1
\$\begingroup\$

for example the LM2678-ADJ is a famous buck converter, there are also many others. I have googled and looked through other regulators, there are many schematics in the datasheets of how you make the output voltage adjustable.

But there does not seem to be any info about current limiting for buck converters. Why is that? Is it harder to make current limiting work on a smps rather than a linear regulator?

LM2678 datasheet: http://www.ti.com/lit/ds/symlink/lm2678.pdf Thanks!

\$\endgroup\$
  • \$\begingroup\$ Next time please put in the effort of linking a datasheet. You get out of it what you put in. Your example is a Simple Switcher Buck converter - not quite a full SMPS driver. Buck converters don't need to monitor current very stringently to stay efficient, boost converters often do. However, there are many, many, manymany converters, both Boost and Buck that have a settable resistor to drive Current-Foldback Protection. \$\endgroup\$ – Asmyldof Apr 22 '16 at 6:59
  • \$\begingroup\$ Sorry about that. updated with datasheet. I see, but what if i wanted to use it in an adjustable power supply and controll the current with a pot. is this possible? \$\endgroup\$ – Xane Apr 22 '16 at 8:04
  • \$\begingroup\$ I see 7A mentioned as the internal current limit in the LM2678 so what's your problem? \$\endgroup\$ – Andy aka Apr 22 '16 at 9:57
2
\$\begingroup\$

You must not have looked very hard to not find a buck controller with selectable current limit. LM5088 is one example, there are many others.

That being said, there's no "natural" place to measure current in a buck controller. With a boost controller you can measure the switch current from source to gnd with a resistor, easy as a pie.

With a BUCK converter you cannot measure anything against GND, it's a differential measurement across the switch. You could certainly insert a resistor in series with the switch and use a differential opamp circuit to measure that. I actually did so with my first custom buck converter back in the day. However that eats up power as you need fairly large voltage drop to measure it reasonably and/or you need an expensive differential opamp to get the job done.

Therefore usually the current sense is integrated to the controller itself and there are various ways this is done. One popular method with integrated switchers is to measure the voltage drop over the (internal) mosfet, not that difficult to do if you're already doing your own custom chip. Other styles are measuring the low-side current as with LM5088 but this cannot unfortunately used directly but requires some processing. Yet another method is to measure ripple current in the output capacitor, which TI calls DCAP control.

There was a point when manufacturers started pushing voltage mode control chips again but you seriously want to stay away from those, they're evil. Well, for some applications they can be OK but have serious drawbacks which is beyond the scope here.

P.s. LM2678 may be "famous" but it's something of a hurr-durr controller. "Simple" may be good for hobbyist or for a designer fresh out of school on his first design (ahem) but for serious work it's far too restrictive.


\$\endgroup\$
  • \$\begingroup\$ For some newer current-mode controllers, you can do inductor DCR sensing, or simply put a few milliohm shunt resistor on the output since the controller has an integrated high-gain differential amp as well. FET sensing as you said works as well, but can be tough to get a precise current limit (especially with external devices) because RDS(on) varies with temperature and VGS. \$\endgroup\$ – Krunal Desai Apr 22 '16 at 16:28
  • \$\begingroup\$ Thank you very much for the detailed explanation, i have never made any smps before only linear ones, and that is because i don't know much about switching yet. I'm reading about eletronics myself (from home) and i don't really have a degree or anything. Regards! \$\endgroup\$ – Xane Apr 23 '16 at 3:33
  • \$\begingroup\$ @Xane nothing wrong with using s simple smps controller. There are quite a few with "integrated" or "internal" compensation. These are fine if you need something that will very likely work with minimum effort on the design. However you're stuck with predefined combinations of inductors and capacitors. Texas has excellent online tools for designing smps circuits, I have designed three in one day for various loads which worked just fine. Earlier that'd involve a lot of calculations, simulations, testing on prototypes.. \$\endgroup\$ – Barleyman Apr 23 '16 at 14:34
  • \$\begingroup\$ As a trivia current mode control is actually much more difficult to simulate/model than voltage mode but almost everyone glosses over the current measurement circuit with perhaps some fixed compensation value by manufacturer. If you so something exotic like 12V to 2kV boost/flyback you're so far out of the envelope you get into trouble. 95% of the cases it's not necessary. \$\endgroup\$ – Barleyman Apr 23 '16 at 14:40

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.