toroidal cores

[Neuromancer]

Well i been playing around clocking my fx55 to 3 gs but need some juice to do so (1.525 vcore)
a8n-sli deluxe mobo


The ferrite rings between the mosfet sink and the CPU are REALLY hot. Now some research shows, this is NOT supposed to happen. (And it doesnt at stock vcore levels)

Also, the square black boxes i see on my cheap motherboards, with the copper stranding poking out of the top, I believe those are PWMs, the ASUS does not have any.

So are the toroidal cores in lieu of PWMs? Is there a way I can "upgrade" these to handle max vcore? Would this also help me lower the vcore requirement for 3ghz OC?

Thanks

[Super Nade]

Toroidal cores are completely different from Pulse Width Modulators. One is a passive elemnt whilst other is an active element. Could you post a few pics of the part you are talking about? I think you may be referring to the phase inductors, which usually have windings almost thicker than 18 AWG.

[four4875]

uhh... the little black things you're referring to are probably inductors, just inside of a casing like. the torroidal core refers to the core of the bigger inductor, and inductors are used in PWMs to smooth the power. you could put in ones to handle more current, you will need to keep the same rating for the henrys tho.

[Neuromancer]

Thanks Nade, yah not really sure all about PWMs except you can use them to control voltage..

http://img110.imageshack.us/img110/1852/hmmmvh2.jpg

[Neuromancer]

So these are phase inductors not PWMs....

Okay I read up a little on inductors, and from what I gather the inductance is directly opposite the input to avoid change. So by making larger inductance loops I would lose heat but further reduce the power.

Just found this article here
http://www.theinquirer.net/default.aspx?article=32945 (yah i know its the INQ but ...)


I have a feeling im just going to say screw it and get a c2d so i dont need to worry about vcore :P

[Super Nade]

Actually the inductors you alluded to are of a special kind as well. They are called ferrite cores/beads. What happens here is, at the rated freq band, electromagnetic energy is completely converted to heat. So suppose you have EMI at say 100kHz, that part of the freq spectrum is quenched by a ferrite core, hich basically takes energy in that freq band and converts it to heat. A very useful and convenient ripple filter. :)

[Skeith]

Actually the inductors you alluded to are of a special kind as well. They are called ferrite cores/beads. What happens here is, at the rated freq band, electromagnetic energy is completely converted to heat. So suppose you have EMI at say 100kHz, that part of the freq spectrum is quenched by a ferrite core, hich basically takes energy in that freq band and converts it to heat. A very useful and convenient ripple filter. :)
Yup. Due to the fact that they waste energy to heat they arent the most efficient form of ripple filter, but they are the cheapest and most reliable. Anything else gets very costly because they usualy use more passive and active components. Not only that but they get big. That is why the inductors are used in PC components instead, reliability and size and cost being the main concern. Hell who needs near 90% efficiency with the massive PSUs on the market today.

[Super Nade]

I think moderm MB's have a mix of both active and passive filters (atleast the expensive ones do). I suppose there is only so much filtering a MB can have to keep it cost effective. As you said, with good quality PSU's available these days (plenty adhering to ATX specs), there is no need to have heavy duty filtering, assuming that everything is at stock ;)

[GigaForce310]

Given that the core power supply is a 3 phase switching buck, passive compenents really don't do much in terms of filtering. They do assist active filters and are important components in a switching power supply.

I will now address the inductors that are pointed at on the ASUS board. It's probably good to know what they do and why they are so important.

The inductor acts as pseudo variable resistor. By nature an inductor resists a change in current. The moment that you place a voltage accrossed an inductor, it acts as if it has infinite resistance. On the other hand, if you have a voltage drop accrossed the inductor for a very long time, it will allow current to flow only restricted by the resistance of the wire. When the voltage is removed, the inductor again resists a change in current. The inductor doesn't convert the energy it receives from acting as a pseudo resistor to heat, but it stores this energy as magnetic flux. This allows the current to continue to flow untill the stored energy is released.

