Quoting Apptech : >>> >> The load will be up to ~250mA, btw. >>> > >>> > Use Bob's method only. Eg. >>> > >>> > R1 1K 1/4W >>> > R2 10K >>> > R3 6K8 >>> > NPN: 2N/MMBT4401 >>> > PNP: 2N/MMBT4403 >>> >>> You had better make sure the load will use all its >>> available voltage else >>> the PNP transistor will fry. >> >> What is that supposed to mean? None of the high-side >> drivers presented >> are short-circuit proof. > >>>> >> The load will be up to ~250mA, btw. >>> > >>> > Use Bob's method only. Eg. >>> > >>> > R1 1K 1/4W >>> > R2 10K >>> > R3 6K8 >>> > NPN: 2N/MMBT4401 >>> > PNP: 2N/MMBT4403 >>> >>> You had better make sure the load will use all its >>> available voltage else >>> the PNP transistor will fry. >> >> What is that supposed to mean? None of the high-side >> drivers presented >> are short-circuit proof. > > I took it to mean (possibly wrongly) that the PNP needed to > be in saturation or close to it. Well, there's no reason for it to be out of saturation unless the load is too low resistance. Not with Ic/Ib ~- 20. Well, I suppose you could drop it into a dewar filled with liquid N2. > Without looking back I don't know if the 250 mA was > specified when those resistor values were given. As will be > seen below, the circuit is in trouble as shown at 250 mA > with that transistor and resistors. It can almost be > redeemed but ideally either a higher current gain transistor > for Q2 or even better a MOSFET would make it work well. > > Because: > > Drive current will be about (12v-Vbe_Q2)/R1 = 11.4/10k = 1 > mA+ I gave a value of 1K 0.25W, which yields 11.3mA, so the remainder of the calculation won't be valid. The criteria used is Ic/Ib ~= 20. Actual power dissipation is 128mW. > For 250 mA load the beta of Q2 wants to be 300 or better. > Beta of an xx4403 is under 100 guaranteed at 250 mA so you > need say 3x as much base drive for Q2 if you want to support > 250 mA. You should expect to operate *well* into saturation in this sort of application. Ic/Ib = 10 to Ic/Ib = 20, NOT Ic/Ib = 100! > An easier way to run cool is to use a FET. > Either a high side P Channel or a single N channel if you > can tolerate the low side drive to the load. Yeah, or a dual N&P channel, but care is required as to maximum gate voltage on the P-channel (might require a protection zener diode), it tends to be slower and much more expensive. The P-channel you mention below has a Vgs(max) of only 8VDC. It's also a lot easier to get it into a linear mode where it will burn up due to a low supply rail, so you might have to add UVLO circuitry. BJTs, being current driven devices (although they model better as voltage driven , but that's another discussion) don't generally have that issue since base current scales with a low supply rail unless you go nuts with low B-E resistance. > Type depends on drive voltage. > The beautiful but not too common in the US CES2310 that I > mentioned a while ago would do it with ease. Ideal, aside from the detail that they're unavailable. ;-) > Failing that the FDN337 with Rdson = 0.082 typ at 2.5V > drive. > Headline Rdsons are always a lie compared to DC steady state > but at 3V gate you'd get well under 0.1V on voltage at 250 > mA for 25 mW dissipation. > With Tja of 250 C/W it would run touch cold. > Probably :-) > > $US0.44/1 Digikey > $US0.12/1000. > A bargain, if you can't get CES2310's :-) > > > http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=FDN337NCT-ND > > http://www.fairchildsemi.com/ds/FD/FDN337N.pdf > > It's FDN338 P Channel stablemate is also your friend at > about the same price. > As expected, not quite as good an Rdson as N channel part, > but still very good in this task. > > > http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=FDN338PCT-ND > > http://www.fairchildsemi.com/ds/FD/FDN338P.pdf > > Use this in place of Q2 in the original circuit and Bob's > your uncle. > R1 10k or whatever. > R2 100k or >>> R2. > R3 almost anything. > Speed gets slower with higher R. Might be a feature if you want it to switch very slowly, provided it doesn't kill the output transistor in the process. SOA, & Pd. > A speedup cap and some R tailoring could make this very fast > if required (for some values of very). And some values of "could". ;-) Think about it-- this is not particularly easy to do reliably without making or buying a full-fledged gate driver. The level shifting stage has asymmetric output impedance. A couple of mA gate drive will make it pretty fast though (microsecond range) for a *small* p-channel MOSFET (a couple of nC gate charge). -- Best regards, Spehro Pefhany -- "it's the network..." "The Journey is the reward" s...@interlog.com Info for manufacturers: http://www.trexon.com Embedded software/hardware/analog Info for designers: http://www.speff.com -- http://www.piclist.com PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist