Connect with us

Linear Power Supply design questions

Discussion in 'Electronic Design' started by John T, Nov 12, 2003.

Scroll to continue with content
  1. John T

    John T Guest

    Hi, I'm sure everyone is pretty tired of basic power supply
    questions but the more I look at the info and try and figure
    it out on my own, the more worried I get that I'll screw it up.

    I need to build a linear power supply for a stepper motor
    project. I need 5v for logic, 24v for some relays and around
    24-30v high current for powering the steppers (perhaps up to 10A).

    I want to try and use as many pieces that I already happen to have.

    I have an old toroidal transformer that I want to use but I don't
    know much about it. It was pulled a long time ago from a working
    printer of some kind so it should still be good. The unit is
    about 5" diameter by 2.5" high and 8lbs so I guess about 240 - 460VA?
    It's encased in a metal "can" for mounting and exposied in from the underside.

    It has the following markings on the side:
    Transformer
    A740D-115
    JK 27T SMD
    (I have a second unit actually, marked JK 14T SMD - so probably a date code?)

    Which, when I did a search I found only the following:
    4408218690 Dataproducts model 2600 Transformer A740d 115v
    PLEASE NOTE: 4408218690 is replaced by 811557-001
    811557-001 Dataproducts model 2600 Transformer A740d 115v
    (http://www.theprinterplace.com/dataproducts/dataproducts-spare-parts54.html)

    I'm having trouble figuring it out but here are some observations (in pairs):
    yellow/yellow 22AWG
    red/red 18AWG
    blue/brown 18AWG
    white/black 20AWG
    purple/orange 20AWG

    I had hooked it up a while back on the white/black (I don't remember what
    my rational back then was for doing that) and so when I tried it like
    that now I get approx. (no load):
    yellow/yellow 30V
    red/red 30V
    blue/brown 30v
    purple/orange 108v

    When I plug a wall wart ac onto the blue/brown (6.7v) I get (approx.):
    yellow/yellow 6.7v
    red/red 6.7v
    white/black 27.5v
    purple/orange 24.5v


    The blue/brown go to a single connector. All the rest go to another connector
    with one of the yellow wires being interupted by another connector and one
    of the red wires also being interupted by another connector. The connectors
    interupting the yellow and red wires were probably going to a switch or
    relay which controlled power in the machine.

    So I'm guessing I've got the pairing correct and that the purple/orange is
    actually another primary.

    There is actually voltage showing up between any two wires when tested but I
    don't know if they are interconnected or leaking or I don't know what I'm doing?

    I tried to test the resistance but my readings didn't seem to make a lot of
    sense. Of course, my meter is really cheap.


    I'm a little unclear on hooking up the primaries in parallel (for 120v use).
    It shouldn't matter which wires I match (black with purple and white with
    orange for example) right? But they are at different voltages (above) so
    they aren't ballanced? that's a problem isn't it? (for parallel)

    While I'm still on the primary side of the transformer let me ask a couple
    of other questions. First, I also have some AC line filters. I don't really
    understand these things except that they are like a capacitor to filter line
    noise. Should I try and put in the highest rated one I can find?

    Second, I have a switch I want to use that is rated at 20A @ 125V and is DPDT.
    Would I be better off using it to switch only the hot wire using both sides of
    the switch or should I have it switch both hot and neutral? (ground, of course,
    is never interrupted)

    Now, on the secondary side of the transformer I put a bridge. I have the
    following ones available (with what I think they are):
    IR 250JB2L (25A 200V)
    IR 35MB10A (35A 100V)
    GI 95582009 - 8419
    GI KBPC25-06-14LD 95582004M - 8536 AC (25A 600V)
    Toshiba 15B4B41od (10A 100V)

    If I want to combine the red/red with the blue/brown to provide more amps
    I join them in parallel before the bridge not after a couple of bridges, right?

    The 35MB10A is probably my best choice even though it's only got 100V inverse.
    And, say, the 250JB2L for the yellow wires.

    After the bridge comes the big capacitor (or several). Going into my
    box I find a whole bunch rated at or above 50V: 31,000uF, 61,000uF, etc.
    I think that they are all electralytic but one says "type DCM" (by Sangamd)
    but maybe that's product line type. I can probably combine them to any
    size needed and I figured maybe a 61,000uF for my high current stuff and
    a 31,000uF for my low current.

