SMPSU design

Discussion in 'The Projects Forum' started by M1EUF, Sep 1, 2010.

  1. M1EUF

    Thread Starter New Member

    Sep 1, 2010
    6
    0
    Im working on a power supply project to use with amateur radio equipment, im looking for 13.8v at 30A or so. Im working on a shoestring budget, so recycled components from sources such as junked PC power supplies and CRT TVs are the order of the day. I reckon the half bridge topology would be best for my requirements. For the time being, I have just been using the half bridge to drive a filament lamp so I know there is going to be no problem with inductive voltage spikes. TV horizontal output transistors seem to be a good choice, should be plenty about but I haven't yet found a cheap source. (Tried BU208 with success but its a bulky TO3 device.) I expect to be blowing a few up over the course of the project so I dont want to be spending much on them.
    For the transformer I intend to use a 2 inch ferrite ring out of the 12V to +/- 36v converter section of a scrap car amplifier (good cheap source of parts, 3GBP from a car boot sale), the original amplifier transformer worked at 40kHz in a push pull arrangement and I expect the power rating is adequate. Should be easier to wind than an EDT transformer too.
    Im using the base drive transformer out of a computer SMPS, so no rewinding to be done and the isolation should be up to scratch. I think they use a single turn to provide positive feedback to reduce drive requirements and make the supply self starting.
    I have already had some exploding transistors (IRF730s) and would like to arrange protection on the primary side. If I used Mosfets I could use the voltage drop across a low value source resistor to short out the gate drive, possibly latching in this state (SCR made fron npn and pnp pair). Tried this with bipolars but it left them in a high dissipation state.
    I do have some high power audio transistors rated at 230V and about 8A. I know the voltage rating doesnt give much headroom, im wondering about fitting a 200V TSD across each transistor as protection for the time being against spikes, probably wont be needed once a suitable snubber is in place. I am also considering the 2 switch forward convertor since no snubber is required with the clamping action, but this means changing to a gapped core.

    Your comments please!
    Regards, John
     
  2. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,670
    804
    What is your supply voltage?
     
  3. marshallf3

    Well-Known Member

    Jul 26, 2010
    2,358
    201
    If you come up with an easy way to hand wind a toroid that contains more than just a few turns the world will thank you.

    One tip I can pass on, unless it's just a few turns start by dividing the wire in half and wind it half at a time. Less wire to pull through on every turn.
     
  4. Papabravo

    Expert

    Feb 24, 2006
    10,166
    1,797
    I believe this is a fools errand. Making a high performance SMPS out of junkbox components is no easy task. The required test equipment alone could bust your budget. Point to point wiring is another problem, you really need to do this on a PC board adding another expense. If your hobby is amateur radio then buy an Astron Power supply on eBay for $100 bucks and devote yourself to building the best antennas you can.

    Good DX
     
    Last edited: Sep 2, 2010
  5. M1EUF

    Thread Starter New Member

    Sep 1, 2010
    6
    0
    Regarding the input voltage, my mains supply is 240V so im expecting about 340V max across the two smoothing capacitors on the input side. All of our appliances here are now marked up for a 230V supply to bring us 'in line with Europe' but nobody has been round changing taps on transformers, they just changed the tolerance I believe.

    Ok on the toroid being a bit of a pain to wind, but im not being paid for it so Im not too fussed how long it takes. I would rather use what I already have to hand rather than wait for items to arrive in the post. I dont imagine its going to have that many turns on the primary that winding would be an issue. I may still use an EDT transformer yet, nothing is set in concrete.

    Yes it would certainly be easier and cheaper to go buy a ready made power supply, but wheres the fun in that, and what will I learn on the way? Im looking forward to the challenge, and while some radio hams are happy to buy a black box and start chatting, some of us would get more satisfaction from building a simple transmitter out of a discarded CCFL lamp. I have just completed a linear power supply for 13.8V at 25A from junk box parts, I even rewound the toroidal power transformer. It has cost far less than a ready made commercial unit and is of better quality. I dont need my SMPS to be state of the art, just more efficient than the linear supply just mentioned.

    I have bought the 'Power Supply Cookbook' by Marty Brown as a starting point, it arrived in the post today so Im on with some good bedtime reading. Thanks for your replies so far guys.

    Regards, John M1EUF
     
  6. Papabravo

    Expert

    Feb 24, 2006
    10,166
    1,797
    Uhhh..Ohhh You won't be putting the mains across ANY smoothing capacitors. You will be using a transformer to drop the voltage from 240 down to a more manageable level. There is NO SAFE WAY to build a power supply without one. If you insist on trying, then the mods will close this thread immediately.
     
  7. Ghar

    Active Member

    Mar 8, 2010
    655
    72
    Stepping down 340V to 13.8V wouldn't be fun either, it's pretty difficult to regulate 4% duty cycle...

    It will be a similar conclusion with all the standard topologies. A transformer really adds a lot of benefits.

    Then again, it does sound like he *wants* a transformer...
     
