Inductorless Switching Regulator

Discussion in 'General Electronics Chat' started by jwilk13, Aug 3, 2011.

  1. jwilk13

    Thread Starter Member

    Jun 15, 2011
    228
    12
    Hello again,

    I've searched all over and can't seem to find what I'm looking for, so maybe someone here has some suggestions. I'm looking for a switching regulator to meet the following criteria:

    Input Voltage: Up to 30VDC (my supply is actually closer to 24, but its a battery so I figured I'd go above a little)
    Output Voltage: 5VDC
    Output Current: 1A

    I'm currently using the linear regulator here to get my 5V, and another 3.3V regulator to power a uC.

    I want to replace the 5V linear regulator with a switching regulator and power the 3.3V regulator from the switching regulator. The only problem is that most switching regulators require an external inductor, and I want to avoid that. Are there any other ways of accomplishing this? SgtWookie and I started talking about this previously, but the previous thread was a bit different. A schematic is posted for reference.
     
  2. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    You have 2 things going against you:
    1) The relatively high input voltage.
    2) The current requirement.

    There are capacitive DC-DC converters, but the only ones I'm aware of that are in production are for low voltages (up to around ~12v ) and limited currents (20mA or so). National has one that can output 700mA, but low input voltage (2.7v <=Vin<=5.5v) - and capacitive DC-DC converters are used for boosting or inverting the input voltage, not for buck regulators.

    I guess I don't understand your objection to an external inductor. It does not necessarily have to be very large. Toroidal inductors will have minimal energy radiation. There are very compact inductors from vendors like Coilcraft that can handle significant amounts of current.

    Now if you want something like a switching 7805 and 7833 replacements, Murata has these:
    http://www.murata-ps.com/releases/cdped259_78sr-2a.html
    And Dimension Engineering has a similar product:
    http://dimensionengineering.com/DE-SW050.htm?gclid=CNq-hbTls6oCFYnr7QodCEqP7g

    They still use an inductive switch, but it's all contained in one package.

    There are also DIP and SIP complete supplies available from various vendors that you can mount on your PCB.

    But if you're going to stay linear, you'll just waste most of the power used as heat.
     
  3. jwilk13

    Thread Starter Member

    Jun 15, 2011
    228
    12
    Thanks for the info. I was trying to stay away from external inductors due to board real estate and cost, but if there's no other way around it, I can make it work.

    Meaning the normally external inductor is contained within the package? Do you have an example?
     
  4. jwilk13

    Thread Starter Member

    Jun 15, 2011
    228
    12
    Ignore the previous post. I just did some calculations and measurements and the total load on the 5V regulator if I connect the 3.3V regulator input from the 5V output is less than 10 mA. All I'm doing is supplying power to the uC and voltages the A/D inputs, so not much current draw there.

    I realize it's still really inefficient to use linear regulators, but wouldn't a significantly larger amount of heat be generated if the load were in the 500 mA range (rather than my 10 mA)? I might just stick with the linear regulators for now if that's the case.
     
  5. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    I could only go by the numbers you gave me. If you give me wrong numbers, then I can't help the results.

    10mA is just enough to maintain regulation for a typical LM317, but you're using something different.

    To get from 28v down to 5v, out of 100% of the power you'll expend, about 17.8% will be in the load, and about 82.2% in the regulator as heat. So, if there's only a total of 10mA current, then that's ~23v*.01 = ~231mW power expended in the regulator; nearly 1/4 Watt. If your 5v load is 44mA, then you'll be just over 1w dissipation in the 5v regulator.
     
  6. mik3

    Senior Member

    Feb 4, 2008
    4,846
    63
  7. jwilk13

    Thread Starter Member

    Jun 15, 2011
    228
    12
    Yeah I apologize for the misinformation. I wasn't sure about my total load on the regulator before I posted.
     
  8. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    OK, well anyway if you need a compact supply to replace a 78xx series in the future you know you have some options. ;) Mik3 posted something very similar to those two links I posted earlier; all about the same size.
     
  9. tom66

    Senior Member

    May 9, 2009
    2,613
    214
    I designed a tiny buck converter which outputs 5V or 3.3V (or any voltage in between, programmed using two resistors) which is the same size as a 7805 in TO-220, plus a few mm on each side. (I'm convinced I could get it smaller by spec'ing in a smaller inductor, but I wasn't too bothered about it.)

    It's based on the MCP16301 buck converter IC. From 6V - 30V it can give you 5V @ 600mA. Lower input range than the 7805 although absolute maximum is 40V for this chip so it does beat the 35V absolute maximum of the 7805. (Is there a "recommended" maximum for the 7805?) For 4.5V - 30V it can give you 3.3V @ 600mA.

    If you assume a typical efficiency of 80% for a switcher, with your situation of 24V in and 5V out at 10mA, the switcher will only dissipate 10mW as heat. However, I'm not sure how well it handles low loads as some switchers have poor efficiency, usually 50% or so. Not as ideal for battery power, compared to a low quiescent current LDO, unless battery life isn't crucial.

    The MCP16301 is tiny, in a SOT26 package, and packs an N-channel MOSFET on the die, and costs about $1 in 25 qty. However, it's a new device, so only available in samples, I've ordered some, but there's a back order until December!

