Connect with us

PSU Fan Direction

Discussion in 'Electronic Repair' started by Caroline, May 14, 2004.

Scroll to continue with content
  1. Asimov

    Asimov Guest

    "LASERandDVDfan" bravely wrote to "All" (16 May 04 02:17:14)
    --- on the heady topic of "Re: PSU Fan Direction"

    LA> From: (LASERandDVDfan)

    LA> Call the phone company and have them correct the problem. They may
    LA> perform a test and say that it's okay, but tell them that this is for a
    LA> computer modem and while the line may be suitable for regular
    LA> conversation, it is apparently too noisy and, therefore, unacceptable
    LA> for use with a modem.
    LA> You're paying for their services, so make them deliver you that
    LA> service in the best way possible.

    Oh, no don't tell them it is for a modem or they may charge you more.
    The thing to tell them is that you hear noise and echos on the phone
    line. When the tech person hears the word "echo" he will tend do an in
    depth test for equalization and noise with the station rather than
    simply picking up the receiver and listening to the line.


    .... I lurk quietly and carry a big OFF/ON switch.
  2. w_tom

    w_tom Guest

    I am not sure where this whole discussion is going - why
    fans were some hardware solution.

    The modem. Until the problem is broken apart into separate
    entities, then problem will remain complex. Your phone line
    enters the house at a premise interface box called the Network
    Interface Device or NID. Open it. Unplug the household
    connecting wire. Connect you computer directly into that NID
    phone jack. Now test. Is problem on their side or in your

    If noise on phone line is a problem, then you can hear it
    when using conventional (POTS) phone.

    But then you can also demonstrate the problem using a
    program (with all Windows) called Hyperterminal. Simply use
    Hyperterminal to call your ISP. It will at least ask for
    Login: Is that clear? In you case, probably so because
    problem appears to be intermittent. But then Hyperterminal
    can also be used to communicate between two computers via two
    separate phone line - and actually see the noise - know
    exactly when it happens.

    Some modems can be more resistant to noise compared to
    others. But problem is not modem. That noise should not even
    be there. Having been exactly where you are, I first
    demonstrated the problem by setting up communication between
    two computer. Demonstrated to the phone tech a problem that
    existed in their digital fiber optic connection from my local
    phone to their local exchange. A problem that would cause
    intermittent failure. A problem that, instead, they kept
    sending line techs out to change phone wires. By
    demonstrating the problem, the telco discovered why they were
    having so many problems with everyone in that part of town.
    But the point is you must make the problem reproducible and
    clearly apparent.
  3. w_tom

    w_tom Guest

    Large power supply does not increase computer performance.
    That is classic urban myth. Either computer gets sufficient
    power to execute the instruction at same speed - or computer
    crashes. The PSU is not a motor. A bigger PSU does not make
    the computer work faster. Either the computer executes at the
    speed of that master crystal oscillator - or it crashes -
    stops working.

    Same applies to CPU heat. Again, either CPU runs as crystal
    oscillator speed or it crashes. Cooling does not make a CPU
    run faster.
  4. jakdedert

    jakdedert Guest

    Freeze up = crash...I think you're arguing semantics here. If the old p.s.
    was undersized for the computer--or if the output had deteriorated for some
    reason--the new, more powerful supply could well be the source of improved

    As far as air flow: IIRC, AT supply fans *exhausted* air from the supply;
    while ATX fans 'suck.' I don't know if that's a hard & fast rule, but I
    seem to recall the above as one of the changes when the ATX standard took

    If you have the technical chops, you can easily reverse the airflow by
    opening the p.s. and turning the fan around. If your CPU has no fan, by all
    means get one. If you are 'really' concerned about airflow within the
    enclosure, buy an auxiliary case fan--orienting it so that the airflow
    supports that of the p.s. fan. Every case with which I'm familiar already
    has a cutout with screw holes for one. Any computer store, electronics
    supply store or even Radio Shack carries these.

  5. Large power supply does not increase computer performance.
    True, a power supply that provides clean power will help make a system more
    reliable, but it won't produce performance gains. - Reinhart
  6. As far as air flow: IIRC, AT supply fans *exhausted* air from the supply;
    My ATX supply blows the air inside the case out, so it's an exhaust. -
  7. The thing to tell them is that you hear noise and echos on the phone
    Good point. - Reinhart
  8. Franc Zabkar

    Franc Zabkar Guest

    Most modems can produce a last call diagnostic report. This includes
    data such as Tx/Rx signal levels, Tx/Rx error rates, S/N ratio,
    numbers of speedshifts and retrains, etc.

