need advice for diy bench power supply

Discussion in 'The Projects Forum' started by Salaja, Jan 27, 2013.

  1. Salaja

    Thread Starter New Member

    Jan 27, 2013
    greetings, this is my first post on the website.

    in the last few weeks i've begun setting up a small electronics kit (breadboard, multimeter, etc), and for my first project, i am trying to make a bench power supply.

    i've attached a 'rough' schematic to this post. i say 'rough' because i haven't included any of the feedback circuitry for the voltage regulators, or circuitry for the volt and amp meters.

    i just want to make sure that there aren't any gleaming errors/design flaws that i haven't noticed yet.

    the goal of this power supply is to generate a floating voltage supply (like most bench supplies), but also a dual tracking +ve and -ve supply for running dual supply op-amps. i want up to 20v, and +20, -20 on the dual line. ideally, i would also like to draw 1-2 amps from the lines.

    i decided to cascade a buck converter with a linear supply so that the buck converter could do the bulk of the voltage control, with the linear supply being used to remove the ripple. i didn't want to use a linear by itself because i don't want to deal with massive heat sinks.

    the volt and amp meters will be controlled by 2 pic micro-controllers, 1 for each floating voltage.

    my main questions so far are:

    1. is the linear regulator necessary? will a few more caps on the output of the buck reduce the ripple to the same extent? is it even worth the trouble? (i'm not doing audio work, just general electronics, micro-controllers, op-amps, etc.)

    2. how important is power factor and passive filters between mains and the transformer? i have an old atx power supply that i could pull the filter out of. is it worth it, and would it even work with a lower current load than it was designed for?

    3. for the +ve and -ve lines, is it okay to use a full wave diode bridge with one of the ac input lines floating like in the circuit diagram? it should 'theoretically' still be the same as 2 diodes, each half wave rectifying the AC into the +ve and -ve DC lines. right? right!?

    4. how sensitive are op-amps to changes in their supply voltages? for example, say i'm using an op-amp to measure the millivolt difference accross a current sense resistor, if it's supply voltages are unregulated, will it still be reasonably accurate? i guess the solution to this would be to create regulated 18v and -18v lines, or something like that. :/

    5. anything i've missed, opinions and suggestions are all welcome. :)
  2. GopherT

    AAC Fanatic!

    Nov 23, 2012
    What are your MOSFETS doing? What is controlling them / what are they controlling.

    What are your lm317 and lm337 getting controlled by (pots?). You need a resistor network to make the voltage variable (output to adj pin then back to ground).

    Your bridge rectifier on the bottom circuit is missing a connection. Are you sure you want that configuration? It looks like you will be heading to a design with 60 hz rectified DC intothe filter caps. You might want to use a center tapped transformer for the bottom circuit to get 120 hz rectified DC into your filter caps.

    Check your MOSFETS I think you swapped your n-type and p-type (one you figure out we're they really go - not doing much where they are but I don't see what is controlling the gate on any of them or why the are there (schematic is incomplete).

    Finally, the 1.5mF cap on the far side of the voltage regulators might be excessive. Read the Datasheets. You are likely better off investing more in caps for the unregulated (left) side of the regulators
    Last edited: Jan 27, 2013
  3. MrChips


    Oct 2, 2009
    A bit overkill for a hobbyist, I would say.

    For experimenting, here is what you need to start off.
    Get one 9VDC 500mA wallwart. Add a 7805 regulator and this is good for a lot of experimentation.
    In many cases, a simple 9V battery will suffice.
    You can also find 5VDC adapters, regulated, all set to go.

    If you need a +/- 12VDC supply, ok, build a simple one as above. But most time the goal is to use single supply op amps.

    After you have done the above you can then venture into building a variable 30VDC 5A supply or similar. You will find out it is a lot cheaper to go buy one. Save your talents for more interesting projects.
  4. Salaja

    Thread Starter New Member

    Jan 27, 2013

    mosfets are the switches for the buck converters, they will be controlled via a 33063 with a pull down resistor (to ground from positive lines) for the P type, and a pull up resistor (also to ground, but from negative) for the N type.

    the lm317 and lm337 will have a resistor between the output and adjust pins, and their voltage will be controlled by a voltage applied to the adjust pin via opamps and resistors. they will be indirectly controlled by potentiometers, but not directly. this is so i can configure them so that their output will be a few volts lower than the bucks output.
    -i didn't include the control and feedback circuitry because it would be complicated, and at this stage i am more concerned with the general layout of the supply.

    yes, the missing connection on one of the bridge rectifiers was intentional. i even asked a question about it in my first post. the reason i did it this way was so that i could use a single transformer, and not have to buy multiple expensive transformers (to my knowledge, there are no dual output tranformers, where 1 output is 18v, and the other is 36v with a center tap).

    i could ahve just used 2 diodes instead of the bridge rectifier with 1 floating input, but i've got them laying around, and i feel like i should use them.

