Cheap split power supply

Discussion in 'General Electronics Chat' started by dude521, Jan 15, 2010.

  1. dude521

    Thread Starter Active Member

    Nov 1, 2008

    I trying to make a very cheap split power supply. The biggest probelm I have is that the transformers cost a lot of money, but I still need a +-15V source. I can get a 15V DC 2A transformer for pretty cheap, but I'm not sure if it can be used to create split voltages. I came accross the following circuit, but I'm not sure if it is a good design. Can anyone give me some feedback.

    Note: This will be used mainly to power op amps in digital circuits, so it should not need much current. The regulators are limited to 1A anyways.

    Last edited: Jan 15, 2010
  2. eblc1388

    AAC Fanatic!

    Nov 28, 2008
    It will work.

    Two comments:

    1. Even with a 12V transformer, the voltage across the input of the two voltage regulators reaches 28V with 100mA output loading and about 30V with no load. The voltage will be even higher if 15V transformer is used. The 25V capacitors is not rated for this. The rating of the two capacitors that directly connected to the transformer terminals is OK.

    2. The ripple caused by voltage doubler is difficult to get rid of, even with a voltage regulator. Be prepare to have some ripples on the +/-15V output.
  3. bertus


    Apr 5, 2008
  4. t06afre

    AAC Fanatic!

    May 11, 2009
    You could also make two +15 power supplies, and put them together. Or purchase two transformers and make a +/- supply this way. What do you prefer?
  5. Mike33

    AAC Fanatic!

    Feb 4, 2005
    +1 on Bertus' first link...easy to make lab supply there. Can't see any reason for much more than 1A for opamp experimentation!!?

    The LM317 Datasheet will have some good stuff on it, too...
  6. SgtWookie


    Jul 17, 2007
    You could also do something simple like the attached with a single 15v secondary.

    The ripple will be higher because each rail is 1/2 wave rectification, but if you add on regulators and with your low current draw that shouldn't be much of an issue.

    Note that Rwire1, Rwire2 and Rsim are only present to allow the simulation to run.
  7. dude521

    Thread Starter Active Member

    Nov 1, 2008
    @eblc1388, bertus: Ok, I didn't know what a voltage doubler was. I just read the wikipedia article. Now I at least know how the circuit works. A little bit of ripple is not that big a deal, but I do need more than 100mA.

    @t06afre: Costs too much, I want to keep the project less than $30. I can get some parts for free, but its the transformer cost that is really killing me.

    @SgtWookie: Thats a good idea for low current. After looking over my plans again, I think I would like to be able to draw 500mA from each rail, so I don't think this is a good idea.

    I was trying to get a wall transformer, but I can't find any cheap enough. I think I will get a power transformer and then build it into a box. What I want to do is make a +-15V, +-12V, 5V and variable voltage (~0 - ~20V) bench power supply. I want to be able to use any combination of rails with a total max current draw of 1A (measured at the transfomer I guess, since each regulator will be 78xx or 79xx it should be ok even if all the current is coming from 1 rail).

    What I will do is get this (digikey part, followed by datasheet):

    I have a few questions:

    1) I'm not sure how to read the data sheet. I want a single primary coil and a centre tapped secondary to give +- 20VAC. On the first page the voltage is given in VCT, which I assume is voltage at the centre tapped node, as shown on the second page. Does this mean that the AC voltage measured from the the CT node to any other node is 20VAC? So if I measured the AC voltage between the two nodes (not CT), then it would be twice 40VAC?

    2) If I load the transfomer unevenly, say I have a +12V 500mA draw, will this cause any major problems with the other voltages, say the +-15V?

    3) What is the best way to go from the +20V smoothed (from capacitor) to 5V. Should I take the 12V output and connect it to the 5V input or should I go straight to the 5V?

    I'll post a schematic of what I mean in a bit. But basically it will be what bertus posted:

    except I will have +-15 and +-12 coming from the center tap, 5V coming from the output of the +12V (I think) and the variable coming from the positive part of the center tap. This should keep it cheap with only two large smoothing caps needed.

  8. SgtWookie


    Jul 17, 2007
    If you ground the center tap of that 20v transformer, the "ends" will put out 10VAC, 180° out of phase with each other. You'll lose about .8v across the bridge rectifier, so you'll have around 13v on the cap. Keep in mind that the LM317/LM317 have a minimum dropout of 1.7v, but that goes higher with increasing current draw; you might lose as much as 2.5v across them at high output.

