Announcement

Collapse
No announcement yet.

Very basic questions you're afraid to ask

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

    Very basic questions you're afraid to ask

    Like a few of you, I've always been intrigued by electronics, but my academic/professional path took a very different turn. As a result, there's a number of very simplistic questions I've been wanting to ask a "pro", but found it difficult to do this without appearing a fool. But on this forum I'll be brave, as usual - my apologies if these turn out to be extremely banal, or poorly worded.


    I invite any others with such questions to "confess" them in this thread!


    1. Why do certain ATX PSUs "post-regulate" their relatively unused 3.3v line with a mag-amp via a "returning" sense-wire, but not the more heavily-loaded 5v and 12v lines?

    2. In a situation in which one is dealing with a predictable, relatively constant load - is it possible (in your opinion/experience) for a PSU using linear regulators to be as efficient (or more so) than a switching/inverter-based PSU?

    3. I understand that with AC power transmission, when the load's apparent power is greater than real power, this translates into some form of "loss", from the generator's point of view. So in a hypothetical scenario with a PF of zero, no work is done, but the generator is running, so where would all the "lost" energy wind up? Heat at the generator? On the wiring leading to the load?

    4. And what is that component, always found in ATX power supplies between the output rectifiers and the smoothing capacitors - a prominent, large, toroid, with multiple, separate windings, usually of red and yellow wire? It's placed in series with the circuit, so my guess is that it's an inductor to reduce high frequency ripple - is this correct? If so, why is it done this way, as opposed to simply using separate coils for each 'rail'?

    #2
    Re: Very basic questions you're afraid to ask

    1) 3.3V line used to be used ( ) to power chipset, memory and AGP slot directly on boards around 1997-2000 - and as the voltage is low (thus current high), voltage drop across the long cables can be significant and cause problems for voltage sensitive parts

    2) For really constant load and constant input voltage, you don't need regulator - a resistor will do as the voltage drop across it will be constant too. A linear regulator can be seen as a "smart resistor" that changes its resistance according to the load. So a linear regulator converts excess voltage to heat.

    4) One wire on the toroid is for 5V and the other one for 12V. The winding is designed to improve cross-load regulation.

    BTW. I'm studying software engineering - but hardware seems to be more interesting

    Comment


      #3
      Re: Very basic questions you're afraid to ask

      Originally posted by Rainbow
      1) 3.3V line used to be used ( ) to power chipset, memory and AGP slot directly on boards around 1997-2000 - and as the voltage is low (thus current high), voltage drop across the long cables can be significant and cause problems for voltage sensitive parts
      Right! So it's partly a vestige of the past, I guess a little like those AUX connectors one still finds on ATX1.3 PSUs...

      Originally posted by Rainbow
      2) For really constant load and constant input voltage, you don't need regulator - a resistor will do as the voltage drop across it will be constant too. A linear regulator can be seen as a "smart resistor" that changes its resistance according to the load. So a linear regulator converts excess voltage to heat.
      So for a relatively predictable, almost-constant load, use of linear regulators would be generally yield greater efficiency? Would the losses of the large transformer at 50-60Hz taken alone generally be less than those of the input rectifier and high-frequency inverter?

      Originally posted by Rainbow
      BTW. I'm studying software engineering - but hardware seems to be more interesting
      I couldn't agree more. Paper/logic/theory is all very cool, but hardware/apparatus has an added appeal - a solidity/plasticity if you will...

      Comment


        #4
        Re: Very basic questions you're afraid to ask

        +3.3v: AFAIK, +3.3v is still is use on most motherboards to power buffers, chipsets, PCI/AGP slots, and even the Vio of the CPU buffers. Almost everything except Vcore, Vdimm and a few TTL-level logic blocks is powered from +3.3v - it's likely to be the highest-current rail. Rapid changes in load current on +3.3V cause voltage changes that are more difficult to regulate to within +/- 5% using the main PWM control loop on the SMPS, so there's usually a post-regulator using a mag-amp for quick, fine control (say, +/- 100 mV or so), in better-designed PSUs.

        Linear vs. Switching:

        Linear regulators are reasonably efficient only for relatively small diiferences between the input and output voltages, and for small swings in the input voltage. For instance, a fairly efficient linear regulator can be built that provides +48 VDC for an input range of 52-56V. Even in the worst case, it will be more than 85% efficient.

        The bulk of the weight as well as the cost of materials in off-line linear PSUs would go into 50 Hz magnetics, in particular, the step-down transformer. The main advantage of switching power supplies is that this size/cost is drastically reduced by switching at ~50kHz or higher.

        For maintaining high efficiency at high output currents over a relatively wide input range (like 70 to 260 V universal-input PSUs), there's no way to do it except using switching PSUs.

        Comment


          #5
          Re: Very basic questions you're afraid to ask

          Single large secondary toroid:

          That's one of the cleverest optimizations in multiple-output forward converters - a single large toroid is less expensive than several smaller toroids, both in material and assembly costs. The toroid and the main ferrite-core transformer are the two most expensive items in a switching PSU, so the multiple-winding optimization is a significant saving. It also provides an inexpensive way to improve the cross-regulation between the +5v and +12v by taking advantage of the common magnetic-circuit of the toroid - the voltage drop on each winding will get reflected on the other winding by the transformer effect.

          Comment


            #6
            Re: Very basic questions you're afraid to ask

            Originally posted by linuxguru
            +3.3v: AFAIK, +3.3v is still is use on most motherboards to power buffers, chipsets, PCI/AGP slots, and even the Vio of the CPU buffers. Almost everything except Vcore, Vdimm and a few TTL-level logic blocks is powered from +3.3v - it's likely to be the highest-current rail. Rapid changes in load current on +3.3V cause voltage changes that are more difficult to regulate to within +/- 5% using the main PWM control loop on the SMPS, so there's usually a post-regulator using a mag-amp for quick, fine control (say, +/- 100 mV or so), in better-designed PSUs.

            Linear vs. Switching:

            Linear regulators are reasonably efficient only for relatively small diiferences between the input and output voltages, and for small swings in the input voltage. For instance, a fairly efficient linear regulator can be built that provides +48 VDC for an input range of 52-56V. Even in the worst case, it will be more than 85% efficient.

            The bulk of the weight as well as the cost of materials in off-line linear PSUs would go into 50 Hz magnetics, in particular, the step-down transformer. The main advantage of switching power supplies is that this size/cost is drastically reduced by switching at ~50kHz or higher.

            For maintaining high efficiency at high output currents over a relatively wide input range (like 70 to 260 V universal-input PSUs), there's no way to do it except using switching PSUs.
            I believe that on most socket A motherboards, including the Asus A7N8X family of motherboards use to +5V rail for the CPU. +3.3V probably is mainly used for the video card. Possibly the DIMM slots, too.

            The Abit NF7 and Abit AN7 motherboards use the +12V rail for the CPU, AFAIK.
            ASRock B550 PG Velocita

            Ryzen 9 "Vermeer" 5900X

            16 GB AData XPG Spectrix D41

            Sapphire Nitro+ Radeon RX 6750 XT

            eVGA Supernova G3 750W

            Western Digital Black SN850 1TB NVMe SSD

            Alienware AW3423DWF OLED




            "¡Me encanta "Me Encanta o Enlistarlo con Hilary Farr!" -Mí mismo

            "There's nothing more unattractive than a chick smoking a cigarette" -Topcat

            "Today's lesson in pissivity comes in the form of a ziplock baggie full of GPU extension brackets & hardware that for the last ~3 years have been on my bench, always in my way, getting moved around constantly....and yesterday I found myself in need of them....and the bastards are now nowhere to be found! Motherfracker!!" -Topcat

            "did I see a chair fly? I think I did! Time for popcorn!" -ratdude747

            Comment


              #7
              Re: Very basic questions you're afraid to ask

              1. The spec for +3.3V regulation was tight enough that it couldn't be met without both a regulator and a sense wire. The sense wire allows the regulator to compensate for voltage drop across the +3.3V "+" wires.

              2. A linear regulator is inherently dissipative, but if the load and I/P voltage are constant, then the transformer (50/60 Hz) turns ratios can be designed to achieve very good efficiency. What hurts the efficiency is that the transformer has to be designed for 90VAC (or whatever the designed low-line point is) at 47Hz and maximum load. Thus, in actual operation, the I/P voltage to the linear regulator is much higher than it needs to be, and that extra voltage translates to power dissipated by the regulator. In a PC P/S, the -5V and -12V often have linear regulators (LM320-X or 790X types). Because the I/P voltage range to those regulators doesn't vary much (the main loop compensates for AC line voltage variations), this doesn't hurt the overall efficiency of the P/S (plus they are low current O/Ps anyway).

              4. An inductor is at the heart of a SMPS circuit, one "inductor" per O/P. It stores energy during the swtich "on" time, and delivers energy to the load during the switch "off" time. The I/P voltage to the inductor is a square wave; the O/P voltage is DC with some ripple,

              V(out) = V(in, peak) * [T(on) / (T(on) + T(off))]

              Having the "inductors" for each O/P share a single core accomplishes two things:

              * It saves space (imagine separate inductors for the +5V, +12V, -12V, and -5V O/Ps, each with its own core);

              * It improves regulation (the numbers of turns for each winding is chosen to accomplish this) on the +12V, -12V, and -5V O/Ps.

