BJT voltage divider biasing - how to decide on the value of collector current?

Thread Starter

atferrari

Joined Jan 6, 2004
4,069
After revising the procedure to calculate values for voltage divider biasing of a BJT amplifier, I couldn't find a concrete indication of how to decide on the collector current.

The "too much" limit is obvious, but, what about the other extreme of a reasonable minimum?

My question: how to decide on the Ic for:

amplifiers
buffers
audio oscillators
RF oscillators?

Thanks for any help you could provide.
 

LvW

Joined Jun 13, 2013
1,078
One important criterion in selecting a "proper" collector current Ic can be the desired voltage gain.
This is because the gain depends on the transconductance gm=d(Ic)/d(Vbe), which tells you how the current Ic is varied for a corresponding variation of Vbe.
Resulting from the exponential characteristic Ic=f(Vbe) the transconductance depends linearly from the voltage Vbe:
gm=Ic/Vt (Vt=temperature voltage)
 

MrChips

Joined Oct 2, 2009
21,649
I would have guessed that it depends on the load.
For example, if you want a 1V swing in 1kΩ load then you would need 1mA.
Rule of thumb says to go with x10 that amount, hence 10mA collector current.
 

LvW

Joined Jun 13, 2013
1,078
I would have guessed that it depends on the load.
For example, if you want a 1V swing in 1kΩ load then you would need 1mA.
Rule of thumb says to go with x10 that amount, hence 10mA collector current.
Yes - but I didn`t say that the gain would depend on the ransconductance ONLY !
If you start with "swing" as a main requirement - I agree. But "gain" is very often the main criterion.
(My formulation was: "One important criterion....can be...)
 

LvW

Joined Jun 13, 2013
1,078
Yes - but I didn`t say that the gain would depend on the ransconductance ONLY !
If you start with "swing" as a main requirement - I agree. But "gain" is very often the main criterion.
(My formulation was: "One important criterion....can be...).
More than that, the OP has asked for the "BJT voltage divider biasing" - and this concerns the relation Ic=f(Vbe).
 

OBW0549

Joined Mar 2, 2015
3,504
The "too much" limit is obvious, but, what about the other extreme of a reasonable minimum?
"How low can you go?" is an interesting question.

Lower collector currents mean higher resistance values in the surrounding circuitry, and this means that stray circuit capacitances, as well as semiconductor junction capacitances, have more influence at high frequencies. The result can be degraded high-frequency response in an amplifier (and, coincidentally, longer rise and fall times in a transistor switching circuit). So the minimum design collector current value can in part be dictated by the upper frequency limit your circuit needs to operate at.

This same consideration applies to operational amplifiers as well, both in the op amp's application circuit and within the op amp itself. Micropower op amps tend to have lower gain-bandwidth products and slower output slew rate limits; high-speed op amps tend to be power hogs. My favorite example of the former is the LT1494, which operates on 1 μA of supply current-- and has a gain-bandwidth product of only 2.7 kHz (in contrast to an LM358, which has a GBW of 1 MHz).

So "lower means slower" is good to keep in mind.
 

Thread Starter

atferrari

Joined Jan 6, 2004
4,069
One important criterion in selecting a "proper" collector current Ic can be the desired voltage gain.
This is because the gain depends on the transconductance gm=d(Ic)/d(Vbe), which tells you how the current Ic is varied for a corresponding variation of Vbe.
Resulting from the exponential characteristic Ic=f(Vbe) the transconductance depends linearly from the voltage Vbe:
gm=Ic/Vt (Vt=temperature voltage)
I would have guessed that it depends on the load.
For example, if you want a 1V swing in 1kΩ load then you would need 1mA.
Rule of thumb says to go with x10 that amount, hence 10mA collector current.
IOW, I need to know the input impedance of the following stage to then decide on the Ic of interest. Entering uncharted waters now... :oops:
 

OBW0549

Joined Mar 2, 2015
3,504
Entering uncharted waters now... :oops:
There are a lot of those in electronics.

People (especially students) often think that everything in circuit design can be calculated, and that there is a formula for everything. In fact, there is a lot in design that comes down to experience and judgement and compromise between competing design requirements (or desiderata). This can be sometimes hard for newbies to take, as for the poor unfortunate fellow a few months ago absolutely insisted we provide him with a formula that would let him precisely calculate the correct value for a pullup resistor for TTL gate inputs.

Unless you are working at frequencies well above audio, or have an application that requires long battery life, you can probably use Ic = 1 mA as a starting point and not go too far wrong (OBW Rule Of Thumb #2).
 

DickCappels

Joined Aug 21, 2008
6,532
Many transistor parameters are tied to collector current. For example, in an RF application you might want to run at a rather high collector current to get higher Ft , or to get faster switching in a switching circuit. Take a look at the charts in the datasheet of a well-characterized and flexible transistor like the ON Semiconductor 2N3904 (attached).
 

Attachments

Thread Starter

atferrari

Joined Jan 6, 2004
4,069
There are a lot of those in electronics.

People (especially students) often think that everything in circuit design can be calculated, and that there is a formula for everything. In fact, there is a lot in design that comes down to experience and judgement and compromise between competing design requirements (or desiderata). This can be sometimes hard for newbies to take, as for the poor unfortunate fellow a few months ago absolutely insisted we provide him with a formula that would let him precisely calculate the correct value for a pullup resistor for TTL gate inputs.

Unless you are working at frequencies well above audio, or have an application that requires long battery life, you can probably use Ic = 1 mA as a starting point and not go too far wrong (OBW Rule Of Thumb #2).
Many transistor parameters are tied to collector current. For example, in an RF application you might want to run at a rather high collector current to get higher Ft , or to get faster switching in a switching circuit. Take a look at the charts in the datasheet of a well-characterized and flexible transistor like the ON Semiconductor 2N3904 (attached).
Thanks you both for the replies.

It all started when I tried to somehow replicate a design of a Colpits oscillator in another forum, in the 2 MHz range, followed by a buffer. Realized that I had to start afresh with transistors and found necessary to carefully revisit biasing. Long time ago I learnt that wildly tweaking designs found somewhere is not the way.
 

Papabravo

Joined Feb 24, 2006
14,231
If you look at a set of characteristic curves and swag a load line from Vce = Vcc on the horizontal axis to an Ic equivalent to a value where the knee of the top most curve is and pick a Q-point at the midpoint of that line that will give you a place to start. This implies a certain value for the load which may or may not be the case. That is the way I would approach it.
 
Last edited:
Top