    My own Rockwell chipped modem produces the data below. Notice these
    highly abbreviated data for a good session ...

    TX/RX I-Frame count : 17654/21091
    TX/RX I-Frame error count : 29/14
    TX Rate (Last/Init/Min/Max) : 28800/26400/26400/28800
    RX Rate (Last/Init/Min/Max) : 46667/46667/45333/46667
    Modulation/Protocol/Compression : V.90/LAPM/V.42bis
    Retrains (Issued/Granted/Fast) : 0/0/5
    Renegs (Issued/Granted) : 2/0
    Retrans per frame/Frames rejected : 1/14
    Error control timeouts in TX : 16
    Error control NAKs received : 29
    Termination Cause : Dte Hangup Command

    .... a not so good session ...

    TX/RX I-Frame count : 10268/34101
    TX/RX I-Frame error count : 15/47
    TX Rate (Last/Init/Min/Max) : 28800/26400/26400/28800
    RX Rate (Last/Init/Min/Max) : 38667/42667/38667/42667
    Modulation/Protocol/Compression : V.90/LAPM/V.42bis
    Retrains (Issued/Granted/Fast) : 0/0/0
    Renegs (Issued/Granted) : 3/0
    Retrans per frame/Frames rejected : 1/47
    Error control timeouts in TX : 7
    Error control NAKs received : 15
    Termination Cause : Dte Hangup Command

    .... and a bad session (water in the cable) ...

    TX/RX I-Frame count : 784/4482
    TX/RX I-Frame error count : 23/211
    TX Rate (Last/Init/Min/Max) : 26400/26400/26400/28800
    RX Rate (Last/Init/Min/Max) : 38667/44000/33333/44000
    Modulation/Protocol/Compression : V.90/LAPM/V.42bis
    Retrains (Issued/Granted/Fast) : 0/1/2
    Renegs (Issued/Granted) : 12/0
    Retrans per frame/Frames rejected : 8/211
    Error control timeouts in TX : 16
    Error control NAKs received : 23
    Termination Cause : Retrain Failed

    Notice that the initial CONNECT speed is not a reliable indicator of
    modem performance, as modems will speedshift as line conditions
    change. Notice also that the modem can tell you the reason for
    disconnect. For example, a "Termination Cause" of "Disconnect Frame
    Received" indicates that your ISP kicked you off.

    - Franc Zabkar
  9. Franc Zabkar

    Franc Zabkar Guest

    Actually, there are three types, "soft", controllerless, and "hard"
    (controller based). Softmodems have a DAA (telephone line interface),
    controllerless modems have a DAA and DSP (digital signal processor),
    and "hard" modems have a DAA, DSP, and controller.

    Among other things, a modem's controller handles AT command parsing,
    UART emulation, data compression and error correction. These functions
    do not impact significantly on the host CPU. OTOH, the functions of a
    DSP are highly CPU intensive, so a softmodem (which emulates the DSP
    in software) may impact noticeably on CPU performance.
    Actually, to be pedantic, "Winmodem" is a USR trademark and refers to
    their line of controllerless modems.

    Examples of softmodem chipsets are PCtel HSP, Motorola SM56,
    Smartlink, and Conexant HSF. Controllerless examples include Conexant
    HCF, Intel HaM, Lucent Win Modem, and USR Winmodem.

    - Franc Zabkar
  10. Franc Zabkar

    Franc Zabkar Guest

    AFAIK, the FCC mandates restrictions on the maximum signal levels, not
    the data rate. It's just that data rates in excess of 53333bps usually
    require signal levels greater than those allowed. At least that's the
    explanation that is often given at comp.dcom.modems.

    - Franc Zabkar
  11. Franc Zabkar

    Franc Zabkar Guest

    .... which would make it an inferior AC97 softmodem.
    Although "soft" USB and PCI modems do exist, they are most often
    controllerless, which makes them better than softmodems. As for
    "parallel" port modems, I've heard they exist, but I've never seen
    one. Serial port modems are always "hard".
    Don't you mean "serial", not "parallel"?

    What did this involve?
    Tell them you are having trouble sending/receiving faxes. The telco is
    not obliged to provide "computer" grade phone lines, only voice and
    Query the modem's last call diagnostic report, as described elsewhere
    in this thread.
    I wonder about this. Unless the phone cord is of extremely poor
    quality, I can't see how adding even 10m to several km of cable is
    going to affect a 4kHz connection ... assuming, of course, that you
    keep the cable away from sources of electrical interference.

