3
\$\begingroup\$

When tackling with the very common need of toggling a current with an output from a microcontroller or similar chip, solutions have varied over time.

The oldest solution is to just use a BJT. This works reasonably well, but one has to verify that the gain is suitable for the use and requires the microcontroller to constantly drive the transistor when switched "on". The current may be neglible, or it may not.

The next solution is to use a MOSFET. Modern MOSFETs seem to switch nicely even with 3.3V voltages and have very low on resistances. Gain is no longer an issue, and no need to constantly drive the MOSFET but the microcontroller still has to drive the MOSFET when it is switching. These currents are sometimes pretty high, so the MOSFET needs to be selected carefully.

Then there are Analog Switches (bilateral switch) which require separate voltage inputs but which seem to offer even nicer characteristics: very fast switching speed, very low power usage, very low drive currents, very low on resistance, etc. And obviously having a SPDT switch is nice in some situations.

So, I am asking, what is the modern world ideal component for these sort of situations? Is analog switch the holy grail, or is there something even more suitable? Or are MOSFETs actually better if you just pick the right one?

The usage situations I am thinking of are just switching low voltage low current lines with no special requirements - the sort of solutions that would've gotten a 2N2222 transistor years back.

The wishes I would have for an ideal component:

  • Supply voltage either 3.3V or 5V
  • Input voltage suitable for LVTTL, TTL, CMOS inputs
  • Low on-resistance (<1 ohm?)
  • Fast switching times (<100 ns?)
  • Bidirectional current flow (although usually not really necessary)
  • Safe for the microcontroller (no back-current?)
  • No need for high current operation, 100-200mA is probably fine.
  • No need for high isolation, although it would be nice if the switched voltage could be 12V
  • Cheap-ish, let's say sub-$1
  • Small SMT package

In general, something that could be used to replace 2N2222 or say BSS138 in most simple situations.


To quote wikipedia on 2N7000 MOSFET:

The 2N7000 is a widely available and popular part, often recommended as useful and common components to have around for hobbyist use

The 2N7000 has been referred to as a "FETlington" and as an "absolutely ideal hacker part."

I guess what I am looking the analog switch equivalent of this, if such a thing still holds any benefit over just using something like that.

\$\endgroup\$
  • \$\begingroup\$ It would seem that you have already answered your own question :) MOSFETs are most commonly used when higher currents are used, rather than data signal switching. You should be able to find an analog switch that suits your spec without too much trouble. Perhaps Texas TS5A6442? ...took me 1 minute of searching to find. \$\endgroup\$ – Lundin Dec 11 '12 at 12:23
  • \$\begingroup\$ Yes, I can find components fitting the specs by google, but I am interested in two things: 1) is there something even better? 2) is there some component which would be common enough to be enshrined to a similar status as 2N2222 or some of the other common "use everywhere" components. \$\endgroup\$ – Nakedible Dec 11 '12 at 12:26
  • 3
    \$\begingroup\$ You do realize that analog switches are not really a different category than MOSFETs, right? Take a look at an analog switch datasheet and consult the internal schematics, and you'll be able to construct your very own analog switch from discrete components! \$\endgroup\$ – Jonny B Good Dec 11 '12 at 13:20
  • 1
    \$\begingroup\$ The closest you will come to a "carefree" device is a solid state relay, in that the intent is that you can simply read the limitations off the data sheet. \$\endgroup\$ – Chris Stratton Dec 11 '12 at 15:41
  • 1
    \$\begingroup\$ @trygvis A couple of really inexpensive, low Rds(on) MOSFETs that have become my jellybean parts for 3.3 and 5 Volt designs are the IRLML2502 N-channel MOSFET, typically under 10 cents each, and the SI4562DY N- and P-Channel 2.5-V (G-S) dual MOSFET, typically under 40 cents each. Those might add to your repertoire of hacker parts. \$\endgroup\$ – Anindo Ghosh Jul 15 '13 at 14:37
6
\$\begingroup\$

No switch is perfect. Switch selection is based on the load that you are going to handle. No load has zero inductance or zero capacitance. Hence there will be surges and kick backs. It should be handled right - if not the circuit will fail under certain inputs.

Now-a-days you could easily find FETs with very low RDS ON (even 2.0 milli Ohms is not uncommon. Typically automotive industry pays pittance for those parts - but open market pricing could be expensive. I have never tried get those outside, but I guess even if it is triple it is still within your target). But once the RDS on is low, if you ever drive it in linear mode, it could easily smoke.

\$\endgroup\$
2
\$\begingroup\$

Taking the comments into account, the answer to your question is: NO.

There is nothing better.

There is nothing better once you eliminate analog switches (two complementary MOSFET's in a package with control and biasing elements), discrete MOSFET's, discrete "BNT" (sic), and IC power switches (ex. TI TPS series).

Most SSR's don't operate at your low voltages or target your low currents.

There is nothing left in the human inventory to answer your question. =(

\$\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.