Just don't forget that the wire has resistance and any current flow will cause heat from the wire. Another source of heat is from the formation of eddy currents in the inductor core. If they used a solid soft iron core, the eddy currents would build up and create a "leak" of energy into heat. They are very interesting in brake applications, but avoiding them in inductor cores is important. Various powder cores are use to minimize the effect of eddy currents. You probably have also seen laminated cores in transformers and inductors. These have many thin painted soft iron plates instead of one solid core to reduce eddy currents.

Back to inductors, there is a charge and discharge curve that the inductor will follow. How much current can flow on the high voltage side is important. The less current it can handle, the shorter periods of time it can have a voltage acrossed the inductor or the current will get too high. More current will increase the amount of heat in the components.

The inductors are used with a pulse width modulator to regulate the amount of current passing them. The inductor is basically just restricts the flow of current and causes the voltage to drop beyond it. When the voltage is cut off, the inductor will release its charge by pushing more current from a diode through the inductor and into the capacitor. The capacitor is used to store the charge given to it by the PWM. One hit won't cause the voltage to rise more than a tiny fraction of a volt.

It takes a constant series of hit to keep the charge from being drained by the core. Actually, the feedback monitors the voltage constantly and the PWM is adjusted to the length that the next hit needs to be. The 3 phases constantly does this to keep the capacitors charged with the correct voltage. The pulse width changes to adapt to the instantaneous power demand. Any changes in power demand will be reflected in the following MOSFET firings.

When you increase the voltage, speed, and therefore the power that the core needs, you also increase the pulse width and the current that the bucks have to endure. All of the wires have resistances and produce heat, the inductor core has resistances that the eddy currents travel through that also cause heat.

The downfall is that the inductor needs to be much larger to handle higher currents and have the same inducatance. Higher frequencies have problems especially since MOSFETs are fast at switching, but not so fast to hard saturate. A couple microseconds to reduce the MOSFET resistance to near zero doesn't seem like much untill you look at the details. They show that the MOSFETs have to dissipate a large amount of heat at higher frequencies and currents.

A single phase buck would produce more of a ripple and would get very hot. Therefore a 3 phase buck has been in place for many years. The 3 phase allows one phase's inductor to discharge while another is charging and the last is somewhere between.

As of more recently, the answer to more current and cooler power supply components is more phases. The Asus P5W deluxe has 8 phases or 8 buck switching power supplies. More phases use more however smaller capacitors and inductors, but also run cooler.

The limitations of Buck power supplies are because of it's power limitations. Adding more phases reduces power per phase but it doesn't actually increase the power over a single buck by that much. Multiphase bucks can handle probably 200 watts max untill the problems overtake that method. A more advanced method is needed beyond this restriction.

When it comes down to your issue with heat, All you practically can do is stick a fan at it and let it cool down a bit under heavy load.

[four4875]

it wouldnt be possible to replace them with bigger ones, of heavier wire, with the same inductance rating, so they dont get as hot from the same load?

[Neuromancer]

WOW, man you guys are brilliant LOL <-- not sarcasm BTW, you guys make my tiny brain hurt

Actually I came up with a beter solution then using a fan!

Im selling the high vcore Fx55 and switching to a nice mellow X2 chip. (Since I never experienced this kind of heat before, I am assuming that the vcore is the problem, jumping back to a nice mellow 1.4v :))


Next motherboard I buy I will definitley look for ones with more phases :) (like the p5B)

[greenmaji]

Neur0mancer.. possibly a more effecent power supply (less ripple) would do the trick?

[Super Nade]

it wouldnt be possible to replace them with bigger ones, of heavier wire, with the same inductance rating, so they dont get as hot from the same load?

Those toroidal cores are an important part of the CPU VCore supply, i.e they are used by the buck-regulator to facilitate DC-DC conversion. Actually, a cap, diode, Inductor and PWM chip form a part of this supply setup. Replacing any inductor should be carefully evaluated. If you put the wrong one in, supply via one phase may be unbalanced. Note that to facilitate control over VCore, a FB loop is employed. Usually, a control system (s-parameter) analysis is done first to get the loop back into steady state quickly, if you unbalance it, it may either ring or take a long time to settle down.

[four4875]

thats why i figured keep the inductance rating the same, just a higher current handling capability, but have to keep same resonant frequencies and all that stuff the same too, prolly be hard to find a match.