Deenergize rc servo - high side or low side switch?

[snoozerman]

Hello,
I'am designing a small watering system for my pot plants. A 3,3V controller will control a pump and a diverter valve. The valve uses a servo to divert the pumpflow to the different pots. Naturally, the pump won't turn on very often, so I wan't to deenergize the rc-servo when the pump is not turned on in order to save energy and minimize wear. I have searched the web but haven't really seen so many final results circuits.

My question to the forum is asking for advice. I have two ideas described below. Which one is likely preferred (and why)? That is, if they work at all of course ???

As far as I've understood by searching the web, behaviour of a servo is undefined when the signal is removed. For instance it may depower, hold in last position or put itself in some predefined position. So simply removing the signal by e.g. servo.detach() won't solve my problem.

I have come up with two designs, both attached as images. One is a high side switch using a p-channel mosfet and the other a low side switch with a n-channel mosfet. A 6V supply is used to power the servo. Two GPIOs on the 3,3V controller turns on the main power (gpio_1) and outputs the servo position signal (gpio_2).

The low side switch is nice because I can control the n-channel mosfet directly from the 3,3V gpio_1. However, when switching it off, the servo signal needs to be "taken care of" because I suppose it will end up equal to the 6V supply. Instead, I've used a npn transistor trying to solve that.

The situation for the high side switch version is rather the opposite. I can connect the servo signal directly to the gpio_2 (3,3 V should be enough for the logic level?) but need a npn transistor to turn on/off the p-channel mosfet. There is also a zener for overvolatge protection, but that is maybe not necessary.

Most greatful if there are any words of visdom from a forum expert!

[Paul_B]

OK, so your low side switch is a worry for a start - particularly as you will be reverse-biasing both of the upper transistor junctions when the lower is turned off - that certainly can be a problem, though you could include a signal diode in series with the base.

Having whole sub-assemblies floating "high" when turned off is in general, a bad idea. All sorts of things can happen!

This looks better. I don't see any point in a Zener. Will a servo run at 3.3 V logic? I would first see how well it runs when you turn the voltage down to 3.0 V or 2.8. Test it first.

Oh yes. With no signal, most common servos will simply lose hold - they will resist movement to some mechanical degree but generally can be forcibly turned.

[snoozerman]

Aha. Wasn't sure but I thought the npn-transistor Q1 in the low side switch design would be able to handle the reverse bias (emitter>base voltage). I imagined that the base-emitter connection could be thought of as a "diode". Anyhow, would a n-channel mosfet work better from that point of view? A nice side effect would then be to use e.g. a SO-8 package with dual n-mosfets. Simplifies the pcb a bit.
Still, the low side switch may be a bad idea if "all sorts of things can happen" . Out of curiosity and own understanding, in this particular design, are there any obvious risks? The servo and the pcb will be very close to each other so there should be little risk to accidently pinch/ground a long cable or something like that. My idea was that, from the servos point of view, there should be no difference.

Regarding the high-side switch, it struck my mind that, it is probably not a good idea to use the 3,0V zener D1 due to fact that the output voltage of the gpio_2 is 3,3V.
B.t.w. replacing the npn Q1 with a n-channel mosfet should be possible here also, wouldn't it?

I actually found some precious info about rc servo signals at
http://www.futabarc.com/faq/faq-bots.html#q572 and
http://hitecrcd.com/files/Servomanual.pdf
At least Hitec seem to claim that the pulse should be 3-5V peak to peak (which I interpret as a pulse 0-3 to 0-5 volts). Futaba doesn't state it that clear although they do state their recievers uses 3,3V internally. Anyhow, if 5V is max pulse voltage, that would be another issue with the low side switch design; the servo_sig needs to be clamped to < 5V.

Thank's for your input, it is invaluable for my learning!