Can you Darlington Darlingtons?

[ambulancevoice]

i was wondering
for homemade darlingtons, you take two bjt's and put them in this array

i was wondering if you could use darlingtons (like mpsa13) themselves as the transistors in the homemade darlington (ok im using that word to much)

btw, when you make a homemade darlington (with bjt's) does the gain of the first trannie (haha) get x by the hfe of the second or do they add together
for example, you had a 100 hfe tran for the first transistor, and a 200 hfe for the second, would the gain be worked out by 100 hfe x 200 hfe = 20000 hfe or would it be 100 hfe + 200 hfe = 300 hfe?

...

[~arph]

Yes, and they multiply. But I can imagine clipping occurs when darlingtonning a darlingtonny.. (you run into supply voltage max). I never tried it but in theory it should work fine

[ambulancevoice]

also, i forgot to ask
could you put 3 transistors in the h.m.d. array?

[ambulancevoice]

Yes, and they multiply. But I can imagine clipping occurs when darlingtonning a darlingtonny.. (you run into supply voltage max). I never tried it but in theory it should work fine

yeah
i was thinking of trying this


EDIT: holy shit, i jus calculated the hfe
50000000 HFE!!

[object88]

Hm, I think it's called a square-wave generator. 

[brett]

Or a switch.
I made a relay driver out of an MPSA13 and a darlington and power transistor (2N3055).
But I did add a little emitter resistance to the MPSA13, IIRC.

I believe that making Dalingtons out of cheap, low hFE power transistors like the 2N3055 have several potential applications in stompboxes.  You should get high hFE (a couple of thousand) and some extra capacitance that probably keeps the top end response under check (the high freq rolloff might start in the upper audio registers ??).

For silican fuzzfaces, I often advocate power devices (especially the BD139) for similar reasons.
cheers

[R.G.]

Sure - you can hook the leads up any way you want.

The bigger question is - when you get done, does it do what you wanted?

Let's do something radical - let's think first.

1. Hfe is not a single number. Hfe varies, so it's always specified at some current, usually about 1ma for small signal gain devices. That's good enough for a simple handle to hand your hat on, but the fact is that all bipolar transistors have an hfe that varies with the collector current. Lower current leads inevitably to lower hfe. That's the principle that analog multipliers work on. Change the transistor's current, you change its gain.
2. The last transistor works at the nominal current for the circuit. Let's call the final transistor Q1, and the previous one Q2, counting backwards from the output. The second one works at 1/hfe1 of that current. The third one works at 1/hfe1*hfe2 of that current. So the second transistor gives you less hfe than you though you'd get, certainly much less than the 1ma spec on the data sheet. The third one is way down in the mud.  One reason manufacturers made integrated darlingtons was to design the low current input transistors to have high gains at low currents. The stages of integrated darlingtons are dissimilar - they're each optimized for where they sit in the chain.
3. Multiplying signal also means multiplying noise and drift. A single transistor has X noise and Y drift. A darlington has hfe2*X noise and hfe2*drift. A triple darlington has hfe3*hfe2*X noise and hfe3*hfe2*Y drift. Stabilizing one transistor against drift and keeping it quiet is something to consider. If the gains you get from preceeding stages is only hfe2*hfe3 = 100, you make it 100 times as hard to keep noise and drift down.
4. Hfe is not voltage gain, it's current gain. The real name of the game for bipolars is to trade off excess current gain to get stable and predictable voltage gain plus stability. As a simple example, what's the "hfe" of a MOSFET? The gate is isolated by a layer of glass 20 volts thick. The current which flows into the gate is is so near truly zero that it's almost impossible to measure. Any current at all flowing in the channel gives an "hfe" of so close to infinity that it's not worth calculating. But MOSFETs do not have infinite voltage gain. Why not?

It's because the current in the drain (or collector) must flow through a resistor to make an output voltage. The current can't be infinite, so neither can the output voltage, so neither can the voltage gain.

