Learning about a one transistor gain stage.

[artifus]

a bit of on stomp boost can be a useful musical effect, for a solo, to punctuate a riff, elicit a bit of amp overdrive or whatever. thanks to kesh for starting the thread and to all contributors - it's been an education.

[Jazznoise]

They're a good beginner project and sometimes the amp is just too far away!  Passive EQ can be added after this, for example, if he wants to expand upon this project.

PRR, do you have any good books written on transistors? Preferably written by you?  I'm still shakey on the HFE side of things. Modern (or should I say current?  ) college courses are very voltage orientated!

[artifus]

PRR, do you have any good books written on transistors? Preferably written by you?   I'm still shakey on the HFE side of things. Modern (or should I say current?  ) college courses are very voltage orientated!

i second that emotion. i have some fine books that no doubt contain all the information required for understanding and can follow the schematics but they soon descend into hieroglyphics in the explanation of the circuit that seem to require a degree in mathematics to decipher. i know i'm putting off the inevitable - having to improve my own knowledge of math - but a good analogy can go a long way in helping one grasp a concept and encourage one to want to learn more, to find a good practical use for that theory. having a goal, ie 'i want this circuit to do this', can be a great impetus to spur one on into wanting to learn the theory that can lead to how to achieve such a result.

[PRR]

> better, more stable buffers or boosters when we have higher hfe.

That's true, if you don't get greedy.

Gus has a series of designs using mid-hi hFE plus bootstrapping to get very-HIGH input impedance.

However transistors are cheap now, and two transistors have effective hFE of 10,000+. You can drive a good output current with vanishingly small input current, high input impedance; and cost/complexity may be little more than a bootstrapped circuit.

[Kesh]

> better, more stable buffers or boosters when we have higher hfe.

That's true, if you don't get greedy.

Gus has a series of designs using mid-hi hFE plus bootstrapping to get very-HIGH input impedance.

However transistors are cheap now, and two transistors have effective hFE of 10,000+. You can drive a good output current with vanishingly small input current, high input impedance; and cost/complexity may be little more than a bootstrapped circuit.

yeah, i'm basically trying to squeeze as much as i can out of a single transistor as an exercise in knowing their limits.

[R.G.]

Even that gets tricky. If you do the math, the gain of a single transistor is pretty much the collector load resistor divided by the emitter resistance. For max gain you want the highest possible collector resistor, which in turn means the highest possible supply voltage before the transistor dies by avalanche failure. And you want the smallest possible emitter resistor.

The emitter resistor can never be zero. There is an internal resistor which arises from the transistor action and varies with emitter current. This is the "Shockley resistance". It puts a minimum on what the emitter resistance can be, and varies, so an external fixed resistor is used to make the varying internal resistor not matter so much. It's back at the idea that whatever cannot be controlled must be made irrelevant.

If you allow yourself to go beyond one device, the obvious thing to do it to make the collector load very large by making it a constant current source. This can be a single JFET or a bipolar plus a few other parts. This kicks the achievable gain up a lot. Gains from single transistors in the low thousands can be done.

... uh, as long as whatever loads the collector is much larger impedance than the collector current source load impedance. Ooops. Gotta have a good buffer, otherwise the input impedance of whatever it drives becomes the load that defines the gain.

It winds up that the limits on a single transistor gain stage are not entirely determined by the stage itself, and usually have to be pretty modest to get predictable results. Getting exceptionally good results in a few cases is good for athletics, very bad for electronics. You don't want a few good examples, you want every example equally good, every time.

[Kesh]

Not after much voltage gain, mostly impedance. I just breadboarded a circuit with a 2n5962 that got from 1M input to 10K out and had a small gain also, maybe factor of 3. It sounded fine.

Wish I had a better DMM. It can't measure the base voltage without messing up the circuit's biasing.

[PRR]

You should NOT try to learn transistors until you are VERY secure with basic Voltage, Current, Resistor problems.

Not mine; but I've mentioned it enough to raise the asking prices.

Analysis & Design of Transistor Circuits
Laurence G Cowles
D Van Nostrand Company Inc
1966

Vintage review:


$20-$40 at ABE.com

I chewed on this for years before I felt comfortable. Some of you sharper guys might absorb it in a semester.

One advantage of Cowles is that he was not highly educated, and learned transistor work as a field engineer for TI.