    The design is reverse polarity protected. And hot-pluggable - it can be dropped into a live system - that is accomplished by using a series 1 ohm 1/3W resistor to limit input ringing which is a problem with the built in ceramic capacitors on the input. Output ripple is 20mVp-p, typical, with the built in 44u ceramic filter. (Untested.)It runs at 600kHz so EMI is reduced, but still a concern as the inductor is not shielded and the layout is not optimised for FCC/CE which it probably wouldn't pass.

    I designed this straight in my PCB editor, didn't even bother with a schematic. It's a 2 layer board and meets 6/6 rules. I'm waiting for the boards which only cost me $2.50 for 6 as they are so small. (0.425" x 0.62") I'm not sure if it will work first time, as I've probably messed something up, but it was something I did in about an hour while I was bored.

    It even has a built in power LED, an optional feature.

    I'm planning to make a 1A version soon. This board could use an LM2734 which is rated to 1A, as the chip is pin compatible. But it only has a 20V input maximum. I'm looking to make it a 6-42V input and capable of 1-1.5A at 5V, but the switcher chip could be more bulky (probably a SO8.) There's the possibility of using a synchronous buck converter, with two built in MOSFETs, to improve efficiency and increase the current limit.

    It's open source, CC-BY-SA, make of it what you wish! I attach the design files.

    --

    Smallest component is 0603.
    C1 is a 100n 35V 0603 ceramic NP0/C0G
    C2, C3 are 10u 35V 1206 ceramic (can use 4.7u with greater output ripple and poorer input regulation)
    C4, C5 are 22u 6.3V 0805 ceramic (C5 can be omitted, with greater output ripple)
    C6 is a 100n 16V 0603 ceramic NP0/C0G (can use same as C1)
    D1, D3 are one of PMEG4010EJ or RB160M-40 or equivalent, SOD323 package
    D2 is BAV99W - dual diode, one half used (or possible BAS19, lower cost, single diode.)
    D4 is any 0603 LED, prefer green. (Optional.)
    L1 is an SDR0604 or PM54 type @ 15uH.
    R1 is a 1 ohm 0.33W 5% power resistor 1206
    R2, R3 are the divider resistors (for 3.3V: R2/R3 = 10k, 31.2k)
    R4 is resistor for LED (optional, recommend ~100 ohms)
    U1 is MCP16301 (=30V input, 600mA) or LM2734 (=20V input, 1A)
     
    praondevou and jwilk13 like this.
  10. jwilk13

    Thread Starter Member

    Jun 15, 2011
    228
    12
    Yep, great resources from you guys as always :). That's why I try to frequent these forums as much as possible. There's always more to learn.

    You think I'll get by with dissipating just under 250 mW in that 5V regulator, or am I going to be pushing my luck? The data sheet says that at a 55°C ambient temperature, the regulator will operate reliably at up to 440 mW power dissipation when mounted in what they call a "worst-case" (no ground plane, minimal trace width, FR4 material) scenario. I don't intend to mount it the way they describe there, and 55°C (131°F) is pretty warm.

    Thanks for all the help :)
     
  11. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    I don't have any information on how you are planning on mounting the board, or if it will be enclosed, or what it's operating environment will be. You will get better heat dissipation if you mount it vertically, as convection will cause air currents to flow across the board; not so much if it's horizontal.

    Adding a copper pour area around the regulator tabs will help a great deal to dissipate the heat.

    The big question is still "how much load are you really going to be putting on the regulators?" Don't leave out a thing.

    And are you willing to set down a power budget, and stick to it? If you don't, you'll burn things up.

    Exceeding the power budget is an all-too-regular occurrence, even among engineers with years of experience. When you already have lots of engineering time tied up in a design, and lots of funds committed to a project to purchase parts and have them assembled, then missing your product ship deadlines because you grossly exceeded your power budget is something that actually does happen; I've seen it happen and it was quite ugly.

    If you don't account for all current requirements, you'll wind up on the short end of the stick. You may have to do work-arounds like have the uC power everything down and take a nap for awhile to let things cool off. As I don't know what your project is supposed to do, I don't know if that's a viable option.
     
  12. jwilk13

    Thread Starter Member

    Jun 15, 2011
    228
    12
    Thanks, SgtWookie. As a relatively young engineer it's good to get advice like that before I end up making the same mistake. I think too many times (at least for myself), we neglect the actual power requirements and substitute a "this is probably more than enough" approach in place of the design theory we learned in order to expedite the process. Thanks to you, I'm not going to let that happen on this project, even if it is relatively simple :).

    I recognize that I still have a lot to learn, and I'm grateful for forum members such as yourself willing to stick with me in the process. Thanks again.
     
  13. bob332

    Member

    Feb 14, 2011
    80
    0
    here is the ds for the 1A one - http://www.recom-power.com/pdf/Innoline/R-78Bxx-1.0_L.pdf

    sadly, the recom site is jacked up and you basically have to know the part numbers to find the ds :confused:

    if you want pics of the actual units i have parts from the 78XX line and the larger 78BXX - and can put them next to a TO220 78XX for size reference if you need that.
     
Loading...