Connect with us

Bench PSU - buy or build?

Discussion in 'Electronic Basics' started by James Harris, Oct 26, 2003.

Scroll to continue with content
  1. James Harris

    James Harris Guest

    Hi. Am looking for a bench power supply that has positive and
    negative supplies. Ideal would be a device that supplies +12,
    +5, -5, -12 and a variable 0-30V DC, preferably with current

    I have been thinking about making such from a computer power
    supply (to get the +/-12 and +/-5) and a torroidal with 2x23V
    RMS to get circa 30V with full-wave rectification. I could then
    use a semiconductor regulator (eg. L200x on the positive and
    maybe something on the negative side also) to adjust the output.

    Frankly the above is a bit daunting. I am OK with the safety but
    unsure of the details of the electronics. It seems to be
    expensive to get any kind of panel meter, let alone the two or
    four required (V and A on both variable outputs). And a decent
    case will be costly also.

    Would I be better buying? I have seen this one,, which
    is the Skytronic unit here, It has
    one +5V and two variable 0-30V for UKP200. It has -, 0, and +
    outputs so does that means it will supply up to (-15 to +15) on
    each output? I've read the manual on the web site which is,
    frankly, dismal. This is a little offputting but other devices
    with similar specs are at least twice the price.....

    Should I buy or build? Any guidance much appreciated.
    - James
  2. Watch EBAY for power supplies. You will occasionally find an old bench
    supply for a reasonable price.

    Building linear power supplies is pretty easy, at least the fixed voltage
    ones. Just get some transformers of the right size, and use the appropriate
    78xx regulator (with bridge, caps, etc.)

    The variable supply is also fairly easy using something like an LM317.
    However, you waste lots of power (and thus generate lots of heat) when using
    something like this at 3VDC and 2A.

    EBAY occasionally has variacs on sale, which is a good solution to the
    waste/heat issue; I've seen them for as low as $16 US (which I lost to a
    sniper, drat!)

    One thing is that unless you are a craftsman, your enclosure won't look as
    nice. Also, the displays, current limiting, control knobs, etc that you get
    with a typical bench supply are features you'll eventually want.

    Bob Monsen
  3. Chris Oates

    Chris Oates Guest

    Years ago I built all my own equipment
    it looked like crap & barely functioned
    but it was so much fun.
    You won't learn anything buying
    'off the shelf'
  4. dg

    dg Guest

    If you want to build it yourself, go ahead and do it, but be warned that you
    will probably be better off in the long run with buying a pre-built supply.
    I built a variable dual 3 amp supply (LM317 iirc), with an additional 5V
    supply built in, in a rack mount chassis. I put in a digital panel meter
    and a dial to read V or A on any of the 3 supplies. It is big, heavy, and
    was not cheap to built. The panel meter is non operational now because you
    need a separate supply to power it (I may buy a DC-DC converter to power the
    meter, that will isolate the supply voltage). I recently bought an RC car
    and all the associated gear. The battery charger I bought came with an
    AC/DC switching power supply much like the typical "wall wart" only this one
    is switching, has a fan, and puts out 12V at 7A! It is VERY lightweight, I
    almost can't believe it. The supply and smart charger were only ~$60. This
    is not what you are looking for, but just an example of what power supply
    technology has become. My advice, buy a nice modern pre-built supply.

  5. James Harris

    James Harris Guest

  6. James Harris

    James Harris Guest

    Variacs sound good to cut down on power and heat. I would buy
    from a retailer but the variacs I can see seem designed to work
    up to mains voltage. I guess they could just be used at the
    lower end of the range. I'll give it some thought. I can see
    myself ending up making more than one PSU.

    I wonder why variacs don't isolate from the mains. I can't think
    of a reason the secondaries need to be linked to the primary.

    Yes, am very very far from a craftsman.
  7. A computer psu might be a good choice, it has most protections already
    built-in. But it does not provide voltage regulation facilities. I'll
    describe an idea how to construct an (a little advanced) linear psu, if you
    have further questions about linear regulators, please ask.

