Transistor bias and adjustment?

[Tom Lauten]

I think I'm on the verge of understanding this but could someone tell me how one actually goes about "adjusting" the bias of a fixed transistor in a predetermined circuit? Is it a matter of exchanging resistors or caps? Can one use a variable resistor somehow?

[antonis]

Of course you can use a variable resistor but it's an overkill - as long as you don't deal with MOSFETs bias..
(variable resistors are used often for gain adjustment...)

You can bias a transistor in several ways but the common practice is either by a voltage divider(Vcc/base/ground) or by a collector-base feedback resistor...

http://www.learningaboutelectronics.com/Articles/Transistor-biasing-methods
http://www.learningaboutelectronics.com/Articles/Transistor-biasing-methods
http://www.allaboutcircuits.com/vol_3/chpt_4/9.html

Have you any particular circuit in mind..??

[GibsonGM]

Another thing to just mention might be that, because you said "...in a predetermined circuit...", you could be 'asking for it', lol.

In an already-designed circuit, Generally speaking, the designer has set the bias on such things to where it will sound good and function right.  There are exceptions to this, of course...some people LIKE a mis-biased fuzz face and so on, or want to tweak something a little.   But moving the bias point of 'done' circuits can lead to reallllly bad sounds.    Do it wrong, and you could pop the transistor, too, depending on how it's set up (ok, not common but it COULD happen).

That said, it's sure a 'do-able' thing, and when designing your OWN stuff could be something to try out!   As long as you have a grip on what is really happening when you do this (load lines, current flow, etc).   

Glad you're interested in how things are biased, and why!  Makes playing with transistors more fun to know what's going on...

[Tom Lauten]

WOW! Ok, that stuff is waaaaaaay over my head! 

I guess I was hoping there was a more layman's', nuts-n-bolts version of determining, measuring and adjusting bias. I would need to learn an AWEFUL lot about electronic engineering and application PLUS advanced mathematics to even begin to get my head around this.

Problem is that it's feels like trying to learn how to speak Mandarin by listening to the directions in Russian when I have enough trouble with English.

[tubegeek]

I'm going to try channeling PRR here for a second, let's see how I do.

Bias is a voltage which sets the no-signal behavior of an amplifier. The current through the device then increases and decreases (compared to the no-signal current) in response to the input signal.

An analogy: hang a spring off a hook, some distance from the ground. With no weight on it the spring will snap back completely and the bottom end of the spring is as high as it can get. With a heavy weight on it, the spring will bottom out against the ground.

If you hang a medium-sized weight from the spring, it will hang partway down, between its highest and lowest positions. This will give the spring freedom to travel upwards a ways before it runs out of room, and downwards a ways too.

Bias is like that partial stretch of the spring: if the current shuts off completely or flows at its maximum, those are the upper and lower limits you bump into.

In the spring example, hanging enough weight to leave the spring exactly in the middle would give you the maximum symmetrical travel possible, so you can see why we often start with "center bias": half the power supply voltage. But since this forum is about stompboxes, I'd be remiss if I forgot to mention that sometimes what we want is exactly the "running out of room" asymmetry in signal swing. So that's why we adjust bias to taste. Maybe we want to bottom out against the ground before we run out of room to the upside: bias heavy. If we want to run out of room upwards sooner, we bias light and so we have less room to go up than down.

[GibsonGM]

Hi Tom, we all felt that way at first, no shame!!  Sometimes all of us still do, when we encounter something new or something we just did NOT consider about what we're working on!!  Tubegeek did pretty well there

If your power supply is a 9V battery....and you input a 2V peak-to-peak SINE WAVE into your little transistor amplifier that you have not set up correctly....it will happily amplify it UP one volt, but when it comes back down to zero, it will STOP amplifying the signal.  So 1/2 of your signal will be lost, or CLIPPED.  <1/2 wave rectified, actually>

Instead, with a BJT transistor - a CURRENT-CONTROLLED DEVICE - we set the collector of the transistor to be about 1/2 the supply voltage.  This, along with setting up the base of the transistor to 'conduct well' (more about that as you learn!), makes the transistor 'rest' at about 1/2 the supply.  SO, when the sine wave comes in, it can go up AND DOWN without being clipped off.   We make the transistor's QUIESCENT state be about 1/2 the supply voltage instead of zero.  Make sense?

When it does make sense, and you've played around with some stuff like an LPB or other basic 1 transistor circuit, that's the time to actually learn how to compute the bias network...it's pretty important to just get a layman's sense of how the transistor amplifies first!

This applies to opamps and FETs as well, with some small differences due to them being of different construction. 

For now, understand that a small current on the transistor's base to emitter makes a larger current flow from collector to emitter.  It is a valve.  In AC work, we need to account for the fact that a real sine wave crosses zero, and DC (which powers our little amplifiers) does not.  So we play tricks to move the reference levels around.  That's all that bias is. 

Coming up once you 'get this'....back to your original question - you pick a current you want flowing thru the BJT at quiescence, set the collector voltage to be about 1/2 supply at this current (OHMs LAW OHMs LAW), set the emitter resistor based on about 1V drop, and then you have to set up the bias network divider based on what base current you need to flow (related to the gain of the transistor).   Not as bad as it sounds, but really knowing what's going on is pretty necessary! 

[Tom Lauten]

Thanks guys. That explaination really helped. It's kinda in the arena of what I was thinking of in terms of the concept.

How does one actually go about changing or setting the changes needed if that's what you decide to do? Varying resistors?

[antonis]

How does one actually go about changing or setting the changes needed if that's what you decide to do? Varying resistors?

The οne word answer is: YES..!!

(but maybe are necessary more "nasty" words like: current consumption, power distribution, Q point, e.t.c.)

[GibsonGM]

You should go on Youtube and check some vids like These :  https://www.youtube.com/playlist?list=PLlm5X-un2rH9W5eihMMXzZ-48s_b1ESVl

Very simply, a transistor has a range it will operate over without distortion (BAD distortion, usually not the GOOD stuff we want).  You can define this in a graph where you see base current vs. collector current...this is a LOAD LINE.  (look that up too).   You'll see a line with curves on either end.  For distortion=free ("linear") operation, you BIAS the transistor to work within the straight portion of the line.  When you ride up or down into the curved portions ("NON linear regions"), you get distortion of your input signal.  Remember, this is usually crappy sounding.

You may WANT this (some fuzz circuits, or other amp classes), at times, but most often it's not very desirable.  So, when designing, you try to set the quiescent operating point of the transistor somewhere inside the linear portion of the load line.  That's done, as described, by using your resistors to 'elevate' the operating point so that your input signal can swing above and below this BIAS point.  You need to use Ohm's law to set the currents flowing within the transistor (more later).

A basic NPN amplifier will consist of input/output caps,  a collector resistor, emitter resistor, and a voltage divider network (2 resistors) that divides the supply voltage and sends it to the base.   Figgering out how to set these resistors up is almost all there is to it!   But first you need to go thru many of the vids/tutorials and understand what's happening (look for oscilloscope tutorials!), and know Ohm's law inside and out.

PRR did a GREAT and fast tutorial on how to simply bias a stage, which I saved.  When I can get it up on Imgur for repost, I will.   So for now, learn from vids what a basic NPN amplifier stage does, and what CLASS of operation means.

We're not 'hiding' anything from you, just encouraging you to understand more about that whole "biasing" thing, and to really get a what is happening - it's important if you want to have control over what you design or 'play with'....from concept to reality

[Tom Lauten]

I appreciate the help. I'll start with the videos.

At this point I'm just building from prepared vero layouts and such. They often mention biasing so thought it best to ask. I'm good to go with assembly and such but how and why components are used is still alchemy to me. Maths was NEVER my strong point so that's kind of intimidating especially not knowing the various roles of components in various type of circuits. I wish the was a basic electronics course in my area but I'm in the Highlands of Scotland...the choices are nil.

I appreciate the hand holding here...gives me some confidence! Thanks guys!

[PRR]

The basic problem:

Audio goes both ways.

All the simple amplifying devices (transistor, tube) only go one way.

Our "trick" is to "bias" the device (no signal) more-or-less in the "center" of its available swing. Then when we add signal, it swings up and down around that center. The external world should not know about this internal matter, and external stuff should not upset the internal bias, so in Audio we use DC-block caps input (usually) and output (always) to keep this bias inside.

TG's weight on a branch is a nice cartoon. We might really want the weight to swing both-ways around some "zero" point. In complex systems we call the "zero" a "ground", really meaning a wire that is common to many parts (both power and signal) of the whole system.

The weight can not swing above the branch (twack), nor below the ground (thud). But we can fake a both-ways swing by starting (biasing) the weight to mid-swing, as in TG's middle picture. Say that is 4 feet up from dirt. Now add a "both ways" swing. Up toward branch, say 6 feet up, and down toward dirt, say 2 feet up. But when we are "outside" this part of the system, we remember to subtract-out the 4 foot bias and read the swing as + and minus 2 feet from zero. Coupling-caps do the subtraction easy/cheap. (They charge-up to whatever the average bias is, and hold that bias through rapid swings.)

In a simple single-transistor, if the Collector is near half of supply, it is probably "good". Small changes from that will not be large changes of sound. Let it be.

If you want "bad", twack/thud, mess-up the bias. The take-away from all that theory is that "all" the resistors affect bias, so fiddle any *one* of them until it sounds bad (good for distortion).

[GibsonGM]

Hey Paul, remember this one??  I picked it up quite a while back, and it really helped me to see what's going on with a BJT amplifier.  Hope you don't mind my re-post as I found this to be very easy, concise and really cut thru a lot of BS....

The text of PRR's reply to a question like "How do I bias a transistor amplifier" follows:
--------------------------------------------------------------------------------------------------------------------------------------------------------------
"Assume 10V supply.

We want the collector resistor dropping a lot of this. "Half" is good.

Transistor base-emitter voltage is about 1 Volt. (It will really be 0.5V-0.7V, but let's avoid tricky math.)

We need an emitter resistor to drop a voltage "not small" compared to transistor base-emitter voltage. 1V is enough.

So far we have 10V supply, collector at 5V, emitter at 1V. We have lots of room (10V-1V) to swing the collector, the output terminal. We waste only 1V in emitter bias.

Remember these relative proportions! Collector resistor and collector-emitter drop most of the available voltage. Emitter resistor voltage is smaller.

How does the emitter know to be at 1V? We force the base so that 1V appears at emitter. We already pretended the base-emitter voltage drop is 1V. Therefore the base must be forced to 1V+1V=2V.

We could use a 2V battery, but not handy; we will also eventually mix audio here and a battery would absorb audio.

We use a voltage divider. Be familiar with the concept.

Pick a collector current. Often 1mA is as good as anything.

Emitter current is the same as collector current.

Now we can pick collector and emitter resistors.

We need to pick a current to flow in the voltage divider. Too small won't force base voltage firmly enough. Too high will lead to small resistors which will absorb the signal we will add later. When you don't know better, pick divider current 1/10th of collector current. 1/10th of 1mA is 0.1mA.

Later you can add your in, out, and emitter bypass caps."

[antonis]

The above reminds me a basic mistake done in enough of my designs...

I was allways trying to bias the transistor at about half of the power supply to have the smallest CE voltage drop (200 - 300 mV)...
(stuck on linear BJT behavior...)

It took me sometime to realize that the obtained distortion was actually from the "negative" cycle clipping and not from the C/E resistors gain...

[PRR]

> bias the transistor at about half of the power supply

For maximum "clean" output: The transistor and the load resistor should *each* drop about-half of the available supply. Under 9V power, about 4V each.

You often have to "lose" some voltage in an emitter bias resistor. For maximum output, this should be "small". BJTs are pretty predicable and you can usually hold this to 1 Volt. Unless you need a large Re for low gain or low distortion or high input impedance.

Biasing "low" (1V C-E, 7V in Rc) will give a bit more Gain but on any significant signal will give negative cycle clipping. Biasing "high" will do the opposite.

Usually, unless you need only small gain or have very favorable impedance requirements, you really need to be thinking two transistors. Two transistors cascade give MUCH more gain than one. You can optimize the first for gain and the second for high max output. Two direct-coupled transistors combine the bias compromises and you often can have less loss due to bias needs. The FuzzFace is a classic 2-Q direct-coupled design.

[GibsonGM]

I'd love to see a 'concise' tutorial on how to set up direct-coupled BJT's with some degree of 'properness'...