Can I use a buck converter but bypass if the input voltage is low?

[sofakng]

I'm trying to use the Pololu D36V50F6 (6.6V Step-Down Voltage Regulator) but I had a question.

Is it (easily) possible to setup a circuit to detect the input voltage like this:  (?)

If Vin = 6V Then pass voltage to MCU
If Vin > 6V Then pass voltage to regulator then to MCU

I'd like to be able to use a 6V power supply but also use a higher voltage (20V) and have it regulated.

[T3sl4co1l]

What about Vin < 6V?  Truly equal voltages are awfully rare. (Maybe pedantic, but precision is everything!)

I don't know if your specific regulator will do it, but regulators in general, simply pass input to output when below threshold, and regulate when above.

The catch is: for linear regulators, the drop is the dropout voltage, which varies by type.  It can be more-or-less arbitrarily small, given free choice of type.

For buck regulators, often the switch is a high-side N type (NPN/N-ch) supplied from a bootstrap circuit, so its driver depends on the output to be switching constantly, to obtain supply power.  These typically have a maximum duty cycle limit, which might be, say, 90% or something.  As a result, there's a dropout equal to the same fraction of the supply voltage; rather than a fixed drop (as a linear might have), it's a proportion.  So, if the limit is 90% and the regulator setpoint is 6V, an input of 6V would deliver 5.4V, and more than 6.6V input would be required to get the output under regulation.

Some types have drivers that don't require bootstrap power, though the price paid may be notable voltage drop anyway (e.g. a PNP pullup into an NPN follower, dropping about 0.8V), so again having a similar fixed drop to a linear reg.  Others use a P-type switch, or an internal charge pump or external bias supply instead of a bootstrap supply, allowing the switch to be held on statically (100% maximum duty) and thus dropping very little voltage in dropout.

Also take note of ULVO / startup voltage, or other enables or faults that may apply.  Obviously, the regulator won't operate at all, below its enable threshold.  So the transfer curve (Vout vs. Vin) will probably be zero at low voltages, then a step up to the Vout ~= Vin ramp, then as Vin passes Vreg, Vout levels off.


Another good application for this kind of regulator, is as an active current limit.  Whereas a linear current limiter device has to dissipate huge power and thus typically can only do so for a short time (say, some ms), a switching regulator can do it for very little loss and therefore a long time (100s ms, or continuously even).  Note that you need a current mode regulator of course; this doesn't work with voltage mode.  (That's fine, voltage more regulators are awful anyway.)

Tim

[tszaboo]

Sure, select one with 100% duty cycle capability.

Tim

Tim, Tim... Giving the theoretical answer for a practical question again?

[sofakng]

Thanks again for the help...

The voltage regulator was a bit expensive ($23) and this is for a one-off project.

Is it possible to use a comparator (or similar?) or something else to still use this regulator?  (ie. I just need to support <= 6V to the MCU and >= 6V to the regulator)

[T3sl4co1l]

Sure, select one with 100% duty cycle capability.

Tim

Tim, Tim... Giving the theoretical answer for a practical question again?

I feel some background is more helpful overall than a one-liner "yes" or "no", if maybe (probably?) not the most useful in the specific case; though there are still occasions for such responses.

If you have suggestions to improve reading of intent, I welcome it!

Tim

[artag]

You may still lose a little voltage as at 100% duty cycle there will be an inductor (and a fet) inline and no regulation to compensate any drop across it.

To avoid that, I guess you could have a PMOS FET bridging input to output and having the gate grounded for voltages below 6V. A depletion mode mosfet might be a more cunning way to do it with very few components (it conducts until biased off).

Getting the threshold right may be a problem. You need some tolerance to design to : if the input voltage is less than 6V, will your circuit still work ? If the bypass doesn't cut off at 6V but, say, 7V, is that a problem ?

[T3sl4co1l]

You may still lose a little voltage as at 100% duty cycle there will be an inductor (and a fet) inline and no regulation to compensate any drop across it.

To avoid that, I guess you could have a PMOS FET bridging input to output and having the gate grounded for voltages below 6V. A depletion mode mosfet might be a more cunning way to do it with very few components (it conducts until biased off).

Getting the threshold right may be a problem. You need some tolerance to design to : if the input voltage is less than 6V, will your circuit still work ? If the bypass doesn't cut off at 6V but, say, 7V, is that a problem ?

Yeah, it's an option.  It's still true what I said, that, if the regulator provides 100% duty cycle, the drop can be arbitrarily small -- simply use one strong enough (i.e. low enough Rds(on)), and a big enough inductor (low DCR).  Granted, maybe that's not the best-scaled solution for some application.  I think the biggest reason might be, if a 100%-duty part simply cannot be found; in which case one may have to resort to this.  (I'm not sure what other situations might apply, where the switch is the preferred solution -- but I wouldn't rule out the possibility.)

I greatly prefer methods that are straight-through, without interdependencies -- as you note, there are many questions about whether and when a bypass switch will function, both in terms of voltage, and time -- note for example that, if it stays on for some time after Vin rising >6V, the output will overshoot by as much.  This likely is fine for the regulator (typically it's made to deliver Vout ~ Vin, but do check the datasheet and your particular wiring of it), but may prove dangerous for the load.  Conversely, if it turns off suddenly, before the regulator is fully enabled say -- the output drops until it's able to start up and "catch" it.  Maybe this only occurs during startup, where you don't really care that the output isn't stable at first; but you also need to check that nothing else triggers it unexpectedly (e.g. transient dips in input?).

So you get all sorts of subtle race conditions and edge cases, when considering topologies like this.  It can be done -- and sometimes has to be -- but it's much harder to prove out, than just the one thing is.

Tim

[sofakng]

Thanks so much again for the help everybody.  Unfortunately I'm still just beginning to learn electronics which is why I purchased the pre-made Pololu module.

Can anybody help me find a suitable regulator on Mouser?  (6V, 5A minimum 10A better for future)

I've tried looking at LDO voltage regulators but can't find anything that supports high amperage with a 100% duty cycle...

[Manul]

For such case I would be inclined to suggest buck-boost type converter. It can step down, as well as step up voltage. No need to create bypass or search for 100% duty cycle option. But your current requirements are quite a monster, plus you want a module. Not many options I guess.

[sofakng]

Yeah, the high amperage is the problem it seems.

I'm using this to power a few servo motors and a microcontroller.  The typical current draw is 2 - 3 Amps but it can peak up to 4A.  Therefore I think 5A is sufficient but a little higher is always better to allow me to use more powerful motors, etc.

[Manul]

I'm using this to power a few servo motors and a microcontroller.  The typical current draw is 2 - 3 Amps but it can peak up to 4A.  Therefore I think 5A is sufficient but a little higher is always better to allow me to use more powerful motors, etc.

Offtopic, but I strongly suggest (at least for the future) to have separate supplies if possible. Generally it is not good practice to use same supply for MCU and big power stuff. Even if it looks ok, there is always a day when bad practice causes time and money wasted. Also the benefit of separate supplies is that it is more flexible and easier to adapt if you choose to use different motors (like higher voltage, more power).