How to choose a transistor used as a switch ?

Hello,

I have my Atmega328p-pu running at 8Mhz with 2 AA batteries (~2.4V).
Additionally I need to power a rf433 module with 5V. I bought a DC-DC 5V step up module but as it's consume too much current I would like to use it only when transmission are performed (that means powering the transmitter with the DC-DC step up and other stuff like atmega and sensor with the battery). I can't plug the step-up module directly to an output pin of the atmega because the load is too high for the chip (I have a drop of voltage if I do it).

So I'm thinking about plugging the DC-DC step up in // to the battery, and using a transistor as a switch in order to drive it. As I'm newbie with it, my question is how should I choose a good transistor ?

Output pin low is 0V, high is ~2.4V,
The DC-DC step up + transmitter will probably use ~20mA,
My goal is to have a low consumption device, I dunno if some transistor are more efficient.

Any advises on which transistor I should use ?

You didn't say what kind of building experience you have so I'll assume you don't want something so tiny it's difficult to solder. Your best solution is to use a Logic Level, P-Channel MOSFET with RDSon as low as you can get it. Generally the higher current the device, the lower the "on" resistance.

Connect your Source to Vbat, your Drain to the positive input of your converter and your gate to the Arduino with a series resistor of about 200 ohms. Since it's a P-channel device, the Arduino output has to be LOW to turn on the FET.

http://www.ebay.com/itm/Siliconix-Temic-SMD10P05-P-Channel-Power-MOSFET-50V-10A-DPAK-5pcs-/161096442263?pt=LH_DefaultDomain_0&hash=item258218a197

At $1 each, these are pretty inexpensive and still easy to solder though you might struggle a bit soldering to the TAB. In general, you can use most any P-Channel FET provided it's specified "logic level".

Maybe fix your link

done

LarryD:
Maybe fix your link

Two things. To post a link, highlight a word to which you wish to attach the link, click the "Insert Hyperlink" icon () add an "=" after the first "url"and paste your actual link URL immediately after this.

Secondly, when posting a link to eBay, remove the question mark and everything after.

Thus the link would be:

At $1 each, these are pretty inexpensive

At $1 each, [url=http://www.ebay.com/itm/Siliconix-Temic-SMD10P05-P-Channel-Power-MOSFET-50V-10A-DPAK-5pcs-/161096442263]these[/url] are pretty inexpensive

Thank you for the answer.

I understand that I need a Logic-Level. How can I know the "level" of this transistor (to be sure that my voltage is enough to "trigger" the transistor) ? I don't see this information into the datasheet given in your link, or perhaps it's depending of the G and D?

with a PNP I'll have to put the arduino output to HIGH when I don't want to power the transmitter. Maintaining a pin to HIGH seems to consume current, and as my Arduino will spend most of his time in sleep (sleep 10min then wakeup for some seconds), is it a good idea to go to a NPN ?

yoshop:
I understand that I need a Logic-Level. How can I know the "level" of this transistor (to be sure that my voltage is enough to "trigger" the transistor) ? I don't see this information into the datasheet given in your link, or perhaps it's depending of the G and D?

First chart on page 6-75. Shows that at 5V on the gate, it will pass one amp with a drain-to-source voltage of 1V. In this area is it more-or-less resistive; the voltage drop will be directly proportional to current.

yoshop:
Maintaining a pin to HIGH seems to consume current,

Absolutely not. Unless you have a resistor (resistance) to a lower voltage or ground, there is no reason for it to consume current.

I don't really understand this chart. :~

My battery is ~2.4v
My command is 0v on LOW and between 1.5v and 2.4v on HIGH

So it means that Vgs will be -2.4v when LOW and between -0v and -0.9v when HIGH ?
Regarding the doc, Vth is between -2v and -4v so i'll never be over the threshold and the transistor will never be saturated ?

In the chart I don't understand the Vds because in my mind if Vds=0, it means that potential on S and D are the same, so the transistor is "on" and current is beeing delivered. But in chart, when Vds is 0, current is 0 too ... i'm lost :cold_sweat:

Additionally, what is the use of the serie resistor of 200 ohms on the Gate given in the first answer ?

That p-channel MOSFET needs -10V Vgs and is wholely inappropriate for 2.4V use
as you have pointed out.