    Not to discourage you from building your own supply, but you could get this for about the same cost of the transformer you're considering:

    They have a 24VAC CT 4A xformer (12v-0v-12v) for less than what digikey's asking for the 20v 1.5A CT:
  9. dude521

    Thread Starter Active Member

    Nov 1, 2008
    Got it, that makes sense, thanks.

    So if I want +-15 V, I need at least 17v peaks and 12VAC RMS will give 12*1.414=~17V. The 7815 datasheet shows a 2V dropout at 1A so that should be fine right (I will be using less that 1A on these rails)?

    Out of curiosty, how do you calculate the total amount of current you can output. So if I had a simple power supply with an AC transformer of 12VAC rms and 1A max current and then rectified, smoothed and then regulated it with a 7815, how do I calculate the total amount of current the power supply can output (assuming its limited only by the current limit of the transformer)?

    I think I'll get this 24VAC center tapped transformer:

    The purpose of this is for a club, we're just doing basic stuff right now so this should be a good intro into soldering and power supply design. I want a project that is both useful and inexpensive. I am however impressed by how cheap that PS was, but it didn't have +-15V. Anyways, we will bulk order 10, so the price is only $13.30 each. Also that site looks like it doesn't export to Canada easily, you have to fax or phone.
  10. SgtWookie


    Jul 17, 2007
    The power supply I linked to had +12v/-12v and +5v too, which is generally all you'd need - but if you want a project, you want a project. ;)

    You have the Vf of the rectifier diodes to consider along with the dropout of the regulator (2v). You'll lose a volt across the diodes, so there's 3v gone.

    You also have ripple voltage to contend with. You have 120Hz ripple that you are filtering, so you have to have large enough filter capacitors after the bridge to keep the input to the regulators from dropping too low, otherwise the ripple will be coupled through the regulator under load. If your transformer is marginal for current and voltage output, you will experience ripple on the output at lower current levels than you would expect. However, if your filter caps are excessively large, you can have very high currents in your bridge rectifier.

    You might want to review this page:
  11. radiohead

    Distinguished Member

    May 28, 2009
  12. dude521

    Thread Starter Active Member

    Nov 1, 2008

    Ok, so this is the power supply. It will have +-12V, 5V and variable outputs. I have a few questions:

    1) I know each regulator is limited to 1A and will turn it self off, but if they are all maxed out, the total draw could potentially be 4A. Since the transformer is only rated at 1.1A, will it automatically limit itself to this amount? What happens if I try to draw more current?

    2) What should I do with the earth ground? My case is made out of plastic, so I can't ground the case. Should I just leave it unconnected?

    3) For the variable, i'm using a Bourns potentiometer. How do I know a knob will fit on this? Are the sizes all standardized?

    4) Anything else you can see?

    Thanks a lot.

    P.S. I'm in Canada, so most companies that have cheap stuff don't ship here. Digikey is my best bet.
    Last edited: Jan 22, 2010
  13. SgtWookie


    Jul 17, 2007
    The output voltage will drop off, and have lots of ripple on it.
    Connect it to the transformer mount; in case the primary shorts to the transformer core, the ground should help to protect you. You can also connect earth ground to the secondary center tap. Sometimes it's advantageous to have a "floating" supply, but that will come later on.

    1/4" is a common size shaft. However, pots are available with many different size shafts. If you are uncertain, look at the datasheet for your pot, or if you don't have a datasheet, measure the shaft itself.

    A) You do not have a fuse in the primary side. Line power (the hot side) should always be fused. Use a 2A slow blow. This will limit damage to your supply in case there is a fault, and won't trip the circuit breaker even if you have a dead short somewhere after the fuse.

    Your schematic should have connected the transformer wires to the pins labeled 1 and 2 on the P1 connector, via the switch and fuse.

    When you are building a project using mains power, the fuse should be the first thing the power goes through.

    B) Your capacitors on the negative rail are shown with incorrect polarity. If you connected them like that in a real supply, you would have a few exciting moments as they turned into stink bombs.

    C) Your rectifier bridge is rather skimpy on it's rating. If you use one rated for higher current, it should have a lower Vf - and will be more reliable.

