Efficient, small voltage regulator.

Discussion in 'The Projects Forum' started by bustaplz, May 19, 2010.

  1. bustaplz

    Thread Starter New Member

    May 19, 2010
    5
    0
    Hi everyone! First post here, I've read through allaboutcircuits.com, though.

    At work we are using some devices that require 7V and 1.2-1.5A and we are powering them with a 24V DC supply(everything else in the panels containing these units use 24V.) Using a simple LM338 adjustable circuit is proving to be more and more trouble due to the LM338 overheating. We cannot fit a large enough heatsink on them due to size constraints of the boards.

    So, I'm trying to come up with a better way of getting the required 7V and keep the circuit small and inexpensive.

    My first thought was running something like a 7812 or 7815 in front of an adjustable regulator, but using simulation programs I can't tell if this reduces the heat production. This post asking for help may be a bit premature as I have not yet tried anything outside of my simulator, but I thought maybe someone could give me some ideas or different methods of making a circuit like this.

    Thank you in advance. :)
     
  2. retched

    AAC Fanatic!

    Dec 5, 2009
    5,201
    312
    How about another transformer to step-down to closer to 7v (or to 7v)

    A step to 12v prior, would help keep the LM338 considerably cooler.

    Get a 24v with a 12v center tap to run the LM338 and the rest of the box. This is the least messy option.
     
  3. bustaplz

    Thread Starter New Member

    May 19, 2010
    5
    0
    Excuse me if this is ignorant.

    I thought that transformers only worked with AC. I guess I could take a transformer from the supply of the 24V DC supply and step it down. If I could take the 120VAC down far enough a diode rectifier circuit would be pretty easy.

    But I don't think there is a transformer small enough to accomplish that, and having 120VAC on this circuit would be a very big hazard.

    Maybe I missed the point of your post? If you could, please elaborate.
     
  4. retched

    AAC Fanatic!

    Dec 5, 2009
    5,201
    312
    I was thinking you were talking about HVAC type 24vAC controls and rectifying to the LM338.

    Thats why you should ask questions kids!

    sorry.

    So in your case, you would need a DC to DC converter.

    You could also, dissipate some of the power in a high wattage resistor prior to the LM338

    The thing is, the power has to go somewhere.. Why have it all leave through the lm338?

    You could use a sandstone resistor, or concrete or, anodized aluminum types to drop the voltage before the 338 to a decent level.
     
  5. retched

    AAC Fanatic!

    Dec 5, 2009
    5,201
    312
  6. oidium45

    Active Member

    Apr 24, 2010
    130
    8
    This probably is not going to help much considering the current requirements but perhaps a DC voltage divider circuit? You would need some good higher wattage resistors! (not radio-shack specials) I would imagine that the resistors would dissipate heat but if the ratings are high enough they should hold up. You could then connect your LM338 to the output of the divider with a lower voltage input to the chip.

    Try this page http://hyperphysics.phy-astr.gsu.edu/hbase/electric/voldiv.html
     
  7. retched

    AAC Fanatic!

    Dec 5, 2009
    5,201
    312
    That crossed my mind. The voltage divider would get him to 12v. That would be fine for the 338.

    I posted a link to digikey's 50w 1% resistors. around $5us. I have a 1o 50w I use for current to voltage readings.
     
  8. Ghar

    Active Member

    Mar 8, 2010
    655
    72
    I've been using some of these guys lately and they seem to work pretty well.
    They're like the 78xx series but switching so they're very efficient.

    http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=102-1717-ND

    Unfortunately they don't have a 7V version and are limited to 1A but they're something to keep in mind. There are a lot of small DC-DC converters on Digikey.
     
  9. ifixit

    Distinguished Member

    Nov 20, 2008
    638
    108
  10. bustaplz

    Thread Starter New Member

    May 19, 2010
    5
    0
    I don't have much experience with voltage dividers, but have just played around with the idea in sim. I left the 338 circuit as is and just added a voltage divider to the incoming power. This seems WAAAY too easy, so I'm obviously missing something. What value resistors would you recommend using? And how much power will these resistors dissipate? Thank you for all the links, too!
     
  11. oidium45

    Active Member

    Apr 24, 2010
    130
    8
    Look into simple parallel circuits a bit and you will figure things out. As for the power dissipation, it depends upon total current. Some simple ohms law calculations should help.
     
  12. mshouppe

    New Member

    May 17, 2010
    3
    2
    I use these switching DC/DC converters for most of my projects. I have had success with the +5v and +9v versions. I don't know what your tolerances are, but TI makes a +8v version:

    http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=78SR108HC-ND


    If you absolutely must have +7v, you can feed the output from this DC/DC converter into your existing voltage regulator to reduce heat.
     