              It may also save some $$.
              PeteS in CA

              Power Supplies should be boring: No loud noises, no bright flashes, and no bad smells.
              ****************************
              To kill personal responsibility, initiative or success, punish it by taxing it. To encourage irresponsibility, improvidence, dependence and failure, reward it by subsidizing it.
              ****************************

              Comment


                #8
                Re: Very basic questions you're afraid to ask

                I'd like to know how much of what's discussed here applies to the laptops?
                They seem so much simpler and can be almost as powerfull.
                A power brick can supply 12VDC at a few amps to a plug in for a board that is already battery powered.
                There are very few if any Electrolytics in it.
                Comparing a desktop to a notebook may be like comparing a car to a motorcycle but, well?
                Last edited by arneson; 04-02-2006, 11:05 PM.
                Jim

                Comment


                  #9
                  Re: Very basic questions you're afraid to ask

                  > I believe that on most socket A motherboards, including the Asus A7N8X family of motherboards use to +5V rail for the CPU. +3.3V probably is mainly used for the video card. Possibly the DIMM slots, too. The Abit NF7 and Abit AN7 motherboards use the +12V rail for the CPU, AFAIK.

                  Most CPUs have split power planes - a standard 3.3v for the I/O buffers that interface with the rest of the motherboard logic, and a lower-voltage Vcore for all the internal logic (there may be additional biasing voltages, as well). Vcore is usually derived from a VRM circuit (either modular, or integrated on the motherboard), which draws power either from +5v or +12v or both. Generally, Socket A VRMs operate on +5v, while Intel VRMs on P4 boards use a separate 4-pin +12v connector.

                  +3.3v is extensively used on the motherboard, but often with separate, small, linear regulators. A Via Socket-370 board I have in front of me has separate regulators for Vdimm (3.3v regulated from +5v), Northbridge Vcc (2.5v regulated from +3.3v), and a biasing voltage (1.5v regulated from +3.3v), in addition to Vcore derived from +5v.

                  Comment


                    #10
                    Re: Very basic questions you're afraid to ask

                    Allow me to thank all of you for the fantastic and extensive replies to my questions.

                    Regarding "question 1", I just noticed that the Supermicro/LiteON PSU I had started a thread about a few weeks ago has returning 'sense' wires for both 3.3v and 5v lines, and even has an unused connector on the PCB for a 12v sense wire.

                    I'll be posting pictures and more details of this in the appropriate thread ( https://www.badcaps.net/forum/showthread.php?t=1537 ) later today.

                    Comment


                      #11
                      Re: Very basic questions you're afraid to ask

                      Originally posted by tiresias
                      Regarding "question 1", I just noticed that the Supermicro/LiteON PSU I had started a thread about a few weeks ago has returning 'sense' wires for both 3.3v and 5v lines, and even has an unused connector on the PCB for a 12v sense wire.
                      Antec SmartPowers also have an unused place on the PCB for a sense wire, +5V, and by installing it I got a 0.1V increase at around 10A. I hae a old, old Powmax (not junk) with a ground sense wire as well.

                      Comment


                        #12
                        Re: Very basic questions you're afraid to ask

                        Originally posted by larrymoencurly
                        Antec SmartPowers also have an unused place on the PCB for a sense wire, +5V, and by installing it I got a 0.1V increase at around 10A.
                        Right - so it was fully functional, but not implemented. Had they just factory-shorted the 5V to 5Vsense somewhere on the PCB? I should take a look at whether the same has been done to the 12Vsense on this LiteON.

                        Originally posted by larrymoencurly
                        I have a old, old Powmax (not junk) with a ground sense wire as well.
                        A ground Vsense, interesting approach! Do know you which line the sensed "ground voltage" is being compared to?

                        Comment


                          #13
                          Re: Very basic questions you're afraid to ask

                          Can you tell me, how can we check the ATX SMPS (PSU) Directly Just By Shorting.

                          From the ATX 24pin connector there are two such wires that if we short those two wires SMPS fan starts spinning.

                          Can you tell me clearly about it.

                          Comment


                            #14
                            Re: Very basic questions you're afraid to ask

                            jarus, Google, jumping PSU. There are many mini-tutorials online.

                            It is considered unwise to power-up a PSU w/o some load on it, like a couple fans.

                            Comment


                              #15
                              Question regarding filter caps near IC's.

                              Recently, I had a hellish time with a simple Analog ckt. I finally figured out that I had missed placing filter caps from VCC-->GND and VEE-->GND on the Op-Amps. I didn't pay attention to this because it was not a high speed ckt nor was the PCB dense enough for me to warrant such a thought. Boy, was I wrong or what! Had a horrible time debugging the severe ringing and RF pickup noise in the order of 10mV!!

                              Now, I wonder if all IC's on motherboards etc have filter caps associated with them? If not, is it advisable to add a pair of 0.1uF tantalum across the IC's?
                              Q6700 @ 3.6 GHz
                              Zippy GSM-6600P
                              Curcial Ballistix PC6400 (4 x 1Gb) Micron D9GMH
                              Abit IP35Pro
                              ATi HD4870

                              Comment

                              Working...
                              X