Transistor Biasing Style. Which and Why

[Johnny G]

ok basically i have noticed that there are two widelly used styles of biasing transistors and i cant work out which is better/worse and why etc.

so i figure ill ask here. the two styles that im talking of are thus.



the one on the left is a simple voltage divider to bias the transistor and then the resistors connected to the collector and the emmiter are there for gain.

the circuit on the right is almost exactlly the same except the top of the voltage divider is connected to the collector of the tranny (i dunno why im saying this you can obviouslly see lol) i realise that this will give a certain amount of negative feedback but i am unsure as to how to work out the total gain of the circuit taking the feedback into account and also which of these is better/more stable/more linear etc etc

anybody who can give me some good advice (feel free to go fairlly indepth)  wouldbe greatly thanked

cheers

JG

[R.G.]

Those are indeed the two commonest forms of biasing the NPN bipolar transistor. There is a third that I highly recommend that I'll get to at the end of this mess.

The first circuit you show, with the top base biasing resistor going to the power supply is commonly referred to as the "Stabilized bias circuit"(SBC), or it was when I first took beginning circuits back in 1971. The second doesn't have a specific name that I know of; let's call it the feedback biased circuit (FBC).

The performance differences between the two are coincidental; you can get good gain, frequency response, input impedance, etc., out of both. They vary slightly in that the SBC can be designed for a higher stability to thermal bias shifts than the FBC. The FBC can be made to have a higher output voltage swing for a given power supply than the SBC, although that usually means giving up the emitter resistor.

In the FBC circuit, the lower base resistor (base to ground) is optional, and is only used for getting some other advantage. The circuit can be, and often was used with that resistor being infinity - an open circuit. In the form shown, it bleeds off "excess" base current from the collector, and allows you to use a lower resistor value from collector to base than you would otherwise do.  The FBC with a high Rcb (resistor from collector to base) and a low or zero emitter resistor has the most possible voltage swing on the collector for a given power supply. It suffers from low input impedance and other woes, like thermal drift. The wide voltage swing is a gift of the feedback resistor moving the bias for you as the collector moves.

The SBC is the classic circuit for "you asked for it, you got it". In the SBC, the gain is essentially always Rc/Re for gains up to about 30 or so, and it's predictable and stable. To get this stability, you throw away the static voltage across the emitter resistor. The gain is very, very predictable and the bias point is rock solid.

The FBC is really only half a circuit. As shown, the gain is close to Rc/Re like the SBC, but you can get more out of it by making Re=>0 and using an input resistor. With low Re, the gain gets high, maybe 50 or 100. Then you use a series input resistor and the gain approaches Rf/Ri like an inverting opamp circuit - which is what it becomes. In that use, it is quite stable, and has a wide output voltage swing.

That's a fairly incoherent and rambling sendup. There's a lot more if I take the time to think and type more. I use the SBC whenever I need a small, fixed gain. I tend to never use the FBC by itself, but in conjunction with a second device to make it more robust.

The third circuit I mentioned is a mod of the SBC. The only real problem with the SBC is that it has a low input impedance, equal to the two bias resistors in parallel. The transistor, having a high Re, will have a high input impedance at its base, so the biasing limits the circuit's input impedance. To fix that, insert a high value resistor between the two bias resistors and the base, and leave the signal voltage coming into the base. Put a capcitor to ground from the junction of the two bias resistors. This makes it into the so-called noiseless biasing circuit, which has a high input impedance, about equal to that inserted base resistor, and lower noise because you've isolated the current noise of the bias resistors from the base.

Ask questions - I'm tired of typing.

[Tim Escobedo]

Thanks, R.G. That's some great info.

I'm curious. On a FBC, how is the voltage swing affected if the neg AC feedback is decoupled? I've noticed in some situations, this can lead to more apparent gain at the output since there's essentially no neg feedback cancelling.

Also, on a FBC, I've found that in some situations, particularly at modest gains, a resistor base-to-ground can reduce distortion without really affecting gain, biasing or quiescent collector voltage. Why is this?

[will]

Hi,

R.G. That’s a great description of biasing bipolar transistors.  I use the 3rd "noiseless biasing for mosfets all the time. I haven’t used it too much for bipolar circuits although it does work well in the Blackfire circuit and the Range-Pig.

I'm curious. On a FBC, how is the voltage swing affected if the neg AC feedback is decoupled? I've noticed in some situations, this can lead to more apparent gain at the output since there's essentially no neg feedback cancelling.

My understanding is you have reduced (eliminated) the feedback for the audio frequencies yet kept the DC bias by decoupling the feedback. By doing this the gain will be set by the collector & emitter (possibly limited by the max hfe of the transistor).

Also, on a FBC, I've found that in some situations, particularly at modest gains, a resistor base-to-ground can reduce distortion without really affecting gain, biasing or quiescent collector voltage. Why is this?

Is this just a case of reduced clipping? Since it is still clipping the gain will largely remain the same.

Regards,
Will

ps: Tim, I enjoyed building a few of your circuits, especially the Fet bootstrapped driver. It made me understand how bootstrapping works. It also made me understand how the Fet Mini booster works. Thanks!

[Johnny G]

Thanks so much R.G. thats exactlly the sort of stuff i wanted to know.

i think im gonna go play with these a bit more on my circuit maker and ill probablly come back with a few more questions lol

thanks again

JG

[Boofhead]

With the emitter resistor (as shown) the second method is usually considered more stable because there are two feedback mechanisms (the two collector and emitter paths).  When you don't have an emitter resistor the two methods are about the same.

While the DC bias stability needs to be considered, both circuits usually have good enough bias stability in practice.  It is still possible to design circuits which have these circuit shapes yet have poor stability - an example is the first method where the emitter resistor is too small (and hence provides no DC feedback to stabilize the bias).

The most important difference between the two circuit is the AC characteristic.   The second method has *lower* input impedance and lower output impedance.  As a result in many applications the second method isn't really suitable to connect to high impedance input sources.

This article provides a reasonable explanation of the DC issues,

http://www.reed-electronics.com/ednmag/archives/1997/121897/26df_04.htm

Often design of this stuff comes to many specific requirements.

[gez]

I'm curious. On a FBC, how is the voltage swing affected if the neg AC feedback is decoupled? I've noticed in some situations, this can lead to more apparent gain at the output since there's essentially no neg feedback cancelling.

With decoupling of AC feedback the resistor between the collector and decoupling cap will now be in parallel with the collector resistor and gain is, approximately, the equivalent resistance of these two resistors, divided by emitter resistance (edit: internal emitter resistance of device plus any emitter resistor used that isn't decoupled).  As you said, you often get an increase in gain as negative feedback is reduced (emitter resistance still provides feedback).  However, linearity is poorer (especially at high gain).  Interesting results can be achieved in distortion circuits by partially decoupling AC feedback (smaller decoupling cap) to thin out the low end.

Also, on a FBC, I've found that in some situations, particularly at modest gains, a resistor base-to-ground can reduce distortion without really affecting gain, biasing or quiescent collector voltage. Why is this?

I'm glad you asked this Tim as it's puzzled me for a while.  I've found most of the info I needed to know about this circuit in text books but not this bit.  I think what might be happening is the extra resistor from base to ground forms a divider with the input resistor (if one is used) and therefore divides down the input signal, hence reduced clipping.

[RDV]

I've got aquestion about 'noiseless biasing'. Does it have the same benefits for an Opamp circuit as it does for a transistor of FET cicuit? I don't see it used much for OA circuits(which is all I've been doing lately). I used it for my own SYRPP/Mini-boo though cause of reading this: http://www.geofex.com/Article_Folders/modmuamp/modmuamp.htm.
I can not stress enough the importance of reading the articles at GEOFEX(to those of you who don't kow this already).

Regards

RDV

[gez]

I've got aquestion about 'noiseless biasing'. Does it have the same benefits for an Opamp circuit as it does for a transistor of FET cicuit? I don't see it used much for OA circuits(which is all I've been doing lately)

You see it all the time for op-amp non-inverting circuits, the resistor from the +ve input to the centre of the decoupled divider sets input impedance.  With transistor inputs there can be a significant voltage drop across this resistor (if large enough) so the output often biases up lower than expected.  This can be compensated for by using the same value resistor from output to -ve input (if it's a problem, it's often not that critical).

The same applies for trannie circuits.  The voltage divider may have to be set higher to compensate for the drop.  High gain trannies work better in this type of arrangement (or when bootstrapping).

