. But it is usually a compact and efficient way to do it so a slight voltage drop over the load range is an acceptable consequence.

The numerous switching topologies are where I need to study. What are the SMPS you work with used for and why is that design more suited?

__________________
Quis custodiet ipsos custodes?

MagicBox

SMPS for audio power stages (class G), one hobby project of mine