>> With say 0.5 mA drive you need a transistor current gain (Beta) of
>> 80/.5 = 160 which is  achievable with many of the more capable 
>> small
>> transistors. I'd use BC337 here but there are many others.

Pre-script: It's surprising how much needs to be thought about with 
such a very very simple arrangement to actually get it right :-(.

I'd too would tend to use a MOSFET if doing this in volume, but more 
for low effective Vsat and for the certainty of available drive 
reasons and not because it has an internal diode. A MOSFET has some 
dangers, such as needing to ensure that it's Vgsmin is not too GOOD so 
that it is not turned on accidentally. This depends on how well the 
parallel port driver saturates when low and this would need to be 
checked as part of what starts out looking like a trivial design. A 
MOSFET can also have very nasty leakage current at elevated 
temperatures but that is not an issue here with a relay load.

> Wouldn't you need an extra diode with a bipolar transistor that you 
> don't
> need with a MOSFET (built-in)? That's the reason I suggested a 
> MOSFET.

Essentially, no, but you probably do :-).

ie         The MOSFET diode won't help spike suppression. The inherent 
body diode of the MOSFET is in the wrong place and of the wrong 
polarity to act as a spike suppressor in this application and will be 
stressed by the spike just as a bipolar transistor will be. (The 
bipolar and the FET proper are in forward blocking mode and the diode 
is in its reverse blocking mode during the inductive spike). Without 
an external diode or similar both would generate a reactive transient 
on turn off which would quite probably require some form of 
suppressor - typically either a reverse polarity diode *across the 
relay coil* or a resistor.

One could use a MOSFET which is specified for avalanche breakdown (at 
slightly above its maximum rated voltage) and thus remove the need for 
an external suppression device.  This is more or less equivalent to 
adding a zener diode as below.

Transient current is initially the relay current at turnoff time and 
energy to be dissipated is 0.5 x Lrelay coil x i^2. A diode clamps the 
turn off transient at 1 diode drop above supply and this can greatly 
increase the turn off time as the energy stored in the relay coil is 
dissipated at low voltage (typically 0.6 x Irelay decaying 
exponentially as current drops). If turn off time considerations are 
an issue then a series resistor will allow faster turnoff as it will 
let the transient rise to V = I.R above rail but will lead to much 
faster turnoff. IR + Vsupply must be no more than the transistor 
rating if breakdown is not to occur. A zener diode may also be used, 
with the advantage that it only dissipates power from the spike. Zener 
will be collector (or drain) to ground for a grounded emitter/source 
driver while a diode will always be across the relay and a resistor 
may be in either location with advantages and disadvantages in each 
case.

MOSFETs are generally more tolerant of breakdown than bipolar devices 
BUT a power MOSFET which is not explicitly designed to spread 
breakdown energy across it's multiple cells (and many aren't) may well 
experience localised overloading and permanent punch through damage. 
At this low power level the manufacturers probably haven't gone to 
vast effort to optimise their devices for forward breakdown so a 
midget power device may be more prone to damage than if the design 
called for much higher power levels.

Due to the modest amount of energy involved in this case (small relay, 
low voltage, lowish current) one MAY get away with no diode but the 
secondary effect of interference with related circuitry will almost 
certainly mean that a diode or similar should be used.

While the end result is a trivially simple circuit there is more to it 
than meets the eye if it is actually to work as intended.



                    Russell.

-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist