4
\$\begingroup\$

Is there a switched-mode DC-DC conversion circuit that makes it easy to configure different input/output voltages?

Circuits given on the datasheets of the LMR12020 and others have a large number of discreet components in certain combinations. Is there a much simpler circuit that is:

  • easy to configure output voltage (using standard components)
  • >= 90% efficiency [edit: 80% is fine]
  • inexpensive in production (BOM under $5)

Example Configurations:

  1. 9-20V -> 14.4V (or 15V) @500mA [to charge lead-acid battery]
  2. 10.5-15V -> 12V @500mA
  3. 5-15V-> 3.3V @100mA

What are some inexpensive solutions with straightforward calculations to set the output voltage?

\$\endgroup\$
  • \$\begingroup\$ Your input range is very wide, for smaller input range and multiple outputs, you can check Intersil. \$\endgroup\$ – AKR Oct 4 '13 at 16:32
  • 1
    \$\begingroup\$ Greater than 90% efficiency is not going to happen on any option unless you use a synchronous buck regulator (with inbuilt low-side FET that replaces the inefficient external schottky diode) - see my answer. \$\endgroup\$ – Andy aka Oct 4 '13 at 20:20
4
\$\begingroup\$

The problem with a circuit that fits all is that it is optimal in none. So, in order to support the wide voltage range that you need, you'll need to pick a part that might not be ideal in some aspects such as size, output ripple, etc.

THere is no "configuring" the input voltage. The input is a voltage range, and in your case you want 20V as max, 5V as min. However, your requirement to go from 6-20V and others means you're looking for a buck/boost solution. The output is usually easily configurable with a pair of resistors in the feedback network.

The biggest problem with your specs is that you didn't specify output current. This is one of the most important parameters and choosing a buck/boost device (or any power supply device) is impossible without knowing this parameter.

I looked online and a nice part that will meet your specs (as much as I have) is LTC3115 from Linear, but this part is not cheap.

Your desire to reduce complexity with one solution is understandable, but you will be better served by finding parts tailored fit for your solution, especially as far as reducing costs. Most DC/DC converters are based on the same 3 topologies and are not very difficult to use.

\$\endgroup\$
  • \$\begingroup\$ +1 for the LTC3115 mention and other helpful hints. The output was specified at 2A, but it was meant to be a ballpark guide, rather than a hard limit. The LTC3115 would be the answer, if it was half that price ($13@1, $7@100), but it's still a great idea where both buck and boost are required. I'll clarify the question. \$\endgroup\$ – Brent Faust Oct 4 '13 at 18:15
  • \$\begingroup\$ @Rubistro: Yeah. Its more expensive than I thought. \$\endgroup\$ – Gustavo Litovsky Oct 4 '13 at 18:51
4
\$\begingroup\$

In a perfect world a perfect buck regulator would be like a linear regulator but with one more component, namely L, the inductor in the circuit below.

enter image description here

This could reduce the supply voltage to (say) 5V just like an LM7805 would but with the added bonus that efficiency would be close to 95% i.e. it won't dissipate a lot of heat.

What you see more usually is this: -

enter image description here

This would be an adjustable "perfect" buck regulator - note the two resistors providing a "sniff" of the output voltage at a ratio that defines the real output level. Hardly any different to a regular adjustable linear voltage regulator except better efficiency of course.

More often than not (such as in the LMR12020) you get something like this: -

enter image description here

The extra components are shown with red circles. The boost capacitor helps the internal power FET turn on better and therefore improve efficiency - this is a "don't-care" on linear regulators because they have poor efficiency anyway. The external schottky diode is present because this isn't always easily implementable in some chips.

There are quite a lot of switchers that do have an internal "pull-down" FET and these are called synchronous switching regulators and are more efficient (and have less ripple voltage on the output) than those which use a schottky diode.

So "simple" is getting a little more "complex" and this is just the Buck regulator - it is only able to provide an output voltage that is smaller than the input voltage - it is a step down regulator.

