Announcement

Collapse
No announcement yet.

To overload or not to overload?

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

    #41
    Re: To overload or not to overload?

    Haha

    Budm, I am having a similar misunderstanding about my posts in a different thread!

    Behemot, you do appear to be jumping to conclusions.
    And missing some well-stated points.

    It's been a pleasure dealing with all present if I get canned here....
    The More You Learn The Less You Know!

    Comment


      #42
      Re: To overload or not to overload?

      Only people that did not do any research or ask question or only build circuits on the a piece of paper or use simulation on the computer. You will be surprised how many people thinks ground at any point are the same because it is tied to ground.
      Never stop learning
      Basic LCD TV and Monitor troubleshooting guides.
      http://www.badcaps.net/forum/showthr...956#post305956

      Voltage Regulator (LDO) testing:
      http://www.badcaps.net/forum/showthr...999#post300999

      Inverter testing using old CFL:
      http://www.badcaps.net/forum/showthr...er+testing+cfl

      Tear down pictures : Hit the ">" Show Albums and stories" on the left side
      http://s807.photobucket.com/user/budm/library/

      TV Factory reset codes listing:
      http://www.badcaps.net/forum/showthread.php?t=24809

      Comment


        #43
        Re: To overload or not to overload?

        Ideal ground, yeah, but nothign is ideal You can help that by beefing it, which I just did yesterday, but I still won't use it. Especially when it is on the other side of the board (and attaching the crocodile to Main ATX is tough ).
        Less jewellery, more gold into electrotech industry! Half of the computer problems is caused by bad contacts

        Exclusive caps, meters and more!
        Hardware Insights - power supply reviews and more!

        Comment


          #44
          Re: To overload or not to overload?

          As for the Intel directives, I understand that as having the two caps on every rail on the PSU, right?
          Less jewellery, more gold into electrotech industry! Half of the computer problems is caused by bad contacts

          Exclusive caps, meters and more!
          Hardware Insights - power supply reviews and more!

          Comment


            #45
            Re: To overload or not to overload?

            I've seen postgrads that have finished their course still put a buck switcher near an ADC, even for a a little microcontroller with a 10-bit one, it matters. Local ground now has a massive current loop and your ADC readings are useless.

            The true test is to fail. If you do not fail to make something work sometime, you will never learn. And I've had plenty of failures, but lots of successes, too.

            My battery analyser worked first time, or so I thought. And it did, with one analyser, on a desktop PC. Then, I connected five of them to a crusty old Win2000 laptop with two USB ports and ran a battery discharge overnight. USB voltage started sagging to 4V as the laptop cycled from charging to running off battery, and with 700mA load from 5 micros running continuously. I did a lot of battery tests then and had to throw away all the results because the ADC was going crazy, dips every 3 minutes, didn't realise it until I saw the power LED flickering!

            But you live and you learn.
            Please do not PM me with questions! Questions via PM will not be answered. Post on the forums instead!
            For service manual, schematic, boardview (board view), datasheet, cad - use our search.

            Comment


              #46
              Re: To overload or not to overload?

              Originally posted by tom66 View Post
              I've seen postgrads that have finished their course still put a buck switcher near an ADC, even for a a little microcontroller with a 10-bit one, it matters. Local ground now has a massive current loop and your ADC readings are useless.

              The true test is to fail. If you do not fail to make something work sometime, you will never learn. And I've had plenty of failures, but lots of successes, too.

              My battery analyser worked first time, or so I thought. And it did, with one analyser, on a desktop PC. Then, I connected five of them to a crusty old Win2000 laptop with two USB ports and ran a battery discharge overnight. USB voltage started sagging to 4V as the laptop cycled from charging to running off battery, and with 700mA load from 5 micros running continuously. I did a lot of battery tests then and had to throw away all the results because the ADC was going crazy, dips every 3 minutes, didn't realise it until I saw the power LED flickering!

              But you live and you learn.
              LOL, haha!

              We all have had our fair share of dumb mistakes! (Hopefully not life threatening though!)

              Like the first time I used a potentiometer, I put too much current through it and it sparked and caught on fire xD Or, when I accidentally (I was really unaware that I had done it until the end!) put a capacitor in backwards. There was a bubbling/boiling sound, then the power led dimmed, I look at the capacitor and it was bulged and tilted to the side, about to explode! I shut off power and ran out of the room
              Muh-soggy-knee

              Comment


                #47
                Re: To overload or not to overload?

