Connect with us

Looking for advise on PC FlowMeter

Discussion in 'Electronic Basics' started by GTD, Mar 16, 2005.

Scroll to continue with content
  1. GTD

    GTD Guest

    Hello all, I have a Water Cooled PC, and I would like to make my own
    flow sensor. The reason I am wanting to do this, aside from the basic
    "I wanna do something I've not done before" , is that every flow
    sensor I've seen has been a weenie little 1/4 hose type (My system
    uses 1/2" Hose and has quite a bit of flow, and I really don't want to
    split and re-route the flow). I can easily build a "paddle wheel"
    enclosue, and my idea is to sense the speed of the paddle wheel, and
    have the circuitry sent an appropriate pulse signal to one of the
    motherboard's fan headers, where it can then be read by MBM or
    whatever software I use. This unit will not have any exposed shaft, so
    I am basically limited to a magnetic pickup arrangment (I will embed
    the magnets in the paddlewheel if need be), or a photocell type (with
    the paddlewheel breaking or reflecting the beam). If anyone has an
    idea of where to find schematics for a project such as this, I would
    appreciate it. I must say that I've done a bit of electronics, but it
    has been quite a few years. Thanks.
  2. Lord Garth

    Lord Garth Guest

    Nice but you will have to have a way to multiply the paddle wheel pulses
    into the KHz range. Most system fans want about 3500 RPM to prevent
    a 'fan to slow' alarm. I am assuming that one revolution of the fan results
    in one pulse on the tach, you'll have to test that theory. Perhaps a phase
    locked loop (PLL) with a divider in the feedback path will do. Look at
    the spec sheet for a 4046 here:

    Don Lancaster has a section in his 'CMOS Cookbook' regarding this device.
    Don's web site is

    I'm guessing that there is a pump motor in this cooling system, what about
    stealing a tach signal from there?

    How about a temperature to frequency circuit?
  3. GTD

    GTD Guest

    Thanks for the ideas, much appreciated. I could and probably will set
    it up with 3 or 4 magnets, so it'l be that many pulses per rev. It
    will have to have at least two just to keep the paddle balanced. The
    software I'll be using is very flexible, and the fan speed alarm do
    not have to be used.
    I'll check it out, thanks.
    This won't help me if a line gets plugged, the pump goes ntts, or
    something stupid like that. Also, I'm interested in seeing just a
    simple drop in flow rate as opposed to waiting for a complete stoppage
    if something bites it.
    I'd rather not wait for the temp to go up, by then it may be too late.
  4. Rich Grise

    Rich Grise Guest

    You could make it dead simple. Mount a reed switch where it's actuated
    by your paddle wheel magnets. Ground one side of the reed switch, connect
    the other side of the switch to your input pin, with 4K7 pullup to +5V.
    Maybe a 1 nf cap across the switch, just because you can. (try it with
    and without a cap, and see if it affects your debounce code).

    Debounce it in software - you've got all day to do that, practically. :)

    Have Fun!
  5. Lord Garth

    Lord Garth Guest

    You could recover the hall effect sensor from an old fan and use it directly
    as there would be no contact bounce. Many newer fans contain an IC that
    includes the coil drive transistors and the 4 or so passives that the old
    needed. I have yet to investigate these for potential reuse.

    If the cooling system plugs up and the heat rises, the motherboards should
    already take care of this situation by slowing or shutting down besides, the
    temperature rise is a simple scaling operation ... how much temp to tolerate
    before it is a problem.

    For the plumbing to stop up, assuming the pump continues to run, you could
    use a change in fluid level within a bottle to detect this situation.
  6. Bob Masta

    Bob Masta Guest

    Another approach would be to skip the moving parts
    completely, and just hang a transistor or diode junction
    in the flow path.
    Apply a small current to get it to self heat, and measure
    the voltage drop across the junction. This is linearly
    proportional to temperature, so you can thus determine
    how much cooling is taking place. There are all sorts
    of schemes based around this concept, often involving
    a second such sensor in a bridge arrangement.
    It's commonly used for airspeed measurement.
    You may want to look up "hot wire anemometer",
    since this was originally done with resistance wire
    instead of junctions.

    Best regards,

    Bob Masta

    D A Q A R T A
    Data AcQuisition And Real-Time Analysis
  7. GTD

    GTD Guest

    I've thought of that, and have been going through all my crap looking
    for a 3-wird fan, no luck so far, but I'm sure I'll find one.
    This is an AMD based system, sop there is no thermal throttling. The
    fact that the system is overclocked/overvolted means the temp will
    rise fairly quickly, but I'm sure the onboardprotection systems will
    work well enough. Basically, my need for a software-monitorable flow
    sensor is more of a want than a need. It is also to help me test
    different pumps, waterblocks, ect.
  8. Lord Garth

    Lord Garth Guest

    These sensors or in every fan as their purpose is to replace the mechanical
    A typical third wire tach fan simply outputs one of the coil pulses. That's
    why you should
    not pay a big premium for this 'cheap to include' feature.

    I have one fan that has a stall sensor built in. I suppose the output is
    constantly reset by a
    moving fan. In any case, you get a signal if the fan stops.
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