    The choice whether to build or buy a psu is yours, but be warned that
    building one may take more time and be more expensive than one may want.
    However you won't learn anything (except hating the prices) by buying one
    while constructing your own psu is likely to be a lot more interesting. If
    you have some spare time left, my advice would be: Try it. If you have old
    components salvaged from some no-more-use equipment, constructing a psu
    might even be cheap (I built mine entirely from this sort of things, it's
    working reliably for 3 years now, params: 1 to 30V, 8A). The enclosure will
    be one of the hardest parts to find, so a poor man's solution may be a
    speaker box with transformer and heatsink on top unless you have the means
    and skills to create a more suitable one. In your post you only gave the
    voltages, no maximal current values, so I can hardly assume anything about
    the power throughput you need. So for the simple reason of not disappointing
    you with a 'weak' psu let's assume 7A approx current at 30V. There are 3
    classical approaches to such a thing: 1: Linear, 2: Switching, 3:
    Triac-controlled transformer with linear fine-regulation electronics. The
    first solution is the easiest, but the most inefficient. The second one is
    the most complicated and exceptionally hard to build at home (do not try
    unless you are really experienced), but the most efficient. And finally the
    third one might be a viable medium-complicated solution with acceptable
    efficiency. The concept of this device is to roughly pre-control the voltage
    on the primary of a transformer by using a triac (rated at two times the
    maximal transformer power), then transform the voltage down, rectify it
    full-bridge, filter it, and finally use a linear regulator to fine-tune the
    voltage, limit the current (if necessary), implement a short circuit
    protection and this like. The disadvantage of this design is that there is a
    need to operate 3 (if you are skilled, maybe 2) transformers independently
    of each other while only one of them delivers useful power. Since I have
    only built similar, but not exactly the same sort of power supplies I cannot
    give you a very exact description, nevertheless I'll try to outline a likely
    possible way of handling this. Everyone who is more experienced is
    encouraged to correct me. First, the easiest-to-test structure of any device
    is a modular one, so I'll try to make use of it. The psu shall thus consist
    of 3 parts (apart from the xformers): 1: the triac circuit, 2: the linear
    regulation circuit and 3: the control circuit. First the triac circuit. Note
    that it is not isolated and operates at mains voltage. To control the
    primary of a transformer, a triac is placed in series with it. To prevent
    the triac from tripping accidentally, a resistor (around 500k) should be
    connected to the cathode and control pins of it. The voltage to 'fire' the
    triac can be taken off a capacitor that is charged from the mains through a
    resistor (resistor value depending on the capacitor used, the more
    capacitance, the less resistance, but not less than 30k, you'll need to do
    some testing). To make the thing work more reliable, 2 zeners in series, but
    with opposite directions and a voltage of about 6 V can be used in series
    with the triac gate. Controlling is done by shunting the cap with a
    transistor (bridge rectifier needed), this will delay the time until the
    triac triggers, if the transistor is open completely, the triac will shut
    down. The transistor base voltage / current is provided by an additional
    power source (very small trafo, maybe half a watt) and controlled via an
    optocoupler (Isolation required!).
    Schematics in ASCII:

    from | R__ to trafo
    mains '--|__|-.
    C |
    from .--||---x--------.
    mains | | to trafo
    | / Triac |
    | R__ | |
    | 2 Zeners|
    | .-------------'
    | |
    |rectifier| .---. R__
    --------- | | .------|__|---- -
    -| |+ | C \ B | R__ opto
    | '-----' |---x-|__|-. .---- +
    | E</NPN | C | |
    | | '-||---x '---- +
    | | | 6V DC
    '------------x------------x------- -

    When the LED part of the optocoupler is on, the transistor will be on, it
    will shunt the triac-controlling cap and the triac will stay off, the less
    current will pass through the opto LED, the more power will reach the main

    The next part would be a linear regulator. with its control circuitry. I'll
    leave it up to you to choose one appropriate for your needs and draw it as a
    simple square in the schematics. Note that if you use a custom regulation
    system (a couple of 2N3055 with appropriate control or this like), an
    additional transformer will be needed to power the control circuit.