Do NOT take this, or any book, as your sole source. Every author has blind-spots.

Do NOT bother with Cowles' other books.

[R.G.]

Correct, Paul.

I like Robert Bonebreak's "Practical Techniques of Electronic Circuit Design", but I suspect I'd also like Cowles. Bonebreak rips through common emitter, common base, and common collector stages in the first nine pages, and it's easy to read.

... and then you go "um, wait a minute; what did he say back here?" for a year or two. 

But there are many such texts of varying ease of reading. I like finding nuggets in used book stores, where they may be well under $10 if you are lucky. Lots of this is on the web too.

Kesh, it's laudable for you to try and figure stuff out from first principles, but all this was done in great detail by people earning PhDs back in the 60s and writing textbooks to keep their professorships. If you just want to know how it works, it's much faster to go do the research and read the books.

[Kesh]

You should NOT try to learn transistors until you are VERY secure with basic Voltage, Current, Resistor problems.

well, yeah, obviously. but that's high school stuff. actually basic transistors is high school here in uk, but i didn't do those electives

Kesh, it's laudable for you to try and figure stuff out from first principles, but all this was done in great detail by people earning PhDs back in the 60s and writing textbooks to keep their professorships. If you just want to know how it works, it's much faster to go do the research and read the books.

i'm ok with laudable.

i'm from a pure maths background, the maths here is no problem to me, what is great is having something i can do the maths for and then put into practice. this is something i never got in  pure maths, obviously. and doing stuff from first principles is a great way to get inside a topic.

anyway, i now have to unravel the mysteries of jfets. a far more sensible way to make a buffer.

[Gus]

Buffers using a NPN transistor

[Jazznoise]

well, yeah, obviously. but that's high school stuff. actually basic transistors is high school here in uk, but i didn't do those electives

http://store.aqa.org.uk/qual/gce/pdf/AQA-2430-W-SP.PDF

Not insofar as we're doing them here. Unless you're implying the A Levels are a rigorous study in just about anything. Which they're not. We don't have an electronics module in Irish schools, but we cover about 2/3's of that syllabus within Physics. And then do 6 other subjects. 

People like AD and TI are putting entire circuits, including transformers, on a single die now. There's no need for the Engineer in the Test or Application department to model the interal workings, just see what circuits they can be implemented in and are the values close to the expected ones.

Thanks for the book, PRR!

[Kesh]

Gus, I'm going for more input impedance and less output impedance.

[Kesh]

well, yeah, obviously. but that's high school stuff. actually basic transistors is high school here in uk, but i didn't do those electives

http://store.aqa.org.uk/qual/gce/pdf/AQA-2430-W-SP.PDF

Not insofar as we're doing them here. Unless you're implying the A Levels are a rigorous study in just about anything. Which they're not. We don't have an electronics module in Irish schools, but we cover about 2/3's of that syllabus within Physics. And then do 6 other subjects. 

not sure how "basic" implies rigorous.

[Jazznoise]

I'd consider not doing a Rc/Re calculation not doing transistors. They're learning about how a transistor can be used as a switch - or at that least that it IS used as a switch. There's abit of V=IxR in there, but the concepts of input and output impedance aren't covered, nor is complex impedance addressed at all. They don't even seem to do Kirchoff's laws! It's this sort of madness that means my brother's Junior Cert (Irish GCSE) was given an unconditional offer for Salford Uni.

I'm just saying you were better off doing maths, really! 

[R.G.]

i'm from a pure maths background, the maths here is no problem to me, what is great is having something i can do the maths for and then put into practice. this is something i never got in  pure maths, obviously. and doing stuff from first principles is a great way to get inside a topic.

Gus, I'm going for more input impedance and less output impedance.

Good for you. I do occasionally enjoy putting everything down into a string of symbols and seeing what relationships fall out. And it is very enjoyable to see what the real world does in response to the equations. As a bit of caution, the parasitics of transistors are much better now, but for the upper half of the audio band you really should start including the parasitics so you see the frequency-related changes in your math. Fortunately, with a math background, you shouldn't be impeded by some complex (in the imaginary-numbers sense) algebra.

