Transistor biasing

[Pezimyst]

Hello everyone. I have an embarrassing "newbie" hardware question about transistor biasing with a PIC. When using a common NPN transistor as just a "switch" for higher currents than the PIC itself can handle, is there a preferred biasing method? Most examples & tutorials I've seen simply use a base resistor on the PIC outpt pin to drive into saturation, throwing the switch on hard and that's seemingly what I'm after. But I've seen two voltage divider resistors used for biasing, or occasionally some other more elaborate scheme. Is consistency between transistors and ambient (?) temperature drift such a problem that using a base resistor alone is a poor choice? If it has any bearing, my application requires making a few hundred devices which will be exposed to climates 10 - 35 C (50 - 95 F). Maybe a bit higher if out in the sun.

[Ttelmah]

Remember that a PIC pin _always_ wakes up as an input. As such very high impedance.
A twin resistor network, ensures the transistor is 'off' when this happens. Also ensures that if the processor gets into an indeterminate state, the transistor defaults to 'off'.

How high is your current?.
What is the nature of the load (resistive, inductive etc.)?.

If there is any inductance present, then far more important, is what happens when the transistor switches 'off'. Any inductance _will_ result the terminal between the transistor and the load having a rising voltage, that will go up until something stops it. The transistor, or another component breaking down. This is why off the shelf transistor drivers will have trap diodes to control this.

If the current is more than a few hundred mA, then the beta of the transistor, may mean that a darlington is needed.

Best Wishes

[temtronic]

also...

depending on the load, a FET might be a better choice than a 'garden variety' NPN transistor

cheap...cost is always a factor

easy to use....especially the 'logic level' types..

tell us the load and we can better advise.

remember to 'overkill' the design ! Think like sending a satellite up,(now that there's no shuttle !).Spending a few extra pennies now will save you big bucks if( when) ANY units come back for service.

hth
jay

[Ttelmah]

Worth also saying, that the cost of transistor arrays is usually less than the discrete parts.
Chips like the ULN2003D1013TR, give you seven transistors, with the resistors, and clamp diodes all in a SO16 package, and in even quite small quantities are under $0.30. Far cheaper to assemble and layout, than seven discrete transistors, fourteen resistors, and diodes....

Best Wishes

[Pezimyst]

Thanks very much for the info. The loads will be mostly small. Less than 100 mA on each pin (and only a few pins). Common things like an array of LEDs, etc. One pin will however have a nearly 1A load on it. But yes, I've been looking at a FET for that one.

[asmboy]

If using an N chan FET buffer - low side - source=gnd
always put a 20k resistor to ground from the gate pin/pic drive pin.

For an emitter follower NPN transistor pull UP -hi side collector +Vdd-
same deal -20k base to ground.

For a low side transistor switch - emitter ground Vss -
a divider of 2,
10k resistors with base tied to the split, and low side to ground - will work.

5.1k resistors for a LOW Hfe type part.

All 3 of these methods keep the load side OFF during PIC power up.

[asmallri]

[asmboy]

[gpsmikey]

The problem with MOSFETS is trying to drive the buggers all the way on if you decide to run your PIC at 3.3v instead of 5v. Lots of 5v MOSFETS out there, but the 3.3v stuff is a bit tougher. The earlier posts are correct though, you need worst case Hfe considerations to make sure you drive the bipolar transistor into saturation (unless you want extra heating that is).

mikey
_________________
mikey
-- you can't have too many gadgets or too much disk space !
old engineering saying: 1+1 = 3 for sufficiently large values of 1 or small values of 3

[asmallri]

[quote="asmboy"]

[Ttelmah]

The 'always assume the worst' lesson, is one that everyone here who designs anything, _will_ have had to learn.

It is so easy to have a prototype that apparently works fine, and then find that production units have chips/transistors/capacitors etc., that are closer to the 'worse case'. This is true with things like watchdog timings, brownout voltages, input switching levels, tolerances everywhere.

Listen to Asmalllri's comment on this, and make sure the design covers both ends of quoted specifications.

Best Wishes