    | regu |
    ----------|-----|-----x---x--| lator |--x---x-----
    from | rec | + C_|_ | | (lin) | | _|_ +
    trafo | tif | ___ | '-------' | ___ OUT
    secondary | ier | - | | | | | -
    | |
    | R__ |Zener 4V
    '-|__|-. '-|>|-----
    1K | -
    | To opto LED
    | +

    Depending on the voltage difference between the connections before and after
    the regulator the triac circuit will be auto-adjusted using the optocoupler.
    When the voltage rises, the LED will light stronger and the optocoupler will
    lower the triac throughput thus limiting the voltage.

    That's all. The whole thing can be repeated with 2 transformers if a
    positive and a negative voltage are needed simultaneously. It is also
    possible to use one power transformer and connect the 2 optocoupler sensors
    in series to control one triac. This may be less efficient though.

    Please note that this design is certainly a waste of parts and time if less
    than some 5A at 30V output are needed. For a less powerful psu just use
    linear regulation with no triacs. Don't forget a BIG heatsink in either

    More questions -> please ask here,
    you can also email me (address valid).

  8. James Harris

    James Harris Guest

    Agreed. This is the one I saw about using an old computer PSU, What do you
    think about the use of just 2Ohms between +5 and ground to
    preload the PSU. This seems an incredibly small value to me,
    passing a large 2.5A just to get the PSU working.

    Another point. Would it be naiive to add zeners on all except
    the +5 to ensure the outputs don't exceed the rated voltages by
    too much?
    - James
  9. Joel Kolstad

    Joel Kolstad Guest

    Just economics -- you don't need two windings if your VARIAC isn't isolated!
    You certainly can buy isolated VARIACs, though -- I own one, and I believe
    they're still readily available. Other than cost, I can't think of a good
    reason you wouldn't buy one! :)

    ---Joel Kolstad
  10. Try a higher resistance. A lot of PC PSUs will work with a small load or
    sometimes even no load at all. Don't waste 2.5A
    Usually the don't exceed the specs much anyway. If they do, use some other
    protection. Do not add zeners, a current of up to 10A (+ 12V line) is
    likely to kill them. Instead try to find out where the 5V loopback controls
    the PSU power and add there a 12V-controlled protection circuit.

  11. Use a variac that goes up to mains voltage w/o isolation. Connect a stepdown
    transformer to it. This will provide isolation and a better efficiency for
    low voltages and high currents. Note that it would be less efficient than a
    variac only for high voltages and low currents.

    As to why a lot of variacs don't isolate: They have no secondary, the (only
    existing) primary is used in auto-transformer mode. There are exceptions
    with (most likely isolated) secondaries though.

    See also my reply about triac control. You may want to substitute the triac
    with a variac and adjust it manually instead of by optocoupler. In this
    case a voltage meter in the place of the opto LED will show whether variac
    adjustments are necessary. It does not have to be exact, just use the
    cheapest one available.

  12. Sorry, there was a mistake in my post. Not a real one, but it could end up
    with a less efficient device. Use a 2V Zener instead of a 4V one.

    | R__ |Zener 2V
    '-|__|-. '-|>|-----
    1K | -
    | To opto LED
    | +

    Part of original message:
  13. The thing about modern supplies is that they are usually switching supplies.
    That is how they get away with such small transformers (they use higher
    frequency transformers, which require much less metal.)

    One problem with these, however, is that they are really noisy. If you are
    doing work with an oscilloscope, you can end up with lots of noise in the
    signal of the device under test from the supply. I was using a switch mode
    PSU from an old cable modem for a bit, and it was great in terms of
    generating lots of different output voltages (35V, 12V, 5V, -5V, -12V, all
    at different amperages.) However, I couldn't get away from the 250k 10mV
    signal it was adding into everything. It didn't even have to be connected,
    it was radiating the signal so intensely that my probes would pick it up.

    Switching <g> to a linear power supply did the trick. Much cleaner signals.

    Bob Monsen
  14. James Harris

    James Harris Guest


    Thanks for all of the info. I get at least some of the
    principles but I think this is a bit above me for now. I'll keep
    it, though, for future reference - maybe a second project! I do
    have a question, though, about displaying the voltage and
    current. It would be essential to avoid dotting my meter between
    circuit and PSU. Can you recommend a cost-effective but
    reasonably accurate solution? If possible I would like to
    display V and A at the same time.

  15. Sorry, I don't have much experience with meters. I actually have only one
    and using it for all sorts of testing (V, A, Ohm, hfe, F, H, Hz, diode test,
    duty cycle, etc) I find myself disconnecting and reconnecting it in another
    mode over and over :). It was not an especially cheap one either.

    I assume, the more experienced users here will help you out.
  16. I think it's more economical to buy than to build. I bought four HP
    66312A 0-20V, 2A power supplies for $200 each, I could never come
    close to getting the capabilities on anything I build. Here in the
    U.S., inexpensive PSes are available, for an example see URL

    @@[email protected]@[email protected]@@[email protected]@[email protected]@[email protected]@@[email protected]@[email protected]@[email protected]@,@@[email protected]@[email protected],@@[email protected]@[email protected]@[email protected]@
    ###Got a Question about ELECTRONICS? Check HERE First:###
    My email address is whitelisted. *All* email sent to it
    goes directly to the trash unless you add NOSPAM in the
    Subject: line with other stuff. alondra101 <at>
    Don't be ripped off by the big book dealers. Go to the URL
    that will give you a choice and save you money(up to half). You'll be glad you did!
    Just when you thought you had all this figured out, the gov't
    changed it:
    @@[email protected]@[email protected]@[email protected]@[email protected]@[email protected]@[email protected]@@[email protected]@[email protected]@@[email protected]@[email protected]@[email protected]@[email protected]@[email protected]@[email protected]@@
  17. There's not a whole lot to learn when you assemble a PS made out of a
    Xfmr, a FW bridge, a filter cap, a LM317, two resistors and a meter.
    Just about all the design is done already.

    @@[email protected]@[email protected]@@[email protected]@[email protected]@[email protected]@@[email protected]@[email protected]@[email protected]@,@@[email protected]@[email protected],@@[email protected]@[email protected]@[email protected]@
    ###Got a Question about ELECTRONICS? Check HERE First:###
    My email address is whitelisted. *All* email sent to it
    goes directly to the trash unless you add NOSPAM in the
    Subject: line with other stuff. alondra101 <at>
    Don't be ripped off by the big book dealers. Go to the URL
    that will give you a choice and save you money(up to half). You'll be glad you did!
    Just when you thought you had all this figured out, the gov't
    changed it:
    @@[email protected]@[email protected]@[email protected]@[email protected]@[email protected]@[email protected]@@[email protected]@[email protected]@@[email protected]@[email protected]@[email protected]@[email protected]@[email protected]@[email protected]@@
  18. Another serious disadvantage of a PS with a TRIAC or switch mode PS is
    that they make a lot of EMI and RFI, so if you are using these kinds
    of PS to power a low level amplifier, you will probably see some
    interference in the circuit. Linear PSes don't have this kind of
    problem. But Dimitrij has some good ideas if you don't plan on using
    the PS for low level amplifiers.

    @@[email protected]@[email protected]@@[email protected]@[email protected]@[email protected]@@[email protected]@[email protected]@[email protected]@,@@[email protected]@[email protected],@@[email protected]@[email protected]@[email protected]@
    ###Got a Question about ELECTRONICS? Check HERE First:###
    My email address is whitelisted. *All* email sent to it
    goes directly to the trash unless you add NOSPAM in the
    Subject: line with other stuff. alondra101 <at>
    Don't be ripped off by the big book dealers. Go to the URL
    that will give you a choice and save you money(up to half). You'll be glad you did!
    Just when you thought you had all this figured out, the gov't
    changed it:
    @@[email protected]@[email protected]@[email protected]@[email protected]@[email protected]@[email protected]@@[email protected]@[email protected]@@[email protected]@[email protected]@[email protected]@[email protected]@[email protected]@[email protected]@@
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