<plug> Following macroman's revelation here that one of the PCMods baybuses used discrete transistors instead of the ubiquitous 317, I built this circuit to check performance. Answer was, very good, with an extra volt at the top end compared to a 317. And, although with no overload or over-temp protection its not idiot-proof, with sane use it should be fine, certainly safer than a rheostat. And its dirt-cheap to build compared to low-dropout regs (or even a Maplin 317T). Construction details here </plug>
Agreed that regulation isn't necessary since you're downstram of a regulated PSU to begin with (and fans don't need regulation anyway).
1. Can't wait to try it out - might even get some help/advice from my EE prof. (Sidetrack - you can tell he's cool b/c on the first week of lab, we needed to heat a resistor and see what happened, and he told my group to go ahead and try to smoke it ;--) 2. Linear - awesome new avatar ;--) Gordon on topic? but...
aww its great when teachers/tutors say that what are the advantages/disadvantages of a transistor-bus over a standard one? more voltage at high end, but not as good at heat dissapation?
I'm always surprised when one of my constructions works first time. Early cooling: My pub landlord, on being asked for ice, would respond "Waft it with yer cap"
Any linear regulator, whether ic, transistor or rheostat, has the same amount of heat to waste - the dropped voltage x fan current. With transistors or ics you can fit a nice finned aluminium heat sink to take the heat away, to increase the area for heat loss and catch any passing breeze. With a rheostat you're relying on the surface area of the package and any exposed winding. PWM doesn't limit the voltage and/or current, it chops it into short pulses of full power by rapid switching, so the bulk of the heat generated is the 0.7v drop across the output transistor x %Duty Cycle x Fan current.
#include <stddisclaimer.h> OK, that's true, but possibly misleading. Since the term 'dropped voltage' has different semantics in each case, the power dissipated is quite different for each case, and I know you know that but I want to make sure it's clear to everyone reading. The voltage drop in a Rheostat will be resistive, and can be quite large. In a regulator, there are current mirrors inside the device as well as protection circuits and regulation circuits that drop a bit of voltage, essentially "a few" diode drops. In the transistor design (bipolar junction) you'll have the drop across the collector-emitter junction which can be 0.2 to 0.7 volts. And in a FET deisgn you can have even smaller voltage drops. So in terms of power dissipated as heat, you've got rheostats > linear regulators > BJ transistors > FETs more or less, YMMV. Anyway, I don't disagree with what you're saying, but I did want to clarify it somewhat.
Okay, Linear, but say we approach it from the other end - the fan is running how you want it with say 8 volts at 8/Rfan amps - we've left 4 volts at the same current across the controller. All the unused power has got to come off as heat, 32/Rfan watts. Could argue there's a wee bit more waste with solid state due to quiescent current and current down the potential divider chain. When the rheostat is turned up to full speed (ie zero resistance) it should waste no heat at all - a solid-state controller up to full speed will still waste heat from the junction voltage drops, and at full fan current. Be nice if a current mirror could reflect it back to the Grid
A Darlington pair is just a 2 transistor configuration intended primeraly to act like a single transistor with high gain. In my case I used it to effectively create a high gain power transistor. (Generally high power means low gain transistors). Whether you use it for on/off or linear control is upto the circuit designer.
how could u use it for linear? and we did it im skewl, so i knew a bit about it.... i was kinda write
Mines not quite a Darlington pair equivalent as the two transistors are different, ones NPN, t'other PNP. But the idea's the same, the first (BC107B) has a gain of 200 but will only carry 200mA, the second (TIP32A) a gain of only 25 but it will carry 3A. Put them together and you can rule the world (or odd bits of Mexborough).
Sorry, didn't realise he was refering to your circuit. I thought he was asking what was a Darlington?
Darlington? Place up North, innit? Not tried it for real, but modelling the circuit with a NPN darlington pair (in ProtoLab 4 - free, easy-to-use, but transistors only) max output is 11.1v, with the NPN/PNP pair 11.4v. Not much, but it helps. And at Maplin prices, TIP122 darlington (5A) = 99p BC547 NPN + TIP42A PNP (6A) = 78p Not much, but it helps.