    im going to use p-type mosfets as high side switches, as they need a lower voltage to drive them, and i can use the low artificial ground. i am using n-type for the low side switch (the negative line in this case) because they like a higher voltage to drive them, so i can use the artificial ground again (which will be way above the negative line).
    -am i using these right?

    both the regulators will have feedback, so i'm not very concerned with noise on the input line. as long as the voltage doesn't fall when i draw current, it should be fine.


    i have been using an 8x AA (~12v) battery pack, and an old atx psu to give me 12v for circuits. i have been playing around with buck converters, and have got circuits that can provide several amps (didnt go higher because i dont want my bread board to melt).

    i am currently a final year engineering student, so i have some experience with this sort of stuff, but i am by no means an expert, which is why i asked here.
  5. Wendy


    Mar 24, 2008
  6. bountyhunter

    Well-Known Member

    Sep 7, 2009
    Looks like a buck switcher followed by a linear. Ambitious, but more efficient than a straight linear.
  7. k7elp60

    Senior Member

    Nov 4, 2008
    I see no problem as long as you use some kind of voltage control that adjusts both the switching and the linear together. Like have the switcher output about 0.5 volts + minimum input/output of the linear regulators. The reason I say this is that my experince with 3 terminal linear regulators has been that they are not suitable for a large range in input/output voltage at currents near the 1 amp range, because of the internal circuity will shut them down because of sensed power dissipation issue. I have tried LM338's, LT1083's just to name the ones I remembered.
  8. Salaja

    Thread Starter New Member

    Jan 27, 2013
    Bill_Marsden: i tried to make an effort to read the rules and search the forum. i had seen all of what you linked (except 1), and i felt that it didn't really touch on what i needed to know. the exception was that i seem to have missed the last link to the "12v to 5v DC high efficiency SMPS buck converter using 34063 IC", which was very useful and pretty much answers by transistor questions. for that, i am thankful.

    bountyhunter: nail on the head.

    k7elp60: i am using the lm317 and lm337, and i think they like to have 2-3v of overhead on their input. i plan on using op-amps to create a control voltage to set the linear output at a few volts lower than what the switching regulator is being told to output. this will reduce the overall efficiency of the circuit, but still be better (and cooler) than a pure linear regulator, while maintaining the low ripple output.

    thanks everyone for your replies. because i didn't get any comments saying stuff like "this will kill you if you do it", i think i will go ahead and begin ordering extra parts and testing. :)
  9. kubeek

    AAC Fanatic!

    Sep 20, 2005
    The primaries of transformers in series are nonsense, unless they are 120Vac transformers, and even then it is not really a good idea.
    The secondary connections, especially the bottom half are very weird. You should have it connected like this: that is both primaries in parallel, both secondaries in series. This will give you two raw DC power rails, +25, 0, -25V. From these two rails you should feed the second half of the circuit - the switchmode and linear regulators.

    If you want the top supply galvanically isolated from the bottom one, then the bottom transformer will need to have center-tapped secondary if you want to get split rail output.
  10. Salaja

    Thread Starter New Member

    Jan 27, 2013

    i don't think i've mentioned this, but im in australia, and our mains are 240v. the transformer im thinking of using has 2x 120v primary windings, and 2x 18v secondary windings, and a VA rating of 120VA.

    also, in the link you gave it showed the fuse after the rectifier and transformer. is this better practice? what happens if the transformer or rectifier has a short?

    but i don't think i've considered running multiple regulators off of the same voltage line to get multiple voltages. it allows me to use a single micro controller, although it comes at the cost of isolation between the supplies.

    i will think about it.
  11. kubeek

    AAC Fanatic!

    Sep 20, 2005
    Ok if it is a single transformer then it is ok to have the primaries in series, with two different transformers it would be tricky.
    That link was more to show the rectifier configuration, yes I would rather put the fuses before the rectifier, and a polyfuse would be better suited than a normal glass fuse + you should have one normal glass fuse at primary.

    I don't think I ever needed a separate isolated supply for anything. Also, since you are having a SMPS in there anyway, maybe you could use the lm317/337 for a regulated low-current low-noise supply for example for opamp supply fed straight from the main caps, and then have the switchers for larger current, but with noisier output, say for feeding an audio amplifier.
    Salaja likes this.
  12. takao21203

    Distinguished Member

    Apr 28, 2012
    Who really needs dual rail supplies these days?

    You don't use these for class D amplifiers or microcontrollers.
    Many modern OpAmps can work from single rail.

    At least save efforts on multiple rectifiers. This might be OK for 1 Amp or two, but will get messy for higher power levels.

    Use one buck converter + one booster if you need more than the transformer provides.

    The secondary linear regulator is largely unneccessary.
    Many circuits only need very small current which must be totally clean. Therefor they normally use onboard linear regulator (some 10mA at best).

