7
\$\begingroup\$

I'm trying to develop a design which involves switching a 5V rail.

The design goals are:

  • As low cost as possible (i.e. Can't use a fancy integrated solution)
  • Relatively low current handling ( typically only a few hundred mA)
  • Trying to keep the overall solution in the 10's of cent
  • Low drop off on the the 5V when enabled ( > 4.8V )
  • Protection from reverse voltage when powered off

The problem I'm having is with the last point.

I've used a p-channel MOSFET as the switch, the gate of which is controlled from an op-amp. Everything is fine except for the "reverse voltage" part. The problem is that the body diode of the FET is going to conduct if an external power source is present.

I'm looking for a low cost solution to this problem, but I'm having a mental block trying to solve it without using some expensive IC.

Ideally I could just use a diode in series but the the drop-off would kill me there. I've seen lots of references to using back to back FETs to solve this type of problem but I haven't been able to figure out a configuration which works.

The following is an overview of what I'm describing as it may be easier to visualise.

enter image description here

Update: Based on the feedback from @endolith I now see how the back to back system is configured which I believe is as follows:

enter image description here

Operation as follows:

  • When gate low
  • First FET GS voltage causes FET to be on
  • This places voltage on second FET Drain
  • Second FET has no GS voltage but the body diode conducts pulling Source high and causing GS voltage, which switches on the second FET

This begs the follow-on question, how much current can the body diode take? Is it safe to use FETs in this manner? (assuming I have understood @endolith correctly)

\$\endgroup\$
  • \$\begingroup\$ Post the image on imgur.com and post a link to it. \$\endgroup\$ – endolith Sep 7 '11 at 20:54
3
\$\begingroup\$

Basic P-MOS switching circuit:

http://www.physics.udel.edu/~watson/scen103/mos5.gif

The problem is that the body diode of the FET is going to conduct if an external power source is present.

You can connect two FETs in series with their body diodes opposite each other so they don't conduct when off, and get low drop-out when both are on. Here's a crappy illustration of the concept from a product:

enter image description here

\$\endgroup\$
1
\$\begingroup\$

Maybe a NCP380 would work for your circuit? NCP382 may be a better choice if you have more then 1 channel and you do not need adjustable current limits.

The wholesale (100 pcs.) price is below $0.5.

\$\endgroup\$
  • \$\begingroup\$ Actually the IC you propose is probably perfect for my application. Leaving question open for now as I think the back to back FET suggestion might also be viable, would like to hear what people think of that. \$\endgroup\$ – Sean Murphy Sep 8 '11 at 21:03
  • \$\begingroup\$ Back to back FET is probably also going to work but with the IC you get over-current protection and a MOSFET driver for free. If this is important depends (of course) on the application. \$\endgroup\$ – jpc Sep 9 '11 at 14:51
1
\$\begingroup\$

If you can handle a very small drop then maybe you could use a PNP instead, as they can block both ways.
Or a reed relay could be another option if the current you are switching is low.
The back to back MOSFET could be accomplished by using an optocoupler to drive the gates, but there are plenty of cheap/simple ICs that will do the job easily, I wouldn't rule these out completely.

Here is a useful app note from Zetex that mentions the reverse blocking capability of a bipolar transistor when base is high impedance (page 5, point 7)

\$\endgroup\$
  • \$\begingroup\$ Will the collector of a PNP actually present a high impedance to 5V source if both the Base and Emitter are low? In any case I think you are right about the IC solutions. The suggestion by @jpc is probably just what I need. \$\endgroup\$ – Sean Murphy Sep 8 '11 at 20:30
  • \$\begingroup\$ No, you would need to let the base float (or drive to collector voltage) by setting the driving pin to high impedance, otherwise you forward bias the collector base junction. I wouldn't use this method to guarantee solid protection though, (e.g. reverse current is expected regularly as part of the normal working conditions) I think the back to back PMOS using body diode would probably work okay for a low current solution. Try them both out on a breadboard and see how they hold up under expected worst conditions... \$\endgroup\$ – Oli Glaser Sep 8 '11 at 23:50
  • \$\begingroup\$ ...If this is an important project though, I would really consider spending the extra few cents on a IC switch or relay for peace of mind, rather than try something "new" out. \$\endgroup\$ – Oli Glaser Sep 8 '11 at 23:50
  • \$\begingroup\$ @Sean - updated answer with app note discussing bipolar switching which mentions the reverse blocking. \$\endgroup\$ – Oli Glaser Sep 9 '11 at 0:18
0
\$\begingroup\$

Dude, Why do you have to make so complex !.

Here is the easy way to do it. Connect the drain of your PMOS to power supply end and source to circuit end. Connect the gate to ground. The PMOS acts as a diode as long as your supply voltage is greater than the threshold voltage of the PMOS.

For more info visit..

http://www.wa0itp.com/revpro.html

\$\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.