Shop OBEX P1 Docs P2 Docs Learn Events
Linear Regulator OR Switching Voltage Regulator ? (need advice, opinions) — Parallax Forums

Linear Regulator OR Switching Voltage Regulator ? (need advice, opinions)

diafysaldiafysal Posts: 92
edited 2005-08-30 19:11 in General Discussion
Linear Regulator OR Switching Voltage Regulator ? (need advice, opinions)

Hi
I'm designing a PCB with some IC's that needs 5volt (Max 500mA.). There is the PSC, compass, accelerometer and some IC's communicating by I2C.
I want to be able to use for example a 12volt battery (and/or 7.4 volt battery).
But should I use linear or switching regulator?

Linear:
low ripple on output
large power loss (12v to 5v)
gets warm?

Switching:
produces EMI
ripple on output
more efficient (12 to 5. But less efficient 7.4 to 5 ?)

I have looked at the LM2574 from National Semiconductor (what is the lowest input voltage?). But my concern now is the EMI and ripple that a switched regulator produce. How will this affect the PSC and other IC's ?

Feel free to have opinions!

Comments

  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2005-08-24 16:17
    Switching Voltage Regulators are a bit larger, but are far more efficient [noparse][[/noparse]80%].
    The Linerars are smaller and cheaper, but eat batteries like taco chips [noparse][[/noparse]33% efficient].

    So, much depends on design goals - small size versus independent micro power.

    The Linears are just fine for stay-at-home and plug-in-wall devices. They may even add a little warmth to a cold corner on a winter's eve.

    Yes, switchers have EMI problems. If you want to use those ultra-sensitive Opamps, you might worry. I don't think robots really are usually EMI challenges. If you must worry, you probably should have an oscilloscope to really see the problems and be able to filter out the nasty EMI.

    If you are just talking shop, nobody knows for sure. They have to build something and see the problems.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    G. Herzog in Taiwan
  • Guenther DaubachGuenther Daubach Posts: 1,321
    edited 2005-08-24 18:19
    Hi,

    I'm using the LM2574 in one of my commercially produced SX-based applications. The product passed EMI tests w/o problems caused by the switching LM2574 - the EMI generated by the SX caused much more trouble. In order to eliminate EMI as far as possible, I designed a 4-layer PCB for this device with two inner supply layers, i.e. one connected to Vdd, and one to Vss.

    So far, none of the about 300 LM2574s I have used for production failed so far. Seems as if this a a rock-solid component.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Greetings from Germany,

    G
  • diafysaldiafysal Posts: 92
    edited 2005-08-24 22:37
    Thank you for the replies.
    I'm thinking of designing the PCB for a “TO220 “ component (regulator).
    And later on try with a small switched regulator PCB to replace the TO220.
    Have not decided yet.

    I have been looking at Lithium-polymer batteries.
    Nominal cell voltage is 3.6v (max 4.2v and min 3v).
    So I would like the regulator to be able to use a 6volt battery (7.2 to 6).
    A low dropout-regulator (linear) can regulate 5.5volt to 5volt.
    Can the LM2574 “work” with as low voltage as 6volt?

    I have also been looking at the LM2674, and I think it can work with 6volt.
    But I'm not sure I have read the papers right though.
  • Ryan ClarkeRyan Clarke Posts: 738
    edited 2005-08-24 22:44
    I'm a big believer in efficiency , so I'd go with the switching supply (especially if I'm going to be running on batteries and want them to last longer).

    Just be mindful of part placement, keeping leads as short as possible.

    Gunther's suggestion of a 4 layer board with ground and Vdd planes is a good one as well.

    Ryan

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Ryan Clarke

    Parallax Tech Support
    rclarke@parallax.com
  • diafysaldiafysal Posts: 92
    edited 2005-08-24 23:44
    Kramer said...
    Switching Voltage Regulators are a bit larger, but are far more efficient [noparse][[/noparse]80%].

    Yes I had a look at component sizes and it gets larger! (might work though)
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2005-08-27 17:26
    YOu know....
    Even if you use those 7.2 volt Lithium cells [noparse][[/noparse]which are very handy], you get more actual use out of them with the switching regulator.