    I know that it has to be sized correctly. Definately big enough but not
    too big. To avoid the startup current that the thing will draw from killing
    the bridge should I put one of those big resistors that I have in series
    before the cap, right? Like 0.1 Ohms @25watts or 5 Ohms @50watts or higher
    like 36 Ohms @50watts. A little extra heat during operation is worth it, right?


    I'd like to use the yellow/yellow for both the 5v and 24v low current needs.
    I guess I'd use a fixed voltage regulators like the 7805 and 7824 after
    the capacitor. Is there a problem with having the two in parallel off the
    same supply? (Maybe I'll throw in a 12v one too). I need a 0.33uF cap on
    the input and a 0.1uF cap on the output of each, if I read the spec correctly.
    My current needs should be well below an amp but I'd get the 78xx versions
    rated for 1.5Amps.

    I also have big heat sinks and there will be a fan in the case. Can I put all
    the bridges and regulators and perhaps some solid state relays onto one heat
    sink? Not because of the amount of heat but because of those metal mounting
    tabs. I know that the tab on the voltage regulator is ground and not
    electrically isolated from the component. Some of the bridges are all plastic
    and some have metal. I have some of those electrical isolation heat sink
    mounting stuff for the regulators (but they come from the same ground anyway).
    The solid state relays will be for switching AC for accessories.

    Then there is the high current section. The controller I'm using can handle
    up to 35V including the potential back EMF from the motors. This part confuses
    me because I'm not sure what will happen once rectified if there is no real
    load. It would be easiest to run it unregulated but if not how can I
    regulate it? I would need a circuit with one of those big power transistors
    (I've probably got many of them anyway).

    I will probably put a relay on the 24v that, when off, will switch resistors
    across the caps to discharge them. When the unit is switched on and the relay
    comes on, I will have it connect the 5v to the controller and enable
    another relay (self latching, controlled by a big push button red/green switch)
    to power the motors.

    Thanks for any help you can provide,
    -John
     
  2. Looks like no one else is biting, so I'll take a couple of these:

    doing?

    Try putting a 10k resistor in parallel with the scope and see if you still
    get a reading. If it's just a leakage, this will swamp it.
    They need to be in phase with each other. To test, take that AC wall wart
    of yours and hook it up to your low-voltage secondary; then hook the two
    primary windings in series, and see whether the voltage sums to close to 0
    or to a lot of volts. "In phase" is when they don't cancel.

    Generally they're a balanced T filter with inductors and capacitors. The
    purpose is mainly to block RF generated in your device from traveling down
    the power line (where it would then interfere with other devices). Use one
    rated appropriately for the current you expect your primaries to draw.

    Hey, if you've got the DPDT switch, you may as well switch both sides.
    Can't hurt, and adds protection in the case where you plug into an outlet
    that has been improperly wired.

    right?

    Yes. And like the primaries, they have to be in phase, or else the voltages
    will cancel. (And again, you can figure out which way is "in phase" by
    putting them in series and seeing if they cancel or reinforce.)

    I don't remember what you said your voltages were at the secondary (I'm
    lazy, sorry). Remember that you need to withstand twice the peak voltage,
    because the capacitor is charged one way as the secondary voltage swings the
    other way; and you should expect the possibility of a 10% mains overvoltage
    condition. A 100V bridge will work for a 30V RMS winding, by that logic.

    That's probably a radical overestimate of what you need. Remember, you
    don't have to get zero ripple from the filter cap - the regulator will take
    care of that. The bridge means you're refreshing the voltage at 120Hz, so
    you've got 8msec worth of time to cover. dV/dT = I/C, which means that if
    you're drawing 1A from a 10000uF cap, the voltage will drop 0.1V/msec, or
    about 0.8v ripple, give or take.

    right?

    I wouldn't worry about it, myself. Remember, the transformer doesn't put
    out infinite current, and the cap has some finite series resistance.
    Sometimes you do see special devices that have negative temperature
    coefficient in series with the primary; the idea is that it starts out cold,
    with high resistance, thus limiting the current; then it heats up, letting
    more current through. But I've seen a lot of high-power supplies without
    any provision for this and other than needing to use slo-blow fuses there
    doesn't seem to be a problem.


    Very common. But remember that if you feed 30V into a 5V regulator, it has
    to dissipate 25V times however much current. If your 5V supply is 1.5A,
    that's 38W of power, which means a big-ass heat sink. You might be better
    off using a switching DC-DC converter, or a separate low-voltage
    transformer.

    The tab on the voltage regulator is usually NOT ground. Check the
    datasheet. The bridges and the SSRs are probably fine.

    Generally speaking unless you're lucky, you need electrical isolation.

    Don't bother with the relay. Just put a 1K or 10K or whatever resistor in
    parallel with the caps. It won't dissipate enough power to worry about (P =
    E^2/R = 0.2W for 30V into a 4.7k); and it'll discharge the caps quickly
    enough (T ~= RC = about a minute to drain a 10000uF cap through 4.7k, from
    30v down to 10v). There's no reason why they have to be instantly
    discharged, right?

    Hope that helps,
    -walter
     
  3. John T wrote:
    (snip)
    I agree.
    That is probably capacitive current driven by the Ac voltage and
    coupled through the winding inter capacitance. If you shunt across
    your volt meter with a 10k resistor or a .1 uf capacitor, this voltage
    should drop to insignificance, if I am right.
    Understood. Winding resistance is too low to distinguish from zero,
    and leakage is too high to distinguish from infinity.

    You would need to pass a controlled DC current (like you get from a
    lab supply operating in current regulation mode) through all windings
    hooked in series, and measure the DC millivolts across each winding to
    get a comparative resistance measurement that had any accuracy. If
    the current were 1 amp, every millivolt would represent 1 milliohm of
    resistance.
    I would be reluctant to hook them in parallel unless their voltages
    matched better than they do. For every volt difference they will
    circulate a current equal to 1/ their ohms of resistance. That can
    get pretty warm. I think they were intended to be used in series with
    a 220 volt source. If you could get the thing out of the case and
    wrap a few more turns through the hole, you could add those turns to
    the lower voltage winding, and make them both match. Without that, I
    would just use the black white pair. and tape up the purple/orange
    pair.

    If you want to use any parallel combination windings, hook the black
    white to your low voltage source and test whether any guess shorts out
    the transformer, before connecting the combinations to the power line.
    They have series inductors to block high frequencies, and shunt
    capacitors to load them.
    The higher the current rating, the smaller the internal inductances
    (usually). So the lower current rated unit might provide more
    filtering. I would use the one that had a little higher current
    rating than you estimate your transformer primary to have.

    (snip)
    Yes, if their open circuit voltages are well matched. Using separate
    bridges will allow slightly mismatched windings to get along better.
    (the lower voltage winding will only contribute when the current sags
    the higher voltage winding.
    10,000 microfarads per amp of output DC is a common rule of thumb for
    low voltage (5 volt, say) supplies. 30 volt windings might get by
    with half of this but up to this amount will probably not cause
    problems as long as you use a slow blow fuse on the primary. This is
    a good idea, to handle the inrush if the transformer core saturates
    due to having the first half cycle polarity in the same direction as
    the previous last half cycle. Switching upstream of the line filter
    will allow the line filter caps to ring against the transformer
    inductance, and degauss the core to some extent on power down,
    reducing this problem.

    (snip)
    The peak voltage from a 30 volt winding is about 1.414*30=42 volts.
    U crude switching buck 30 volt buck regulator is probably what is
    practical at high current. It can be about 85 to 90% efficient.

    (snip)
     
  4. Mac

    Mac Guest

    Try measuring impedance, instead. Do this with no power connected to the
    device. If the resistance is less than a few Ohms, then they are
    connected. If the resistance is more than a megohm, then they are not
    connected. Anything in between is weird.
    Oh. Not sure what to say. Buy a meter you can trust? ;-)
    Oh, it matters which ones you match! If you do it wrong, the magnetic
    flux induced by the two windings will cancel, and the transformer will
    look like a short circuit (well, there's still the resistance in the
    wires, but you get the idea) and you'll blow a fuse or start a fire for
    sure. When I had no equipment to test this on a transformer, I just
    trusted the fuse and plugged it in. The fuse blew immediately, so I
    switched the polarity and it worked OK. I don't recommend this practice,
    BTW.

    Also, as another poster mentioned, your primaries don't match very well.
    so it probably isn't a good idea to parallel them. Stick with white / black.

    Theoretically it is always safe to switch only the hot wire, if you are
    using a polarized plug. If the plug isn't polarized, then you have to make
    sure you insulate both conductors from any user accessible areas. I
    wouldn't do this, personally.