All but the first example provided by the OP could use a simple buck regulator and, depending on the output current requirements and input voltage range, may need to have their inductor value changed to optimize performance.

The first example in the question is both a step up (boost) regulator and a step down (buck) regulator because to get 15V out the input voltage range is 6V (boost required) to 20V (buck required).

The LMR12020 would only give you half the circuit, namely the buck part - however there would be a problem because the required input voltage range extends to 20V and this is at the spec limit for the device. Let's say the OP would be happy with 6v to 18V on the input; this could be converted down to say 4V then you could put a boost regulator on the output to "jump" this up to 15V - but there's another problem and that is the maximum current that the LMR12020 can provide - 4V @ 2A (limit) is 8W and this power is the maximum that is available (via a boost regulator of any type) to the output (15V) - this means o/p current is more like 0.5A.

Conclusion Depending on your opinion, your skill set and your experience, you may regard this as simple or you may not.

\$\endgroup\$
  • 1
    \$\begingroup\$ +1. Wow. I was about to start whining that the original question was too broad and not helpful - but your answer addresses exactly the issue I'm having with it, and in the best possible way. \$\endgroup\$ – zebonaut Oct 4 '13 at 13:54
  • \$\begingroup\$ @zebonaut I'm pleased to assist. BTW do you live in a park in the centre of Munich LOL? \$\endgroup\$ – Andy aka Oct 4 '13 at 18:16
  • \$\begingroup\$ Heh. That's rounded to the first decimal only ;-) Now that you mention it: The spot ended up being just outside a graveyard. I haven't even checked that carefully. 10 m to the West and it would actually be quite macabre... \$\endgroup\$ – zebonaut Oct 4 '13 at 20:07
  • \$\begingroup\$ @zebonaut yeah I noticed what looked like a church but google earth aint too hot at spotting individual graves!! \$\endgroup\$ – Andy aka Oct 4 '13 at 20:11
  • \$\begingroup\$ +1 Brilliant. Very nice explanation of why the external discreet components are required. I've since found the TLV70433 which requires only 1 external component and would work for 3,3.3,3.6, and 5V out and a huge input range of 2.5-20V (where < 150mA is required). Would like to find an equivalent in the 2A range. Is there an easy formula to set the output voltage on the LMR12020 (is it the ratio of R1 to R2)? \$\endgroup\$ – Brent Faust Oct 4 '13 at 20:57
3
\$\begingroup\$

3.3V

For your 3rd example, the TPS560200 is just the ticket. It is a 500mA Synchronous Buck Converter that accepts a wide range of input, has a configurable output, and requires very few external components:

  • Vin = 4.5-17V
  • Vout = 1-6.5V
  • requires very few discreet components: (1) inductor (2) capacitors (2) resistors - to set the output voltage:

circuit

To top it all off, it's only $.35 in quantity and to configure the output voltage, the calculation is simply:

Vout = R1/R2

Only 3 external Components

Even better, there is a fixed-output version for 3.3V - so you don't even need the two resisters for 3.3V output (!) leaving only 3 extra components (inductor + standard Vin/Vout caps).

\$\endgroup\$
  • 1
    \$\begingroup\$ This device is OK providing it's not feeding sensitive analogue circuits because it uses a fixed "on" period topology meaning the switching frequency can be quite low under light loads = higher ripple voltage and because its at a lower frequency it means you might need an extra series L and decoupling. \$\endgroup\$ – Andy aka Oct 7 '13 at 8:20
  • 1
    \$\begingroup\$ @Andyaka Nice observation. The switching frequency is fixed at 600kHz during normal operation. So are you referring to Eco-Mode operation? There is apparently no way to shut this "feature" off, but at this price/performance, an LC filter on the output seems like a reasonable way to limit the ripple for those applications that require it. \$\endgroup\$ – Mike Oct 7 '13 at 18:07
-1
\$\begingroup\$

Buy a step-down converter that creates the lowest voltage of the three (case 3 with 3.3 volt). Connect a step-up converter on the output to make the other two voltages (case 1 & 2, 14.4V & 12V).

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.