    D) You should have 0.1uF (100nF) caps on both the inputs and outputs of every regulator to ground. These are cheap insurance against possible high-frequency oscillations/instability in the regulators. It's better to add them up-front and not need them, than discover you need them later on and try to band-aid the design.

    E) For R2, you need a 2.5k pot. If you use a 5k pot, you will only be able to use about 1/2 the travel. It would be OK for range if you had 30v coming in to the LM317, but you don't.

    You could increase R1 to 470 Ohms and make the 5k pot usable, but that means you will have to put more of a load on the LM317 output until you will get the guaranteed regulation. LM317's guaranteed specs for regulation percentage are when 10mA <= load <= 1.5A. If R1 > 120 Ohms, you won't get guaranteed regulation with no other load on the output.

    It's essentially the same thing with the 7812/7912 regulators. You'll need 5mA <= load <= 1A to get guaranteed regulation. With those regulators, it's easy - just put a 2.4k resistor from the output to ground, and there you have your 5mA load.
  14. dude521

    Thread Starter Active Member

    Nov 1, 2008
    Ok, thanks.

    I understand connecting earth GND to the case of the transfomer, so if there is a short, the current goes to earth. But is connecting the GND of the main circuit to earth a good idea? Won't that mean earth is being used as a return path for the current in normal operation? Isn't that illegal in most countries?

    Ok, so the diameter is fine but what about the shape? For the potentiometers, I see slotted and flatted actuators, but in the knobs I don't see any mention of what type of actuator they will accept in either the description or datasheet (looking through digikey). Do knobs accept both types?

    Is a PTC fuse a good idea? I found one with a 2A trip and 1.2A hold.

    Stupid mistake, fixed.

    Ok, so I got one that is 4A instead. I don't understand how higher rated diodes have a lower Vf?


    Thanks, I should have checked that first. I made the pot 2.5K.
    Last edited: Jan 23, 2010
  15. Audioguru


    Dec 20, 2007
    Did you calculate the size of the heatsinks for the regulators?
  16. SgtWookie


    Jul 17, 2007
    The idea is that if the HOT shorts to the case of the transformer, the current to ground will blow the fuse. If you are on a GFCI breaker, the GFCI breaker will trip. (GFCI = Ground Fault Circuit Interrupt, these usually trip at low currents, <40mA)
    No, it's not used as a return path, merely as a ground reference. Any current coming out of the transformer secondary must also go back into it. There should be zero current on the earth ground wire during normal operation.

    Generic knobs are usually for round shafts. If they are for flatted shafts, you can always use a file to put a flat on your shaft. This is a good thing to do anyway, as relying on tightening a screw to hold it in place is a bit iffy.

    No, if a 2A fuse blows, there's a problem. You want it to stay off until you fix the problem.

    Look at the datasheets for diodes/rectifiers, particularly at Vf vs current plots. As they approach their rated current, the forward voltage drop across them increases.

    You won't find a 2.5k pot; they come in 100 Ohm, 500 Ohm, 1k, 2k, 5k, 10k, etc.
  17. dude521

    Thread Starter Active Member

    Nov 1, 2008
    @Audioguru: I got a sink that is 4W @ 40C for the -+12V regulators (input voltage is slightly less than 14). And a 10W @ 50C for the variable and 5V regulator (they use the same input voltage). Does that sound good? Note: at 13V input, power = 1A * (13V - 5V) = 8W, so I figure going to 50C is ok, should be less since I have a fan in the case.

    @SgtWookie: I can't thank you enough for your help. You made my design much better.
  18. cumesoftware

    Senior Member

    Apr 27, 2007
    You will need a supply voltage never lower than 17V. It means that, considering that you don't have an infinite filter capacitor, you will have peaks with voltage greater than 17V. The ripple cannot go below 17V.

    Also, note that a centre-tapped 15V + 15V (RMS) transformer will barely do, considering that your mains voltage varies in the range of -10% to +6%, and the voltage drop on rectifier diodes (something in the order of 0.8V for the expected load current, 0.6V minimum). Also, I'm guessing that you will need an unnecessarily larger filter capacitor. You would be more confortable with a 16V + 16V transformer or higher.

    As a final note, you might not want to exceed a ripple of 10% after the filter capacitor, even if the voltage at the dips (in full load current) is greater than 17V. Excessive ripple might be "seen" as noise after the regulator.