    If you have a well-grounded metal case, and a three-prong inlet, I would
    say it is definitely safe to switch only the hot wire.
    I think so. If you have
    Those are big caps. I'm not too sure about the voltage rating. Your 30
    volts is RMS, so the peak DC voltage with no load would probably be around
    30 x 1.4 which is 42 volts. I was always taught to use a x2 safety margin
    on capacitors.
    36 Ohms is way to big. At 1 amp it would drop 36 volts. You may have to
    work this out yourself. I've used vitreous wire wound resistors for this
    application, because they are good at handling short power surges well in
    excess of their steady dissipation.
    The thing is, if you need one amp at 5 volts, and you regulate down from
    30 Volts, you're going to be dissipating 25 watts in your regulator. That's
    a lot. Consider a switcher?
    As long as you make sure you don't create an undesired short circuit I
    don't see why you can't mount everything on one heat sink.
    If you can run unregulated, that is the way to go. It might be nice to
    have a seperate bridge and filter caps for the stepper motors, now that I
    think about it. They will probably put out a lot of noise.

    I hope that doesn't screw everything up. I don't have any experience
    designing with stepper motors, but I know that most DC motors are pretty noisy.
    I would just put large bleed resistors across the caps. Shoot for an RC
    time constant of 5 seconds or so. Make sure to calculate the power for the
    resistors. If it seems too high, your caps are probably too big.
    HTH,

    Mac
    --
     
  5. Yeah, that one's pretty weird. I wonder if the purple/orange was intended
    for a japanese (100V mains) application, or some such weirdness. Although I
    don't know why they wouldn't just use a tap for that. It seems really
    strange to have two high-V windings with that ratio - all the transformers I
    know about have two identical primaries, designed to be used in parallel for
    120V mains or series for 240V mains. But what do I know.

    Any possibility that this xfmr is just plain defective, as in
    mis-manufactured and then surplussed out??
     
  6. John T

    John T Guest

    First, thank you to all for your help. I'm sorry I am slow to respond
    but at the moment I can only do testing between the time my kids go to
    sleep and the time I get too tired to be playing around with AC power.

    My original testing was really just taped together because I had hoped
    this would be pretty standard stuff.



    Nope. I have two of them, both pulled at the time from working equipment
    and they are reacting the same.



    Worked! But that doesn't mean it still won't be weird...
    With my wall wart plugged in on the blue/brown secondary (6.75v) I get: (aprx)
    Purple/Orange 23.60v
    Purple/White 27.21v
    Purple/Black 0v

    White/Black 27.16v
    Black/Orange 23.80v
    White/Orange 3.5v

    (this is what I'm thinking)
    /-------White
    { (27v - which scales up to about 118v)
    {
    +-------Black
    +-------Purple
    { (23v - which scales up to about 102v)
    {
    \-------Orange

    Doesn't look like they are in phase :-( In fact it seems that the coils are
    conneted and not being equal, they cancel each other out to small volts (3.5v).
    If I short white and orange then everything is down to about 0v.


    Yes, point taken. I will look into the DC-DC converters but I remember
    them being very expensive. I may go with another transformer since
    my 5v needs are pretty low.

    Is there any point to using the extra primary (if I could)? Ordinarilly
    it would just allow me to put out more current right?



    and

    (input also appreciated)

    -John
     

  7. For clarity, let's rearrange your readings into a half matrix, like this:

    BLK PUR ORN
    WHT 27.2v 27.2v 3.5v
    BLK - 0 23.8v
    PUR - - 23.6v

    Your surmise is that black is connected to purple internally, and that
    orange and white are at the same phase. Seems reasonable. If you take the
    wall wart and connect it from black to orange, you should measure around
    (but not quite) zero volts from white to orange, but around 7v from white to
    purple.

    My best guess is that one winding is for Japanese 100v market, one for US
    120v market. But it is weird that they would tie black and purple
    internally and yet bring both wires out; and also weird that they would
    bother with a whole separate winding. I would have expected a single
    winding, with a tap for the 100v. Or two entirely separate, isolated,
    windings, if one of them were actually a high-V secondary.

    Is there any possibility that you've confused the secondaries and primaries,
    and what we actually have here is two high-voltage secondaries? It would be
    somewhere around 500V, so I'm not sure what kind of equipment that would be
    for. And I can't imagine why they'd be the same phase, that just mystifies
    me.