  13. oidium45

    Active Member

    Apr 24, 2010
    130
    8
  14. bustaplz

    Thread Starter New Member

    May 19, 2010
    5
    0
    Looks like Digi-Key also carries the 7.15V model. Any idea if the pins on these would fit into holes layed out for a TO-220 package. If they could, I could keep the 500+ boards I have and simply put this on. The only issue I see being the cost. $30 is quite a bit just to get 7V. I'm going to show the 78SR174VC-ND to my boss and see how he feels about the price. Meanwhile I'm going to keep trying to come up with a less expensive alternative if possible.

    You have all been so very helpful so far. I'm glad I finally decided to join the forum. :D
     
  15. John P

    AAC Fanatic!

    Oct 14, 2008
    1,632
    224
    It can't possibly be pin compatible with an LM338, as that has no ground pin. A 3-terminal non-adjustable regulator will have input, ground and output.

    I hate to say this, but anyone who makes 500 of a board which throws away over 20 watts and has inadequate heat sinking, needs some basic lessons in circuit design. You can make a very inexpensive and efficient switching voltage regulator with just a few components, but it won't fit the boards you have.

    However, having made that grumble I've got a suggestion. How about making a new little PCB which would stand off the existing one, using wires into the present components' outline as "legs"? This new board would carry the components of an efficient step-down switching regulator. You can do it for way less than the cost of the TI parts, it'll run cool and won't make you generate power at 24V just so you can raise the air-conditioning bill.

    Time for some thinking outside the box.
     
  16. Wendy

    Moderator

    Mar 24, 2008
    20,764
    2,535
    I like this idea best. Anything using a resistor or linear regulator will generate gobs of heat, while a simple switcher will be as cool as a cucumber. Another benefit, if you have 7V at 2A, the 24V side will have less than an amp draw. Switching regulators are converters, which means they don't generate nearly as much heat.
     
  17. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    I think you're going to be better off using IFixit's idea on a daughterboard. You might mount the daughterboard vertically to save space. Use a 90° 3-pin header to make the connection between the daughterboard and your LM338 pads. You'll need to remove your LM338's R1 and R2, replacing R2 with a piece of wire to supply the daughterboard with ground.
     
    Last edited: May 20, 2010
  18. bustaplz

    Thread Starter New Member

    May 19, 2010
    5
    0
    Well I did not design the circuit or the board. But the board actually uses two separate circuits.

    LM338 for adjustable output.
    78XX for fixed output.

    The LM338 side works okay when adjusted to higher voltages, but this is a newer application and the 7V regulation is just too much. The boards are perfectly capable of being used with both circuits, or more commonly, just the 78XX fixed circuit. So having 500 or so of them is not really as bad as it seems, and I will create a new board for this project if necessary, but being use them would obviously be nice.

    The main thing I'm trying to account for is the fact that every board we use and produce is designed to mount in snap-track(76.5mm long.) I came across this this item, and it seems like it might be worth a try, but the $25 price is still not ideal.

    http://www.dimensionengineering.com/DE-SWADJ3.htm

    Could anyone show me a schematic of a stupid-simple switching regulator circuit? Maybe if I can understand the basics of this type of circuit I could come up with something on my own that fits my needs.
     
  19. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    I started to post the dimension engineering product, until I remembered that the pin layout of their regulator is not the same as the LM338; it matches the 78xx series of regulators, which has a different pin layout.

    Really, IFixit's schematic is quite simple as far as switching regulators go.

    I'll have to respectfully disagree with Bill Marden though; since the buck-type regulator really isn't converting power, you will need just a bit more current in than you will get out due to losses in the inductor and the regulator. If you want greater efficiency, you will need to use an offline push-pull converter with a toroidal transformer primary:secondary ratio of roughly 3 to 1. I certainly don't think you're looking for something that complex, though.

    The most basic buck-type regulator consists of an input capacitor, an inductor, a switch (usually a MOSFET), a diode across the switch, an oscillator/regulator to control the MOSFET, and an output filter. It would be a lot more complex to try to build something like that out of discrete components.
     
  20. Wendy

    Moderator

    Mar 24, 2008
    20,764
    2,535
    Yep, and you said the same thing about puck buck on 3W LEDs, but I have the experimental data to back it up. :) 300ma in, 700ma out, the exact input current depends on input voltage. Up the input voltage and the input current goes down, while the output current is a constant and the buck puck stays cool.

    Any decent switching scheme coverts voltage/current and tries to keep power constant. That is the strength of the system, and it is also why they run cool. Electrical wattage in equals electrical wattage out plus the losses due to inefficiencies, which show up as heat. As a general rule of thumb, anytime you aren't dissipating a lot of power in a power supply regulator, there is conversion going on. This is why they are running cool.

    Linear regulators are not converters, they dump the excess power as heat, which is why they make good space heaters.
     
    Last edited: May 21, 2010
Loading...