    - Franc Zabkar
  12. Franc Zabkar

    Franc Zabkar Guest

    Controllerless modems (eg USR and Lucent/Agere "winmodems") have
    little impact on the host CPU. IMHO, their only significant down side
    is that they require drivers which makes them OS specific. "Soft"
    modems, though, are another story ...
    I have always found ACF/ACF2 chipped Rockwell/Conexant "hard" modems
    to be reliable perfomers.

    - Franc Zabkar
  13. Hi!
    Every "regular" sized ATX supply I've ever seen pushes air out the back with
    a fan mounted in the back of the supply, just like AT units have done for
    years. It is usually the oddball sized or really cheap models that seem to
    do things differently. Some ATX PSUs even seem to have fans at both ends,
    which strikes me as odd.
    My question is why do almost all computer manufacturers today bother putting
    these holes in place? Some (like HP) use them and put a fan there, but many
    do not. Most of them are even left open. I've only seen them closed up on
    Gateway 2000 computers. I find that especially odd considering that the hole
    for the extra fan is usually right under the PSU...which makes me think that
    all the air comes in that hole and not necessarily through the case as it
  14. Hi!
    Perhaps not. For the basic concept of a bigger PSU not making the computer
    work any faster I agree with you. Obviously it cannot do so. However, for a
    system starved for power, a crash may not be the immediate result. The CPU
    may operate with reduced efficiency and/or speed if the power supply is
    insufficient as a protection measure.

    Therefore the addition of a bigger PSU that can handle the load better than
    the original can create the illusion of the PSU somehow making for a faster
    Definitely not always. Some systems (especially laptops) are able to
    downclock so as to protect themselves from heat damage. Even some "smarter"
    desktop motherboards can monitor their temperature and fan RPMs so as to
    shutdown or downclock in the event of a thermal problem.

    As for the statement about cooling not making a CPU run faster, ask any
    overclocker about that. The more you cool the chip, the harder you can push
    it, up until a point of failure occurs or reliable operation can no longer
    be acheived.

  15. w_tom

    w_tom Guest

    Computers do not work at reduced efficiency if starved for
    power. This made so painfully obvious even from a data sheet
    for any simple logic IC. Either the computer works 100% fine
    or it crashes: 0% performance. Either it crashes / freezes or
    it keeps working. There is no inbetween. CPU does not
    change speed and efficiency like an eletric motor. Basic
    computer knowledge makes that woefully obvious.

    Either the PSU is insufficient and computer crashes - also
    called a complete and total failure. Or the computer works
    at 100% performance.

    This discussion is not about laptops - that have properly
    sized power supplies. Discussion limited to a system that
    somehow will run faster by increasing power supply.

    For that matter, take off the heatsink on an Intel CPU. It
    too will run slower - and not destroy itself like an AMD. But
    that is well beyond the context of this discussion - and
    should not be discussed in this thread.

    Will cooling a CPU that runs at the constant master clock
    frequency work more efficiently or faster when cooled more?
    No. But even worse, not one good technical reason is provided
    to justify these erroneous speculations. Even the reasons for
    cooling an overclocked CPU is not valid. Neither more cooling
    nor more power in a supply is going to make a CPU run faster.
    Either it works at full speed or it crashes.

    Most every desktop system works just fine on a 250 watt
    power supply. Not obvious from the so many technical experts
    who never even learned basic electrical principles; then
    advocated "More Power" as a solution. Far more damning are so
    many clone computers missing essential power supply
    functions. So instead of buying a supply with essential and
    necessary functions, those computer assemblers recommed "More
    Power". Just another reason for 500 watt power supplies that
    often cannot even output 500 watts.

    Bigger PSU is often a solution to failure by the computer
    assembler - who failed to learn basic facts. Bigger PSU does
    not make a CPU work more efficient or faster. Even concepts
    in overclocking are being misrepresented.
  16. w_tom

    w_tom Guest

    Franc is properly summarizing basic science as it conforms
    to FCC regulation. Concept defined in one of the world's
    greatest papers, written by Claude Shannon, reprinted in the
    Bell Labs Technical Journal in July and Oct of 1948, then
    reprinted again in Scientific American July 1949. IOW
    Shannon's Law defines modem speed. Underlying concept should
    be common knowledge to any computer power user. Concepts that
    Franc Zabkar has properly summarized.

    Power increases to permit 56K would create too much
    crosstalk. Power is limited by FCC regulation. Shannon's Law
    therefore says maximum data rates would be 53K.
  17. Ken Weitzel

    Ken Weitzel Guest


    But with all due respect, that is less than half of the

    Imagine this if you will. My system is running as your
    refer to it 100% fine. (the power supply is more than

    Then an additional demand is placed on the power supply,
    ie Explorer spins up all HDD's, a cd reader, a dvd reader,
    a floppy, and perhaps interrogates some USB devices.

    The available current is now perhaps woefully inadequate,
    voltage drops, and (in your words), is now 0% fine :)

    Most will say that the system is frozen or crashed.

    And they will say that the problem is in opening
    Explorer :)

    Take care.

  18. Ricky Eck

    Ricky Eck Guest

    This is what I originally meant, however, this discussion has made a major
    turn for the worst, so I backed out of it. Everyone has their "Facts" and
    "Opinion" that they read on papers and in magazines. However, I use true
    actual life facts. Mostly what I can see and touch. I seen where people
    built (including myself) their own computer. Well, like you said. The PSU
    was perfect for the original config. However, it was just for a CD Rom,
    Floppy, HDD, ect. The basic system. Then I added a CD Burner, A DVD Player
    (A total of 3 CD Drive Units), then added two more HDDs. Before I knew it,
    many of the apps I ran, were no longer running at the same rate (crashing,
    dragging, ect.). I was thinking that maybe there were additional programs
    running in the back ground. But there wasn't. So I started removing the
    power cables from the units installed, and lowe and behold, locks almost
    completely ceased. So I thought, what the heck, I upgraded the PSU, and I
    connected a higher wattage unit to my main. Connected everything up, and
    never had a problem with it.
  19. Ricky Eck

    Ricky Eck Guest

    I remember a while back, in the P1 chip days, a group of people built a
    tower system, that used a form of coolant. I don't think it was Freon, but
    something similar to it. I am not going to get fully into, mainly because I
    forget all the logistics of it. But I know their studies proved to increase
    the power of the chip almost 4 times the amount. It was a 75 MHz and ran at
    a 300 MHz rate. Like I said, it was many years back, and I just remember
    the basics. I seen only a year or so back, a tower case that used water to
    cool the chip. It worked the same as a radiator on a car. There was a
    "radiator" that sat on the chip, kinda like the Heat Sink does now. And
    there was another one in the back of the Tower, that fans blew through,
    blowing the heat out, and re-cooling the water, to be returned to the
    "Radiator" on the chip. If I remember correctly, this was for a dual chip
    system. But could be used for a single chip also.

    Another thing, too. Ever since I can remember, computer rooms have been ran
    around 68-72 degrees, to keep the computers cool. So, if cooling a chip is
    not important, how come companies continue to improve on the cooling systems
    of the Processor Chips in computers? How come Muti Million dollar companies
    spend more money on the conditions of their computer rooms (not meaning
    actual hardware), then they spend on insurance in a year (Just
    exaggerating)? There has to be some logic to it. I am sure Microsoft would
    not spend money where it is not needed.

  20. Computers do not work at reduced efficiency if starved for
    That is so, but a computer is composed of many more components than the CPU.
    Power supply problems can cause errors not so much on the CPU but with the
    drives, for one.

    What do you think will happen on a modern computer when the hard disk trips on
    account of inadequate power?

    Every component in your computer requires power to run. If the power supply is
    unable to operate to the demands of the entire system, you will have problems.
    While the CPU may be running hunky-dory, other parts may not run so well and
    you can have crashes and hangups. In power, you've got to deal with volts and
    amps, and both indicates watts.

    Volts measure flow potential, amps measure current level, and watts measure
    actual ability to perform work.

    Here's something for you to check out:

    As for overclocking, pushing a CPU at a higher rate of oscillation will put
    more strain on it. This effect is well documented.
    Not if you are running a high end P4 system which requires more than 250 watts
    and uses it's own dedicated power socket.
    A power supply that has a higher level of overhead helps the entire system run
    without problems. Running a power supply that is insufficient for the demands
    of the computer will strain it and the computer will not be reliable.

    As for overclocking, all it's about is pushing the CPU and RAM to levels beyond
    what the manufacturer had indicated. However, you need better cooling to pull
    it off as you do operate them at higher bus and multiplier levels, which puts
    more of a strain on them. - Reinhart
Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day