    I see DC-DC converters with 5V output, in the 1 to 2A range, 18-36V input,
    going for $17-$50 in the Mouser catalog. The lower end of that range isn't
    terribly much more than a transformer, bridge, and some caps are going to
    cost.

    Right. But it sounds like it's not going to help you, here.

    -walter
     
  8. John T

    John T Guest

    yes, I get around 7v from white to purple and actually about 1v from white
    to orange.

    Well, I've got one weird transformer.

    So perhaps there are a few turns between purple and black.
    That might give:
    orange to purple 100v
    white to black 120v
    orange to white 220v

    or something like that.


    On the secondary side I'm also a bit confused now.
    The yellow pair seems separate. With 120v in on black and white I get
    Blue Brown Red 1 Red 2
    Blue - 30v 30v 0v
    Brown 30v - 0v 30v
    Red 1 30v 0v - 30v
    Red 2 0v 30v 30v -

    So maybe:
    /--- Blue
    { (30v)
    +--- Brown
    +--- Red 1
    { (30v)
    \--- Red 2

    With the two coils not being in phase and adding up to 0v.
    If I short Brown and Red 1 I get the same thing.
    I didn't try shorting Blue to Red 2.

    If I connect them through 2 bridges won't the phase not be a
    problem anymore? Or would one be -30v to the other one's ground?


    One a positive note, I found some DC-DC converters. A couple which
    are 36-72v input (from telco equipment) and one that is 18v - 36v.

    So I can probably use my transformer at this point with the black/white
    as the primary. The yellow/yellow for my 24v (with a 7824) and 5v with
    my DC/DC converter. I have bridges and my capacitor needs are lower
    than I expected. I can use Red/Red for my 30v motor power or the
    blue/brown (possibly both - depending on the answer to the above
    question).

    At the very least I have a better understanding of what I need if
    should I want to just go out and buy a transformer which matches
    more what I'm doing.
     
  9. No, because putting a voltage from orange to white is just going to cancel
    itself out. Nice idea, though.

    Are you still measuring with the 10k resistor in parallel with the meter?
    I'm wondering if some of what you're seeing is being confused by leakages.

    If that's not the issue, then again, this seems weird; I've never seen a
    center-tapped xfrm secondary with the two sides wound in antiphase, and
    that's what you're describing (with the caveat that the "center tap" is
    brought out on two wires, itself very very strange).

    I'd much rather believe this was two separate 30v windings. If you measure
    the resistance from brown to red 1, is it high or low?
     
  10. John T

    John T Guest

    Still measuring with the 10K resistor. Between all points I get about 0.3 ohms
    but my meter then suddenly jumps down to 0 sometimes. I think that is just the
    transformer soaking up the power from the meter. Sometimes I get a different
    value like 0.6 but whatever the number it is consistent between any of those
    four wires and it's infinite between any of those four and either yellow wire.
    The yellow wires have slightly higher resistance.

    Maybe I'll open up one of the two of these that I have. I can try using it
    like I have it now or I can say it's useless until I figure it out, in which
    case I either throw it out or open one up.
     
  11. Well, you're at the limit of my limited expertise - I've never seen a
    transformer like that :-(

    By the way, if you've got a current source (e.g., a bench supply that will
    work in constant-current mode) you can put a fixed current of, say, 1A
    through the winding and measure the DC voltage across it, 1mV/milliohm.
    Just one way to get a low-resistance measurement.
     
  12. John T

    John T Guest


    Thank you very much for the help you have given me. I can work with
    my transformer at this point and I have learned some stuff so you've
    also provided me with tools for next time. I tried to see if I could
    take one of the two that I have apart and it appears exposied well into
    the can. Oh well.

    Thanks again - John
     
  13. No, the transformer can't do that. The effect you are getting indicates
    a problem of some sort with your meter (maybe an intermittent fault).
    This, too, indicates a problem with the meter or its connecting leads.
     
  14. Ben Bradley

    Ben Bradley Guest

    What type of meter is this? What's the resolution? Most DMM's have
    a resolution of 0.1 ohms.
    It's not that easy to measure such low resistances with the average
    DMM. You must make very good connections. Put the leads across short
    bare copper wire (both leads touching the wire, but not each other).
    You may get a measorement of perhaps 0.1 to 0.3 ohms. This is your
    "zero" value and should be subtracted from any low-resistance
    measurement you make.
     
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

-