I personally have always been fascinated at what happens to equations when one starts deciding which parameters matter and which ones do not. It's amazing to me how a string of symbols going all the way across a page reduces to three or four variables when the parameters which only cause minor changes in the result are presumed to be fixed. For instance: Vbe. Vbe is the forward voltage of the base-emitter junction and is a logarithmic function of base current with some additional oddities from transistor-action physics and the physical construction of the device. However, it will always be between 0.45V and 0.7V for small-signal silicon devices, and one can in most cases find answers to an acceptable degree of accuracy by replacing it with 0.6V for silicon. It does vary, but the other things in the equations matter far, far more.

The most important issue for bipolars is beta. The input impedance for CE and CC stages is essentially beta times the sum of Re and re (the Shockley resistance, which was evaluated empirically, and is itself approximated based on current). There are limits on the values of Re one can use in a circuit and the values of re one can buy in a transistor, so the biggest single effect is beta. For output impedance, in the CE stage, that's so close to the collector resistance in almost all cases that the rest can be ignored. For the CC stage, the output impedance is [a bunch of stuff involving Re] divided by beta, so again, beta is the biggest determiner. For lower output impedance, get more beta.

The mantra of the engineer is "that which cannot be controlled must be made irrelevant", and that's what good design with transistors does. Once the huge effect of beta was understood, design processes were developed to make choosing components simple given only that beta was big enough, since beta cannot be controlled well.

To get significantly better results for high input impedance and low output impedance than a high-beta device in a CC (emitter follower) circuit, one needs to go to more than one device with techniques like emitter current-source biasing, buffered bias (of which bootstrapping is one simple and elegant form), huge betas with darlington-connected devices and so on.

Actually, operational amplifiers are another stage of this process. Once you decide you can allow yourself to use more than one amplifying device, feedback techniques allow you to get higher input impedance and lower output impedance, both well-controlled by simple circuit elements, with essentially no consideration of the internal construction of the amplifier at all (at low frequencies of course).

If you're launching into FETs, be alert to look for the effect of dynamic processes on that nominally high input impedance. JFET gate impedance drops precipitously as a result of leakage and dynamic effects. In another way, so does the effective input impedance of MOSFETs. Watch for that as you dig out the equations. Also, the equations require a model of the internals of the device. For good results, you have to pick the model carefully.

Good luck, and happy equating.

[Gus]

Gus, I'm going for more input impedance and less output impedance.

That is why I posted the screenshots.  The NPN is bootstrapped for high input resistance and an EF can have lower output resistance than a simple Jfet circuit.
I only use a jfet when I have to.
How low an output impedance do you need?

[PRR]

> unravel the mysteries of jfets

Best single source:

Designing With Field-Effect Transistors
Second Edition
Siliconix Incorporated (Author), Ed Oxner (Author)
Publisher: Mcgraw-Hill (Tx); 2 Sub edition (October 1989)
(some citations show 1990.)
(There is a 1981 edition which is similar but not the same.)
ISBN-10: 0070575371
ISBN-13: 978-0070575370

$35-$55 (and up) on Amazon or ABE (1981 ed 9780070574496 can be found for $5.)

________________________________________________

> jfets. a far more sensible way to make a buffer.

The input is near-infinite, fine.

Problem is that bias voltage for a given part-number may vary 4:1, and the higher voltages are a large fraction of an assumed 9V supply. (Most JFETs still on the market are used for RF applications where they are run zero-bias, wide-open, with inductive load.)

I won't mention that in this post-1972 world there are fine chips to do such work with negligible brain-work.

_________________________________________________

> I'm going for ...less output impedance.

What is the Zout of Gus' circuit?

The small-signal Zout is under 100 ohms. (Large-signal is different.)

However it has unity voltage gain.

[PRR]

> Shockley resistance, which was evaluated empirically

I believe Bill derived it from fundamental physical principles, well before functional devices were "clean enough" to get good numbers experimentally.

However Bill's big book is too expensive for me, and also said to be very dense.

Chenming Hu has a modern treatment for free download.

_____________________________________________

If that leaves you hungry:
Modular Series on Solid State Devices: Bipolar Junction Transistor
Gerold Neudeck
ISBN: 0201053225 / 0-201-05322-5
cheep

The whole series may be of interest.

_____________________________________________

But as R.G. is telling you: _practical_ circuit design is done by assuming 0.6Vbe and ~~100 ratio of collector to base current and finding likely candidates. Then rough-check with other Vbe and hFE to see if approximations fall apart.

[Gus]

http://www.diystompboxes.com/smfforum/index.php?topic=98928.20