[RDV]

On my HMP I've got a 10K/10K/100K arangement for the Vr. I was a bit gig-addled when I posted last night(@3:00am) and didn't realize that I already was using the 'noiseless biasing'. I wonder if I'd have been better off to use a 10K/10K/470K arangement however?

Regards

RDV

[Iwannalearn]

is it possible for someone to dumb down what biasing is all about for some of us newbs?

[R.G.]

is it possible for someone to dumb down what biasing is all about for some of us newbs?

I don't know that dumbing down is needed, just an intro to the foreign language that you're trying to learn.

Every electronic amplifying device has some active range over which it operates, usually from fully off (conducting no current at all) to fully on (or saturated; conducting as much as the external power supply and other components will let it).

Neither fully on nor fully off can amplify. We have to set the active device somewhere in the middle so it *can* amplify. This process of setting the active device to conducting a fraction but not all of the current it can is called biasing.

You can see some of this at GEO in "How It Works" where I draw some pictures to go along with it.

I'm going to do a made-up example.

Let's say we have a transistor that has a current gain of ten, and is running with a power supply voltage of 10V, and a collector resistor of 1K ohms (that is, one volt per milliampere of current) between its collector and the +10 power supply. The emitter is grounded, and we have put no current into the base.  

No current flows in the transistor at all, because its base current is zero, so its collector current is zero, and the voltage across the collector resistor is 0ma*1K = 0V. So the collector sits at +10V. If we dribble a little signal into the transistor base, it will be dramatically distorted because half of the signal will be trying to drive the transistor further off, which is impossible, off is off. Only the bits of signal trying to turn the transistor on have a chance.

Similarly, the max current we can put into the base without saturating the transistor is 1ma. At that currrent, the collector current is 10ma (gain of ten, remember) and the collector voltage is zero because the collector resistor has 10ma*1K = 10V dropped across it. Any signal we put into the base when the base is already conducting 1ma will not come through to the collector because the transistor is already saturated, and cannot pull any more current through the collector resistor.

The trick to get amplification here is to put a steady current of 0.5ma through the base. Now the collector conducts 5ma, the 5ma causes a voltage drop of 5V across the collector resistor, and any signal we put into the base adds to and subtracts from the steady bias current to be amplified at the collector.

Bias voltages and currents are what we provide to make the active devices work within their active region - or outside the active region when we want them to distort. Each active device has its own active region, and you have to know what the device does and what the rest of the circuit is around it to get it biased correctly.

Bias is an offset, somewhere between fully off and fully on to get the active device to amplify or distort in the way you want.

[Rob Strand]

is it possible for someone to dumb down what biasing is all about for some of us newbs?

Biasing simply means setting the DC voltages and currents for a device.

In each circuit the bias point you choose may vary widely depending both the device type and the type of circuit. For example the currents in the power transistors of a 300W amplifier will be somewhat different than those in the preamp.  For an amplifier you choose the bias voltages so the amplifier has a good voltage swing - you (usually) don't want the transistor to be biased fully on or off in an amplifier.

There is more than one way to achieve a design, as the above two circuits demonstrate.  Each circuit has it's pro and cons and it's up to you, the designer, to choose the best one for the job.  Choosing the best circuit isn't a simple choice and can involve finding a balance between many conflicting design decisions - this is a job for the pros.  Most common circuits are around for a reason, they have stood the test of time and are often good enough.

In principle that's all there is too it.  How well you achieve you goal is up to you - you have to study and experiment (there's no easy way out here).

[RDV]

To all Noobs:

READ everything at GEOFEX, AMZ, GGG, ROG, & this site. It's the only way, it has to be digested.
A little over a year ago I first ventured into the DIY waters, and now look at me! I know just enough to be dangerous! But I've built scores, and designed one, and come up with a few decent mods as well. By reading.

READ

RDV

[Eric H]

Location: Gone

Interesting place.
Nice to see you, Rob


-Eric

[gez]

Location: Gone

Interesting place.
Nice to see you, Rob


-Eric

You can change the name, but you can't change the style!  

[Rob Strand]

LOL! couldn't have said it better myself.......think I already did say that to someone......

Yeah I'm still still alive.....