    That 33% versus 80% is about all the voltage drop that goes out in heat. Divide 80%/33% and it is obviously more than twice as many Amp/hours.

    Sadly, the T0220 footprint will not work with the switching supply. If you are going to pay for boards to be made, try to accomodate both into nearly the same area of the board. YOu then could pull the TO220 and cover the holes with the larger device.

    This would be a win-win design.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    G. Herzog in Taiwan
  • Robert KubichekRobert Kubichek Posts: 343
    edited 2005-08-27 18:26
    There are switchers that use the same pin-out as a 220, but cover a tad bit more area.
    Remember the heat sink takes up room as well. Some of the switchers can go up to 85-90%
    efficiancy....

    Look at the specs for the 78stxxx series from TI.....

    Bob N9LVU scool.gif
  • diafysaldiafysal Posts: 92
    edited 2005-08-28 01:17
    Robert Kubichek said...

    Look at the specs for the 78stxxx series from TI.....

    I can't find this?

    No, I did find it !
    focus.ti.com/docs/prod/folders/print/pt78st100.html

    Post Edited (diafysal) : 8/28/2005 1:28:27 AM GMT
  • Robert KubichekRobert Kubichek Posts: 343
    edited 2005-08-28 01:45
    diafysal said...
    Robert Kubichek said...

    Look at the specs for the 78stxxx series from TI.....

    I can't find this?

    It is the 78ST1XXYC, at 1.5 amps max which XX=volts, Y=chip package
    And the 78ST2XXYC, at 2.0 amps max which XX=volts, Y=chip package

    XX= 3.3v to 12v, Y = (V)ertical (S)urface (H)orizontal, mounting

    Made by Texas Instruments

    www.ti.com

    According to the pdf spec sheet, it has about the same footprint as the 7805 with a heatsink, but can also be mounted flush...

    Bob N9LVU scool.gif
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2005-08-28 15:53
    The 5.0 volt unit seems to require 500ma of current to reached higher efficency.
    You can use it with less, but the efficency falls off.

    Nonetheless, this is really a convienent conversion.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    G. Herzog in Taiwan
  • diafysaldiafysal Posts: 92
    edited 2005-08-29 01:05
    The difference in footprint is quite large.
    4.8mm x 10.4mm (STMicroelectronics LF50 TO-220)
    8mm x 23mm (http://focus.ti.com/lit/ml/mpsi048/mpsi048.pdf)

    But one could easily but three short wires between the PCB and the component. Placing the component a short distance away from the PCB.
  • Robert KubichekRobert Kubichek Posts: 343
    edited 2005-08-29 01:33
    diafysal said...
    The difference in footprint is quite large.
    4.8mm x 10.4mm (STMicroelectronics LF50 TO-220)
    8mm x 23mm (http://focus.ti.com/lit/ml/mpsi048/mpsi048.pdf)

    But one could easily but three short wires between the PCB and the component. Placing the component a short distance away from the PCB.

    Sure those dimensions are for the bare chip, when you add on the heatsink it will take up more room, plus you only get 1/3 the current, plus a lot
    of heat. There is more current with the ti packages, plus less heat...

    Bob N9LVU scool.gif
  • Dave PatonDave Paton Posts: 285
    edited 2005-08-29 13:52
    My personal preference has always been for both. I use a very high efficiency buck or flyback regulator to turn the 8-12V into about 5.7V, and then use an LDO (low drop-out linear regulator) to clean things up before they hit my 5V bus or my more sensative +/- voltage analog busses. The National Simple Switcher series has always done well for me, though I've recently gotten interested in the offerings from Micrel. It's the best of both worlds, if you can afford the board space.

    my $0.02

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    This is not a sig. This is a duck. Quack.
  • Jim G.Jim G. Posts: 27
    edited 2005-08-30 19:11
    I like your approach of using both, Dave...

    I have been working on an SX circuit to control 24VAC sprinkler valves, and regulating 30+ VDC to 5 VDC warms the 7805 for even small currents.· This way you could in some cases·power LEDs and relays using the intermediate voltage and maybe get away with a low-power, TO-92 regulator for the CPU.

    It think I'll try it out, thanks.

    Jim
Sign In or Register to comment.