  8. Papabravo

    Expert

    Feb 24, 2006
    10,166
    1,797
    Yes but the SMPS part doesn't start with the mains. If it does the mods will close this thread, because you cannot make such a supply safe -- period. You get a step down transformer convert 240 VAC to 18VAC, bridge rectifier, Pi-filter to get 24VDC unregulated. Now you build an SMPS that outputs 13.8V @30 A. This is where it gets tricky because you need really thick wires and traces, inductor cores that won't saturate, and really beastly semi-conductors that will withstand tremendous abuse. Without a really good oscilloscope and an LC Bridge you won't have much of a chance of building a 30 Amp supply from junkbox parts in this lifetime.

    Just for grins and giggles assume you need an inductor that will handle twice the peak output current or 60 Amps. A conservative estimate of the wire size at 700 circular mils per amp would be 42,000 circular mils. Question boys and girls: What wire gage has 42,000 circular mils of cross sectional area. What is the diameter of the wire? Can you really wind an inductor with that stuff?

    Answer: AWG 4 has 41,738 circular mils and is 204.3 mils in diameter.
     
    Last edited: Sep 2, 2010
  9. Ghar

    Active Member

    Mar 8, 2010
    655
    72
    Oh I see what you mean, I was confused for a minute.
    I can't blame him for this, having a power frequency transformer defeats one of the major advantages of an SMPS.
     
  10. Papabravo

    Expert

    Feb 24, 2006
    10,166
    1,797
    Wait a minute. You can still have the efficiency benefits of an SMPS even if you start with a step-down transformer. The losses in an equivalent linear power supply come from the (Vout - Vin)* Iout. The losses in the step-down transformer are small by comparison especially at 30 AMPS.

    The step-down ratio is 240/18 ≈ 13.33..., so 30/13.333 = 2.25 Amps from the mains. I suppose if you used a 1:1 isolation transformer and did the DC-DC conversion from 340 VDC to 13.8 VDC you would be back to a very narrow duty cycle range and there is no real benefit to that.

    I too have a ham shack and I prefer my Astron Linear supplies. If I had a kilowatt amp I might have to reconsider
     
  11. Ghar

    Active Member

    Mar 8, 2010
    655
    72
    Yeah you'll have the efficiency but you'll also have a rather large 60 Hz 400+ W transformer.

    Pretty much all commercial SMPS switch rectified mains and use a small high frequency transformer.
     
  12. Papabravo

    Expert

    Feb 24, 2006
    10,166
    1,797
    I'm not understanding. If you start with 240VAC 60 Hz, are you saying that the next step is to convert that to 240VAC 40kHz. How would you do that?

    240VAC 60Hz to 340VDC to 240VAC 40kHz for example
    Now use a small transformer to step the voltage down to 18VAC at 40 kHz and proceed as in the 60Hz case.

    NOW - How do you make this constuction SAFE???
     
  13. Ghar

    Active Member

    Mar 8, 2010
    655
    72
    Well for example check out this old computer power supply:
    http://www.smpspowersupply.com/ATX_power_supply_schematic.pdf

    You rectify the AC to a poorly smoothed high voltage DC, then switch that at 40 kHz or whatever across the transformer.

    You still have a transformer with all the safety it brings, it just so happens you have some electronics on the primary side. The outputs will still be safe but working on and testing the primary side electronics won't be.

    That transformer will be a fundamental part of the down conversion, so while a buck converter will be Vout = D*Vin a forward converter (like a buck with a transformer) will be Vout = D/n*Vin, so the turns ratio shows up in the conversion factor.

    Edit:

    So, that would be 240 Vrms AC -(rectifier)-> 340 Vdc -(xfmr + switching)-> 13.8 Vdc
     
    Last edited: Sep 2, 2010
  14. M1EUF

    Thread Starter New Member

    Sep 1, 2010
    6
    0
    Just to clear things up, I intend for the power supply to be of the ISOLATED half bridge type, driving a high frequency step down transformer. A 1:1 isolation transformer will be used on the input side during the experimental phase for safety purposes. I do have a dual trace oscilloscope and LCR meter amongst my collection of test equipment by the way. I may consider using a 50Hz transformer ahead of the switcher, but it seems a shame to weigh it down with all that iron.
     
  15. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,670
    804
    As usual. Transformer tested for high voltage leakage current and output ground connected to mains ground.
     
  16. Papabravo

    Expert

    Feb 24, 2006
    10,166
    1,797
    That feature was not made clear from the original post
     
  17. M1EUF

    Thread Starter New Member

    Sep 1, 2010
    6
    0
    Thanks for the further replies so far. I apologise if the original post was unclear, I thought it would be assumed that I would be using a high frequency transformer to provide isolation when I mentioned my plans to go for the half bridge topology. For those that are interested, these links show 3 SMPS designs made for amateur radio purposes, some based around junk box components:
    http://w5jgv.com/hv-ps1/
    http://www.qrp4u.de/docs/en/smps_new/
    http://ludens.cl/Electron/PS40/PS40.html

    I think I am pretty much convinced that half bridge is the way to go. The remaining unknowns are what power switches to use; bipolar, Mosfet or IGBT, and whether I should have the control IC (I have available SG3524, TL494 and SG3525) on the primary or secondary side. Im considering using synchronous rectification to improve efficiency along the lines of the '250W smps with power FETS' in the link above. My main requirement is rapid shutdown if there is a short on the primary side, say if the output rectifier fails s/c, so that the switching transistors will survive. I am under the impression that current mode control will satisfy this requirement.

    Regards, John M1EUF
     
Loading...