The simplest solution is use a PNP high-side switch, no problem with 2.4V drive,
available in through-hole package. The STX790A or ZTX550 are good high-performance
PNP devices although many will adequate for such low currents. Just use 1k base
resistor and pull this LOW to turn on the PNP (emitter to Vcc, collector to load).

yoshop:
I don't really understand this chart.

I'm not really surprised, it isn't easy.

yoshop:
My battery is ~2.4v
...
Regarding the doc, Vth is between -2v and -4v so I'll never be over the threshold and the transistor will never be saturated ?

Ah, yes, I missed that, I was just addressing the interpretation of the datasheet. You are wanting to use a mere 2.4V to control it. You will not find a FET suitable for that, you will need an "ordinary" junction transistor as MarkT points out (and I should have 8)).

yoshop:
In the chart I don't understand the Vds because in my mind if Vds=0, it means that potential on S and D are the same, so the transistor is "on" and current is being delivered. But in chart, when Vds is 0, current is 0 too ... I'm lost.

As I explained, the interpretation of the chart is that in that area, the device is essentially resistive, current is proportional to voltage. The only way you can have zero voltage is if no current is flowing. You would have to accept that there is some resistance in the "switch", but as long as it is acceptably low, no matter. A BJT ("ordinary") is no different.

yoshop:
Additionally, what is the use of the series resistor of 200 ohms on the Gate given in the first answer ?

It is to limit the current that the Arduino would have to provide to charge up the gate capacitance. Expecting it to charge this capacitance instantly is unreasonable, representing a transient overload of the Arduino output.

Thank you very much for all these clarifications :slight_smile:

I'll go look for the ones pointed by MarkT :smiley:

That p-channel MOSFET needs -10V Vgs and is wholely inappropriate for 2.4V use
as you have pointed out.

Are we looking at the same data sheet? I see VGSth at 2 typical 4 max. True, from the curves it might be pretty linear there but the IDS is nothing

VGSth is threshold. At a drain current of 250uA.

rmetzner49:

That p-channel MOSFET needs -10V Vgs and is wholely inappropriate for 2.4V use
as you have pointed out.

Are we looking at the same data sheet? I see VGSth at 2 typical 4 max. True, from the curves it might be pretty linear there but the IDS is nothing

http://rfbayinc.com/upload/files/data_sheet/smd10p05.pdf

The threshold voltage is nothing to do with the MOSFET being switched on, it is
all to do with it being effective OFF, its the gate voltage at which a tiny current
is just starting to flow, and marks the Vgs you must go below to guarantee turn-off,
here below 2V. If you want to guarantee turn-on you need to go the the Vgs
specified in the Rds(on) specification.

There is another voltage, the plateau voltage, at which most of the actual switching
happens, during which the gate/channel capacitance charges up, which is between
the threshold and on voltages. These three voltages, threshold, plateau and on
tend to be in the rough relationship 1:2:4, here 2.5V/5V/10V in fact.

A 5V logic level MOSFET will have a threshold voltage spec of typically "0.5 -- 1.0V".

Oh, yes you can get MOSFETs which work at 3.3V or 1.8V logic levels, but they are
all surface mount, since new devices are surface mount only in practice.

The Rds(on)/Vgs figures in the datasheet are guarantees for all devices across the
full temperature range specified in the table - remember any graphs in a datasheet
are usually "typical values" and are specifically not a guarantee of performance.

In particular the threshold and plateau voltages display large variations between
devices, which is one reason the on-voltage is significantly larger than the plateau voltage
as seen in any graph. Its also one of the reasons you need gate resistors when
paralleling MOSFETs.

MarkT:
Oh, yes you can get MOSFETs which work at 3.3V or 1.8V logic levels, but they are
all surface mount, since new devices are surface mount only in practice.

Clearly by definition you can get MOSFETs working at this voltage - because that is exactly what the millions in the current CPU chips have.

This looks interesting:

Avalable in TO-220 package
https://www.fairchildsemi.com/datasheets/ND/NDB6020P.pdf

Paul__B:

MarkT:
Oh, yes you can get MOSFETs which work at 3.3V or 1.8V logic levels, but they are
all surface mount, since new devices are surface mount only in practice.

Clearly by definition you can get MOSFETs working at this voltage - because that is exactly what the millions in the current CPU chips have.

Power MOSFETs with vertical current flow and body-diodes, not signal MOSFETs

That is a nice P-Channel MOSFET, and Digikey seems to have them. $1.49 if you buy 10. I knew these were out there, I just didn't look hard enough given my first post.