Transistor Configurations

[raulgrell]

I have been seeing very few circuits that use darlington transistors, and none that use Sziklai pairs (complementary darlingtons). But where the darlingtons are used, the often have much better performance than a single BJT.

What justifies the use of a darlington? What makes it better (or worse) than a single BJT?

I was also thinking whether it would be reasonable to make a "fetlington" or a darlington made with FETs instead of BJTs... or Maybe a FET and a BJT... Hell, while we're already here, why not a "mosfetlington" ?

The idea also occurred to me of making a darlington with 3 transistors instead of just 2... Would this have any advantages/disadvantages?

Cheers...

[aron]

I don't know, but am guessing that for the most part, we don't need those high gain for most of our circuits. It seems that most circuits are trying to tame the gain, rather than slam more into the circuit.

[Sir H C]

Darlingtons allow you to get a much higher current gain than a single BJT.  THe problems are that you often get slow response and you now have two Vbe's that you have to drop (can't do a 1 battery thing very well if at all).  Triple darlington is usually not good because with the multiplications of the current gains, you will either have leakage at the input quashing anything at the output or have stability issues.  There is a great book on power amplifier design ( http://www.amazon.com/High-Performance-Audio-Power-Amplifiers/dp/0750626291/ref=sr_1_3/103-9450657-1221447?ie=UTF8&s=books&qid=1187554692&sr=1-3 ) that has a whole section on "triples" which are three transistor (though the last stage could be multiple devices) configurations that can do a lot of cool stuff and the plus/minuses of these.

Szikli pairs are not used much now as there are decent PNP devices.  I have done a CMOS Szikli for an IC years back.  Worked well.  There is a device called an IGBT (IIRC) that is a MOS input bipolar device.  Usually used for high power switches but Threshold used them for power amplifier stages in some of their amps.  The FET darlington would be pretty moot as there is no (well usually) DC gate current that you would want to buffer against.  

[raulgrell]

What do you mean, slow response? In switching circuits, I guess that makes sense, the transition from "on" to "off" being slower... but in audio applications, I can't see how this would be a problem... May even be beneficial for some circuits...

Regarding the high gain nature of the darlingtons: Extremely high gain could be very beneficial for certain circuits, I know some (eg Bazz Fuss) work better with a darlington than with a single BJT for that reason...

What about the (MOS/J)FET ideas? Any viability?

[Sir H C]

Realize that a darlington is high current gain.  MOSFETs and JFETs have pretty much infinite current gain so it is moot to do the darlington configuration with them.  Also, since the gate is high impedance, unless you have a path for the current off the second device gate, the first device will be a very poor control for the second. 

By slow response, look at some Ft's for darlingtons vs. discrete devices.  Often they are a factor of ten or more off.  The problem is getting that charge out of the second base, often that is why there is a resistor to the emitter of the second device.  In general you want to control where the response slopes off with external capacitors, but in this case, you can have the parasitics of the device creating poles low enough in frequency that they start to roll off response and add phase shift that will really hurt you if you have negative feedback going on.  Too much phase shift and that negative feedback becomes positive and you have an oscillator and not an amp.

[raulgrell]

So the issue with darlingtons is that their parasitic capacitance meddles with the frequency response of the circuit? I was reading an article the other day about capacitance, and it said an esy way to solve the problem of this parasitic capacitance was to have a high value resistor between the base and the emmiter, although it was referring to a single BJT... i think it could apply to a (home-made) darlington. I'll experiment with different resistor configurations and report back as soon as I get some new trannies.

Regarding the phase shift added, is there any way to control it? Or by solving the capacitance problem, is the phase shift is eliminated?

If these problems are overcome, are there any instances where the high current gain is a possible benefit?
Is the gain usually delivered by transistors voltage gain, current gain, or both? This discussion has got me a bit confused, as I always hear of people talking of signals in volts and never amps... and google has just made it worse =(

Damn, I needa take an electronics course =P

[Sir H C]

Bipolar transistors and FETs are voltage controlled current sources.  That is the fundamental math that controls them.  So the current gain of a bipolar is a byproduct of the physics but not necessarily something you want to have.  Infinite current gain would be best like you have with FETs.  In the end the darlington is good for situations where you want large current outputs but a high input impedance.  Since the input impedance is related to the resistance on the emitter times the current gain, you can get a high input impedance and get a good output current with a darlington.  The curse with the resistor to speed up the darlington (some put it in the 3 terminal ones some don't), is that if it is a 3 terminal device and it is not there, you can not access that terminal to put the resistor in.  For audio, you shouldn't have a problem, but it is possible that one might have issues if you put three in a row as you would have little current to drive those other devices. 

Since the output current is across some load, that converts it to a voltage.  You are right most things are viewed just as voltages, in power amplifiers and "current mode" circuits (a whole field unto itself), you deal with the currents more directly, but in most preamplifier circuits the current usage is more what you want to use not what is needed precisely for driving some load.

[jaytee]

You must be just looking at modern circuits. There's lots of old designs that use darlingtons. The univibe and sola sound tonebender are 2 examples. I think the reason they're not used as much now is because opamps came along. Designers only used to have discrete transistors to work with. Opamps are easier to use. High input impedance, low output impedance and tons of gain so closed loop gain can be set accurately. Opamps out perform darlingtons in most instances. Having said that, for quirky stompboxes we might not want better performance. It's more about tone. You might want that old timey sound that's a bit noisier and dirtier. I have a peavey amp that has all darlington stages in the preamp. Overall the amp sounds crap but the overdrive is quite nice.