Connect with us

Digitally Controlled Power Supply with Digital POT, Keeping the Digipot from being fried

Discussion in 'Electronic Design' started by Meek the Geek, Feb 24, 2006.

Scroll to continue with content
  1. Hi, My name is Mike, I'm a mechanical engineering student

    I'm building a digitally controlled electromagnet for a project,
    the system needs to be controlled between 0V to 24 Volts DC with up to
    4 Amps of current and I do not think that PWM will be a suitable
    control on this devce for a variety of reasons (it gets complicated,
    but trust me though, no PWM).

    I found a design on a website for an adjustable power supply, and
    given that for the purposes of experimentation we will be using a
    lab-bench DC power supply set to provide 30V DC power, and then
    regulate that current. It means that we can eliminate the transformer
    and bridge rectifier for now. The design I am proposing involves using
    a digital potentiometer to replace the manually controlled
    potentiometer shown in the design below

    http://www.aaroncake.net/circuits/supply.htm

    My problem right now is that the digital potentiometers that I
    have seen seem to be limited to only 5VDC. At the voltage which I
    intend to use, I am very sure that I will be blowing it up when I reach
    certain voltage levels. Is there a way of buffering the digital Pot
    such that I can avoid barbecuing it, while still being able to use it
    to provide digital power control. I was provided with an x9c103
    digital potentiometer from Xicor/Intersil (10K Ohm, with 100 wiper
    positions) http://www.intersil.com/products/deviceresults.asp?i=10763

    Is there a simple way to buffer this digital Potentiometer (use
    OP-Amps, a transistor, magic faerie dust) so that even though it only
    has a 5 Volt survival range, I can use it to control the larger
    voltage?

    Thanks in advance
     
  2. Ken Taylor

    Ken Taylor Guest

    Look up R/2R Ladder Networks, or mosey on down to:
    http://www.csse.monash.edu.au/courseware/cse2111/tut6.pdf

    and extrapolate from tehre.

    Cheers.

    Ken
     
  3. Rich Grise

    Rich Grise Guest

    Are you absolutely constrained to the LM338K? There are lots of adjustable
    power supply designs out there on the web, that are not too much more
    complicated, and do exactly what you need. Google up "Adjustable power
    supply" or whatever, and poke around.

    Good Luck!
    Rich
     
  4. Guest

    As Rich Grise says, the LM338 is not a good choice for your job. He
    should have mentioned that if you really want change the driving
    voltage from 24V down to 0V, the LM338 falls short because you can't
    pull the outptu voltage lower than 1.25V (check out the data sheet at
    http://cache.national.com/ds/LM/LM138.pdf).

    The thermal overload protection built into the LM338 is a very nice
    feature, but you can now get this in some power MOSFETs. Farnell lists
    "Philips Intelligent switches" where the BUK106-50L and the PIP3206R
    look quite attractive, the equivalent "fully protected MOSFET
    switches" from International Rectifier, and OmniFETs from ST where the
    VNW100N04 looks nice
    (http://www.st.com/stonline/products/literature/ds/4593.pdf).

    Since you want a linear circuit, you are going to have to dissipate up
    to 37.5W in the MOSFET (assuming a 30V supply and a coil resistance of
    6 ohms) which is going to mean a big heat sink and very good themal
    coupling between heat-sink and MOSFET.

    If you assume a maximum junction temperature of 150C, and a maximum
    ambient of 40C
    you are going to need at total thermal resistance of less than 2.9
    degrees C/Watt - of whixh at last 0.6C/W is going to be the junction to
    case termal resistance of the MOSFET (VNW100N04) another 0.5C/W between
    the MOSFET and the heat sink leaving you needing a 1.8C/W heat sink,
    which is on the large side (see Farnell order code 414-438, but I don't
    know if the central gap is big enough for a TO-247 package).

    Once you've got your series element, you've then got the fun of
    relating the voltage drop across the solenoid, which has to have one
    end tied to the +30V supply rail, and the other end tied to the drain
    of your MOSFET, to the votage that you et up with some kind of A/D
    converter, which will probably be referred to the 0V rail.

    You can use a differential amplifier (otherwise known as a subtractor)
    to generate a 0V-referenced voltage in the range 0V to 5V that is
    proportional to the voltage drop across the solenoid, and then you can
    use a standard op amp circuit to adjust the gate voltage of the MOSFET
    to product the desired voltage drop across the MOSFET. Note that the
    input capacitiace of the VNW100N)4 is big - at around 12nF - and you
    either need to drive with an op amp that will tolerate a large
    capacitative load on its output, or you have to isolate the output from
    the gate with a resistance (see the op amp data sheet for the
    appropriate value).

    All this will make the feedback loop relatively slow, so do a Bode plot
    for the loop as a whole at the very least ...
     
  5. I thank you all for your input

    Just so it is known, it is easy enough to incorporate a computer
    controlled switch at voltage close to 1.25V such that I can shut it
    off. I do not need to keep voltages of this low a value, because this
    is below an initial power value needed to produce any sort of results
    (in short, if I have power in this range, odds are I will be getting
    little if any result from the system).

    I am now seriously starting to think that maybe I would be better
    off with an array transistor/resistor simple DAC type network instead,
    coupled with an off-switch. Would this fare me much better? One of
    the problems is that getting weird IC components costs me too much and
    I have a limited timespan.

    What might serve as a better type of controller?
     
  6. I thank you all for your input

    Just so it is known, it is easy enough to incorporate a computer
    controlled switch at voltage close to 1.25V such that I can shut it
    off. I do not need to keep voltages of this low a value, because this
    is below an initial power value needed to produce any sort of results
    (in short, if I have power in this range, odds are I will be getting
    little if any result from the system).

    I am now seriously starting to think that maybe I would be better
    off with an array transistor/resistor simple DAC type network instead,
    coupled with an off-switch. Would this fare me much better? One of
    the problems is that getting weird IC components costs me too much and
    I have a limited timespan.

    What might serve as a better type of controller? cheap DAC
    resistor transistor network, or x9c103 digital pot with only the +5V
    limit before burnout? I sometimes cannot get everything where I am.
     
  7. Guest

    Why is 1.25V magical?
    Don't futz around with either, but use a proper DAC - one with a
    built-in voltage reference.

    They aren't expensive, and the last thing you want is to re-invent the
    wheel.

    None of the componets I mentioned were weird - you can buy them all off
    the shelf from Farnell, or any other broad-line distributor (like
    DigiKey). I don't know where you are, so I don't know how fast Farnell
    delivers in your area. In Europe, they seem to claim that if you order
    before about 4.00pm, you will have the parts in your letter-box the
    following morning.
    E-mail me if you need more details.
     
  8. Rich Grise

    Rich Grise Guest

    From your original application, it sounds as if what you're really looking
    for is a digitally-controlled current source. You want regulated AMPS
    through your solenoid, not necessarily regulated VOLTS, right?

    A current regulator isn't hard at all. Do you want it on the high side
    or the low side? Are you good enough of a designer to do it with PWM?
    Ever built an R-2R ladder?

    If you can afford the power budget, then one mongo transistor on a hefty
    heat sink, a current sense resistor, a feedback amp, and you control it
    any way you want to. :)

    Good Luck!
    Rich
     
  9. Joseph2k

    Joseph2k Guest

    I have no idea what others have said yet. Get the manual for that Lab
    supply and look for the characteristics of the voltage and current control
    interfaces. Post that info and you will get better results.
     
  10. Joseph2k

    Joseph2k Guest

    For that matter if it is really a digitally controlled power supply is the
    control interface RS-232 (serial) or IEEE-488 (or something else)?
     
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

-