    You can also get pretty good results with one additional coil + small capacitor.

    The MC34063 isn't so good for larger currents. It is old, and while it is good to use for upto 1 Amps, not many people use them for larger currents. Use a TL494 instead.

    It's upto you of course but my opinion about the design simply put is you can get much more for the value.

    Better build one single rail buck converter based on TL494, and drop the secondary linear regulator.

    Microcontrollers can work from quite wobbly supplies. Only for A/D you need clean voltage and a few other purposes. So you normally use a small 78L05 fed from a 12V auxillary supply.

    12V is used for instance by most cooling fans. And it is a good intermediary voltage- to convert down to 5V or 3.3V, or to step up to higher voltages.

    Rectifiers get quite hot if you exceed 1 Amps. You can use these large 6 Amps diodes for 1A current, but for larger current, you need a bridge with a hole + a heatsink.

    I made one PWM regulator based on TL494, fed with 60V, and I have run it with 70W 12V bulbs. I have also run it with nearly 200W LEDs at ~30V.

    It is using two large coils, and one special PNP transistor. In theory it should be able to handle more than 100V input voltage.

    If you use linear regulator after the switcher, how will you adjust both at once?

    Build what you like but here's my opinion. I have built many switcher circuits.
    Salaja likes this.
  13. Salaja

    Thread Starter New Member

    Jan 27, 2013
    yeah, sorry about my 'schematic'. i couldn't find a symbol for dual winding transformers, so i just used two in series.

    i feel like i don't know enough about re-settable fuses, so i wasn't sure if they were suitable for this application, so i just went for glass. i'll look into them.

    one issue i noticed with having the secondary windings in series is that it reduces the available current. lets say (to keep it simple) it can supply 1A to each line (+25 and -25, for 2A total), this means if i have op-amps drawing 100mA, then there is only 900mA left for the single supply. where as my configuration can supply 1A to the single supply, and 0.5A to each side of the dual supply(for a total of 2A).
  14. Salaja

    Thread Starter New Member

    Jan 27, 2013
    the dual rail is for dual supply op-amps. why do i want to use dual supply op-amps? because i'm used to them.

    i wasn't aware that most IC's had their own internal regulators. if this is the case, then the argument for the linear regulators pretty much vanishes.

    the full current isn't actually going to pass through the 33063, it controls a mosfet which will handle the current. i don't plan on drawing more than 1-2 amp anyway.

    the diode bridges i've got laying around are rated for 10A, and have a hole to mount a heat sink if needed.

    i was planning to adjust both of the regulators with op-amps, which would be controlled by a potentiometer. because they would take different 'adjust' voltages to control them, i was going to derive a mathematical equation to relate them, then use op-amp maths to perform the equation, less a few volt for linear overhead. although, after what you said before, i may not even include the linear reg.

    thank you for your input.
  15. #12


    Nov 30, 2010
    Here. This always keeps me from expecting the wrong amount of output from my rectifier designs.
    Salaja likes this.
  16. Salaja

    Thread Starter New Member

    Jan 27, 2013
    i'm not sure what "V (ave.) DC" is for. i assume it is the average voltage when you draw current equal to the transformers limit?

    otherwise "I D.C." comes to just over 2 amps for the single supply and about 1 amp for each line on the dual. V peak is about 25v as expected, and V ave, is about 16.2v which is still very useable.

    did i read it right?

    EDIT: never mind, i've figured it out.
    Last edited: Jan 28, 2013
  17. bountyhunter

    Well-Known Member

    Sep 7, 2009
    Anybody who works in electronics.

    I just had to repair a digital TV converter box and the main board runs off two voltage rails (5V and 3.3V) so I needed two supplies just to power it up.

    As always, it was the switching power supply that had failed and it was because of the electrolytic caps.
  18. bountyhunter

    Well-Known Member

    Sep 7, 2009
    Good reference, the FWB with cap filter bears memorizing since it is used in about 99% of generic Dc supplies. Note that since the DC current is about 0.6 X the transformer secondary current, it follows that the trasformer secondary current rating must be AT LEAST around 1.7 times the DC (output load) current for the design. Without question, this is the most common mistake made.
  19. THE_RB

    AAC Fanatic!

    Feb 11, 2008
    Huh? Transformers are usually rated at RMS amps, which is equivalent to the DC amps after rectification and filtering.

    It is assumed by transformer manufacturers who label the "amps" on the transformer that these days it will be used as a DC supply and the voltage and amps listed on the label reflect the final DC volts and DC amps.

    How old is that info?
  20. bountyhunter

    Well-Known Member

    Sep 7, 2009
    As I said, for the FWB with a capacitor:

    Transformers are indeed rated in AC (RMS) voltage and AC (RMS) current ratings on their secondary windings, but the relationship to the DC load current is certainly not 1:1 for a FWB.