Consider an imaginary transistor with a single hfe of 10,000,000. It has a collector resistor of 10k ohms and an emitter resistor of 1K. What's the voltage gain?

It's 10;   10K/1k. What if the transistor has an hfe of "only" 10,000, same resistors. What's the voltage gain? Yep, you guessed, it's 10. That's because the transistor's current gain is not what determines the voltage gain. True, the CURRENT going into the base may only be 1/10K or 1/10M of the current in the collector, but that's neither here nor there. The resistors set the gain. Even if the emitter resistance is zero, the transistor's internal resistance sets the gain, and that's not dependent on hfe, only the collector current.

As a practical matter, hfe of somewhere between 100 and 1000 is sufficient for anything you want a discrete bipolar device for. The higher gain ones tend to have lower noise and act more "perfect", but the amount of advantage you get is not always commensurate with the increase in hfe. Diminishing returns sets in fast.

[ambulancevoice]

i actually plan on using this device for noise and maybe to amplify quiet sounds incredibly loud
and additional noise doesnt really bother me

[R.G.]

It will be an average to poor noise generator for a couple of reasons and there are easier ways. I don't mean to discourage you, and you might get exactly what you want; it's just that there are other ways to do what you're trying to do that are not as problematic.

The issue of drift can rise up and drive you crazy. If you want to amplify the input noise of a transistor, you also amplify the drift by the same amount. No way around it. I was the lab instructor in a circuits class when I was in school, and some of the students came up with high gain circuits that you could drive into blocking - no signal gets through at all - by breathing on them from several feet away.

Not all noise is the same. Most transistors have bad flicker noise, also called 1/F noise. It rises as frequency decreases. Flicker noise is bursty and intermittently crackly, while thermal noise is "white" - like hiss. Flicker noise is interesting, but may be what you want to generate, may not.

And signal to noise ratio will drive you crazy. You obviously want a low-noise amplifier to amplify tiny sounds up so you can hear them. As counterintuitive as it sounds, you also want a low noise amplifier to amplify up small noises as a noise generator. This is because presumably you want to listen to the noise on the input sensor, not the noise generated by the amplifier itself; the amplifier may be adding 1/F or other kinds of noise that are not what you wanted.

If the objective is to play with parts you have, kewl! If there are two objectives, to generate noise and to amplify up tiny sounds, you'll do better with a low noise opamp run at high gain amplifying a reverse-broken base-emitter junction for noise generation and a low noise mike for signal.

But experiment! Have fun!

[Steben]

What about negative feedback, RG? Won't it help reduce all of the disadvantages?

[R.G.]

Sure, negative feedback will reduce the drift disadvantage. But you can get to substantially the same gain you'll get after feedback by using fewer parts. There is a strong effect of diminishing returns. Somewhere between hfe=100 and hfe=1000, you get almost all the advantage you can get out of ordinary bipolar parts. That's a gross oversimplification, but it's a good rule of thumb.

Negative feedback doesn't do much if any to help with noise.

[R.G.]

Sure, negative feedback will reduce the drift disadvantage. But you can get to substantially the same gain you'll get after feedback by using fewer parts. There is a strong effect of diminishing returns. Somewhere between hfe=100 and hfe=1000, you get almost all the advantage you can get out of ordinary bipolar parts. That's a gross oversimplification, but it's a good rule of thumb.

Negative feedback doesn't do much if any to help with noise.

[soulsonic]

It's supposed to sound like a big distorted mess. I suggest building it up on a breadboard and just fool with it until it does it's thing. It's hard to say though if there would be an advantage to doing it like that as opposed to some other way to get it grossly distorted. But of course, each kind of massive destructive overload kinda puts it's own little flavor onto it. When it comes to Noise, anything goes, so as long as a sound comes out of the circuit, I'd consider it usable for your purposes. Hook up some contact mics attached to some metals and...... well, you know what comes next...