18
\$\begingroup\$

What are the popular equivalents to the 2n7000, for 3.3 Volt MCU circuits to switch higher voltage devices?

Background: For 5 Volt microcontroller based prototypes and experiments, my go-to solution for low frequency, low side switching of 50-200 mA currents or 10-40 Volt devices driven by a digital / PWM pin, has been the dirt cheap ($0.05 retail here in India) and ubiquitously available 2n7000 MOSFET.

The best aspect of using this MOSFET is, I have made a bunch of little building block PCBs with a 100 Ohm gate resistor and 10 k gate pull-down, and simply plug them in to almost anything that isn't high frequency or high load. It just works, and is almost bulletproof. If I could find any 4 x 2n7000 array parts locally, I'm sure I would make 4-channel building blocks too.

When working with 3.3 Volt MCUs and boards (e.g. TI MSP430 Launchpad) what is the equivalent easily available, robust switching solution, if any, that one turns to for non-critical quick prototyping?

I currently end up using either 2n7002, whch does not turn on hard enough, or various IRL/IRLZ parts, though they cost as much as 10-20 times as much. IRL/LZ are often unavailable off-the-shelf at the "electronic components market street" where I pick up random parts for my personal component shelves when I am not working to a BOM or a plan.

The AO3422 suggested in comments to this question simply is not available locally in retail.

I want to avoid BJTs for this purpose as they tend to run hotter than MOSFETs, and I create enough magic-smoke gizmos anyway.

I know there isn't necessarily one correct answer to this, but good advice would, I am sure, benefit many people jumping into 3.3 Volt devices.

\$\endgroup\$
  • 1
    \$\begingroup\$ At low drive voltages, BJTs can be good solutions. Your objection that they "run hotter" doesn't make sense since you said this is for currents only up to 200 mA. With 200 mV or so saturation, you wouldn't even notice a SOT-23 getting warm. Even with 500 mV C-E drop (well above saturation at 200 mA), the dissipation is only 100 mW, again fine for even a SOT-23 package. \$\endgroup\$ – Olin Lathrop Nov 26 '12 at 14:49
  • 2
    \$\begingroup\$ Use math instead of superstition. According to the Fairchild 2N7000 datasheet, the Rdson with 4.5 V on the gate and 75 mA channel current can be 5.3 Ohms, and it goes up with higher current. Even using just the 5.3 Ohms at 200 mA means it can have over 1 V drop. A saturated NPN is going to have a lot less than that. Your board may not be supplying sufficient base current. \$\endgroup\$ – Olin Lathrop Nov 26 '12 at 16:59
  • 1
    \$\begingroup\$ It's interesting that the knee-cappers got to this question just under 3 years after it was asked and long long after the several useful answers to a question that many people would find useful. This question meets the prime site directive well in all but letter-of-the-law-that-kills , so I guess it had to go. Not that beeing gone-g 3 years on made any difference. Ain't life interesting. \$\endgroup\$ – Russell McMahon Jun 21 at 5:17
  • 1
    \$\begingroup\$ @PeterJ IMHO the discussin alone makes this worthwhile for people asking such questions. HOWEVER.|| DMN4800 - Sunnyskyguy - $0.36/10 Digikey 10000+ in stock. || 2N7002PW $0.22/10 87000+ stock Digikey. || The ROhm parts are gone. || A better answer, which I cannot give because the question is closed, = ... \$\endgroup\$ – Russell McMahon Jul 26 at 1:33
  • 1
    \$\begingroup\$ ... @Peterj : A good solution is to use the "selector guide" from companies such as Digikey with a search set to parameters of interest. eg This is logic level drive FETS up to 1.8V Vgs quantity10 in stock Digikey arranged in ascending $ cost.|| Many of those knock the 2N7000 into a cocked hat, as it were. || \$\endgroup\$ – Russell McMahon Jul 26 at 1:41
6
\$\begingroup\$

The UM6K1N / 2SK3018 from Rohm work very well with 3.3V circuits (max 1.5V \$V_{GS(th)}\$) and are rated 100mA \$I_D\$ and 30V \$V_{DS}\$. They also have gate protection clamping diodes.

The only downside is that they're surface-mount only - you'll need to spin a small PCB to use them.

\$\endgroup\$
  • \$\begingroup\$ Surface mount is not a problem at manageable pitches. Thanks for this suggestion. I'll have to explore local retail availability, but that's an offline problem. \$\endgroup\$ – Anindo Ghosh Nov 26 '12 at 16:18
4
\$\begingroup\$

My favorite for 3.3 Volt and even 2.5 Volt circuits is SI4562, N and P MOSFETs in a single package, with much lower Rdson for both N and P channels than the 2n7000. They are not expensive at around $0.37 per piece on eBay. Even if you only use the N or the P, it is a good deal. It is still much more expensive than 2n7000 but even then it is a much better MOSFET than 2n7000. Current is 4 to 6 amperes for each channel, not milliamperes like 2n7000. I made a hbridge using it for driving a small motor and it was not even warm while using BJT it was getting very hot.

RDSON for P channel is 0.05 ohms and for N channel is 0.035 ohms at VGS value of 2.5 volts.

\$\endgroup\$
  • \$\begingroup\$ Thanks. Very expensive compared to 2n7000, but if I needed both N and P channels, this looks like a terrific option. Added to my list. \$\endgroup\$ – Anindo Ghosh Dec 6 '12 at 13:26
2
\$\begingroup\$

My suggestion is Manufacturer Diodes Inc DMN4800LSS-13 Description MOSFET N-CH 30V 9A 8SOP

in 1K qty approx. 17 cents

enter image description here

\$\endgroup\$
  • \$\begingroup\$ Thanks, added this one to my list too... Is the Vgs graph knee a bit unusual on this one, not as sharp as some of the others? Or is that just because of the scale used? \$\endgroup\$ – Anindo Ghosh Nov 26 '12 at 17:57
  • \$\begingroup\$ just the scale. \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Nov 26 '12 at 19:40
2
\$\begingroup\$

What about the 2N7002PW from NXP? Basically the same device at the same price point in a smaller package (SC-70), but with a lower RDSon (1 Ohm vs 2.8 Ohm) and should turn on harder according the graphs.

\$\endgroup\$
0
\$\begingroup\$

I usually just stick to 2N7000 and a bigger drain load resistor (10k or more) and it is pretty reliable as I have found out. I used Raspberry Pi which virtually cannot source or sink any current (7mA max or the $35 board is toast), so my designs are usually peppered with 2N7000s and with a higher drain load, and downstream CMOS parts are pretty reliable.

\$\endgroup\$

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