I have an autopilot unit on a boat. It carefully checks the nominal-"12V" supply voltage before driving the servos, making sure that it's not too low (you might be running down the batteries by accident) and not too high (you don't want to apply too-high voltage to a motor that's not designed for it).

As it happens, my boat's "smart regulator" sometimes charges the batteries at a voltage that the autopilot deems is "too high" by a few tenths of a volt. Let's say the autopilot cuts out at 15v, but the regulator sometimes charges at 15.2v. Most of the time, the (deep cycle) battery voltage on the boat is between 11v and 13.5v, and everything works fine.

I'd like to make a one-sided regulator to put in front of the autopilot, one that cuts any voltage above 15V down to just 15V, but below 15V does nothing at all (or at least has minimal loss).

Peak current for the autopilot is about 5A, but maybe it's safer to say 10A. So we're talking about dissipating 0.2V at 10 A = 2watts.

I guess I could put a resistor in the B+ wire to limit current, and a zener diode between the two power leads for the autopilot, but that resistor would still be heating up (with the full 11-13.5V) the rest of the time. Surely there's a better way, which someone here can suggest. I feel as if this is the sort of thing where a reasonable solution should involve about 3 components.

To help bound the design limits, let's assume that the B+ to ground voltage will never exceed 18V.

Any and all suggestions appreciated.

  • \$\begingroup\$ The autopilot unit is not configurable? Maybe you can change the threshold. \$\endgroup\$ Oct 15, 2020 at 17:22
  • 1
    \$\begingroup\$ Can you adjust the alternator regulator to keep the charging voltage below 14.5 Volts or so? \$\endgroup\$ Oct 15, 2020 at 17:27
  • \$\begingroup\$ Both good questions. Adjusting the autopilot means making a new RaspberryPi kernel or something, which is beyond my skills. Adjusting the regulator...might be possible, but the higher-voltage fast-charge mode is actually a GOOD thing that I'd rather not hobble. \$\endgroup\$
    – John
    Oct 15, 2020 at 17:38
  • \$\begingroup\$ "Let's say the autopilot cuts out at 15v, but the regulator sometimes charges at 15.2v" - are these the actual voltages, or just a guess? What is the minimum voltage the autopilot accepts? \$\endgroup\$ Oct 15, 2020 at 18:00
  • \$\begingroup\$ They're just a guess --- but they're right within a few tenths of a volt, and if you provide a circuit sketch, I can adjust the parameters enough to make it work, I'll bet. Why don't I have the exact values? Well, the problem happened just this morning, when I was alone on the boat, and without the autopilot working I didn't want to go below and dig around to find the voltmeter, etc. The overvoltage amount might be as much as a half-volt, I suppose, but I doubt it, given the symptoms I saw. \$\endgroup\$
    – John
    Oct 15, 2020 at 18:19

1 Answer 1


Here's the simplest thing I could come up with quickly off the top of my head. It uses the requested 3 active components, but there's a bunch of passives including the not shown decoupling caps.

When the input voltage is below the setpoint (set by voltage reference/zener D2) the comparator output is low, and the PFET is on, shorting out the dropping resistor.

When the input voltage rises above the setpoint, the comparator output goes high turning the PFET off, and the dropping resistor drops the appropriate amount of voltage necessary to keep the load working.

R6 provides some hysteresis to avoid oscillation around the trip point.

So you're only dissipating significant power when the input voltage rises above whatever you set with the reference voltage.

Component values are for example only, adjust to meet your requirements.

You need:

A comparator (probably with open-collector output) that can operate from the full range of the input voltage. Maybe TLV1701.

A PFET with an RDSon much less than the dropping resistor you want to use, and Vgs max high enough for your max voltage.

A reference that could be a Zener (A 5.6V zener has near zero tempco) or a shunt reference like a TL431. You would likely want a 0.1 uF - 1 uF cap decoupling the reference too.


simulate this circuit – Schematic created using CircuitLab

  • \$\begingroup\$ Thanks. This is the first time I've ever heard of a PFET. Cool! I also like the idea of shorting out the dropping resistor --- that's the part of the story I couldn't get figured out at all. \$\endgroup\$
    – John
    Oct 15, 2020 at 19:09
  • \$\begingroup\$ There's a diode drawn in above the PFET...or maybe it's part of the PFET. Can you explain it, please? \$\endgroup\$
    – John
    Oct 15, 2020 at 19:13
  • 1
    \$\begingroup\$ @John it's part of the PFET. It's called the "body diode". I drew it explicitly just for clarity. A PFET is also called a P-channel MOSFET. \$\endgroup\$
    – John D
    Oct 15, 2020 at 19:18
  • \$\begingroup\$ Note that you can't drop more than maybe 0.8V with this scheme, because the body diode of the FET will conduct at that point clamping the available drop. But it sounds like you don't need that much drop anyway. \$\endgroup\$
    – John D
    Oct 15, 2020 at 19:33
  • \$\begingroup\$ Right...if the Vin gets to be 18V for some reason, I'm HAPPY to have the servos cut out, because something else is very wrong! But thanks for the warning. \$\endgroup\$
    – John
    Oct 15, 2020 at 19:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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