                Well I several times onced soldered backwards whole ASUS board and one of them even turned on. Luckily only couple caps vented But that's bloody ASUS and their lame backward drawing.

                As for that load… Intes says caps accross load, but load is on +5 V SB, +3,3 V, +5 V, +12 V and -12 V so is it fine soldering two caps per every rail? Now that tantal has ESR of 1 ohm and 390 mA ripple. Would Nichicon PW 22 uF with 0,95 ohm and 120 mA be adequate replacement?
                Last edited by Behemot; 01-03-2013, 09:20 PM.
                Less jewellery, more gold into electrotech industry! Half of the computer problems is caused by bad contacts

                Exclusive caps, meters and more!
                Hardware Insights - power supply reviews and more!

                Comment


                  #48
                  Re: To overload or not to overload?

                  Since we opened several matters here, I have another one.

                  On some PSUs - not on all, usually on those with very bad ripple - I have noticed there are not just slight differences in ripple depending on where do I take the readings from, but sometimes differences in tens of percent.

                  Now I use spare ground wire with no load flowing through it. Than what is generaly the best place for voltage reading? Directly on Main ATX/other loaded connector, before/after the recommended capacitors, or on totaly different connector with no load attached? Because, it seems, the decoupling capacitors are doing very much difference. Seems Johnny G has them directly on the probe itself.

                  I found different site suggesting to take the readings as close to the measure point as possible.
                  Last edited by Behemot; 01-23-2013, 02:14 PM.
                  Less jewellery, more gold into electrotech industry! Half of the computer problems is caused by bad contacts

                  Exclusive caps, meters and more!
                  Hardware Insights - power supply reviews and more!

                  Comment


                    #49
                    Re: To overload or not to overload?

                    (dunno if this was the most recent place you discussed this subject, but anyway...)

                    did any of you measure impedance of mobo on 5 and 12V rails vs. impedance of those resistors?

                    i feel measuring in these 2 conditions is like measuring load of pure wood on the truck vs. a water container always half empty (same weight as wood) .and with truck constantly stopping/starting to move.
                    (looking the load on truck's engine/gearbox)

                    why not just measure the temperature of coils and caps on mobo that make vrm (on full cpu load), more ripple=higher temp (in the end that temp. destroys caps...)

                    or, just take the noise alone (ie ac components of the signal, via capacitor) and power a rather small bulb with it, more light=more noise! instant indicator of noise!

                    or rectify that (now ac) noise and measure with precise voltmeter.

                    intel pdf also suggests diff. scope probes: anybody used those?
                    (read the small print:



                    checked the price of those? checked how hard a those tek probes to calibrate?
                    http://www.eevblog.com/forum/testgea...-power-supply/ )

                    even the method is called "differential".


                    i would suggests different road: have one decent psu and one mobo to use as a load, and then try few ways of measuring and pick the one that gives you highest "resolution". then compare other psus to that one.
                    definition of "decent psu" is this: one that lasts a long time and that doesn't destroy mobos.

                    (particularly funny thing is that you even have dmms that can detect ripple/noise, and these need not be expensive ones...perhaps mastech makes such where it just so happens that sampling periods coincide well with noise waveform so a lot of last digit fluctuation=lots of noise )

                    either way, as others suggested, play with it, that way you'll learn the most.
                    while at play, think about your safety too!

                    to that extent about place to measure: why not measure at few places and see what you get, try mobo, try the load you made, try on molex, try on molex in faraday cage further away from mobo etc.

                    5 and 12V rails will be loaded the most on usual pc.
                    Attached Files
                    Last edited by i4004; 09-28-2014, 10:38 AM.

                    Comment


                      #50
                      Re: To overload or not to overload?

                      I think nobody who does not have stuff like Chroma is using differential probes. Anyway, using motherboard or something like that is not even wild guessing, that's useless. Sticking to Intels method is at least trying to be close. OFC everybody is probably measuring something slightly different, but at least people are close to it and constant. Messing with MB is not only pointless, it's wasting time for bad data.
                      Less jewellery, more gold into electrotech industry! Half of the computer problems is caused by bad contacts

                      Exclusive caps, meters and more!
                      Hardware Insights - power supply reviews and more!

                      Comment


                        #51
                        Re: To overload or not to overload?

                        The Intel guideline says that the filtering capacitor are to be used at the probe end. So when you take readings with your oscilloscope, you should fit those capacitors in your probe.

                        The probe does not need to be differential if you are using an isolation transformer to connect the PSU, or if your oscilloscope is battery powered.
                        Originally posted by PeteS in CA
                        Remember that by the time consequences of a short-sighted decision are experienced, the idiot who made the bad decision may have already been promoted or moved on to a better job at another company.
                        A working TV? How boring!

                        Comment


                          #52
                          Re: To overload or not to overload?

                          https://www.youtube.com/watch?featur...3eduRj4#t=1913
                          (wouldn't hurt to watch complete video either! <wink>
                          and now the thread has mikeselectricstuff and dave jones videos, what more do you need? hehe)

                          unique, i would say your psu will have "dirty ground" even with isolation transformer, because it's the way it works. that's why you need differential probes to do this properly.
                          (and that's why i'm suggesting totally different concepts...heh..because it soon gets rather complex and/or expensive...would you like pre-amplifier to better see those signals? i would...)
                          btw. where does intel guideline mention where to put caps?

                          behemoth i don't think any measurement is a waste of time. if you see nothing you still learned something. <wink>

                          i would, for example, like to use 10ghz scope so i can see all the noise and can analyse it precisely (you would eventually learn what noise is cpu producing, what south bridge etc)...hehe...but these are not cheap.

                          did you try something basic, few different psus powering the pc and scoping the 5 and 12V on molex which would be as far away from pc as possible? with those diy probes dave mentiones... <wink>

                          aim should be to know what kind of noise and ripple is flowing thru the wires that connect psu and mobo. you don't know that with purely resistive load.
                          which is not to say it's totally useless, but you're not measuring what's really happening.

                          btw. you know what jonnyguru lacks? tests with new and then (after few years) used psus. i would prefer a psu with slightly more noise that remains on same noise levels to the one that falls apart after few years and destroys my mobo/hdds....
                          Last edited by i4004; 09-28-2014, 03:55 PM.

                          Comment


                            #53
                            Re: To overload or not to overload?

                            "i would prefer a psu with slightly more noise that remains on same noise levels to the one that falls apart after few years and destroys my mobo/hdds." falls apart after few years due to?
                            Never stop learning
                            Basic LCD TV and Monitor troubleshooting guides.
                            http://www.badcaps.net/forum/showthr...956#post305956

                            Voltage Regulator (LDO) testing:
                            http://www.badcaps.net/forum/showthr...999#post300999

                            Inverter testing using old CFL:
                            http://www.badcaps.net/forum/showthr...er+testing+cfl

                            Tear down pictures : Hit the ">" Show Albums and stories" on the left side
                            http://s807.photobucket.com/user/budm/library/

                            TV Factory reset codes listing:
                            http://www.badcaps.net/forum/showthread.php?t=24809

                            Comment


                              #54
                              Re: To overload or not to overload?

                              Originally posted by Th3_uN1Qu3 View Post
                              The Intel guideline says that the filtering capacitor are to be used at the probe end. So when you take readings with your oscilloscope, you should fit those capacitors in your probe.
                              I tried embedding them (not directly on the tip but between the probe and O-scope), it filters everything. But I am doing the measurement as close to them in circuit as possible. Actually, at some points I am connecting the probe to the electrodes of the caps. Sure it can be made better, just not now.

                              i4004: as a matter of fact, I tried some of that and I can tell you this, you have no idea what you are talking about. The thing is, nobody cares about actual noise. According to Intel guidelines, you are simulating motherboard and than measuring what comes out of that. That includes limiting the band to 20 MHz (so some GHz are ), connecting or terminating especially the long wires so you don't catch radio frequencies etc. If you just stick a probe into that you will read things several hundreds percent off.
                              Less jewellery, more gold into electrotech industry! Half of the computer problems is caused by bad contacts

                              Exclusive caps, meters and more!
                              Hardware Insights - power supply reviews and more!

                              Comment

                              Working...
                              X