Why Use Kohm for Bias Resistors?

[ulnv4ever]

professors talked about this school but i forgot after the exam. Why do Op Amps and Transistors always use Kohm for bias resistors? does it have to do with the bias current or the input impedance? What happens if one uses ohms resistors to set the gain or bias transistors?

thanks.

 

[raybuzz]

I think it is to have a large input impedance. If the bias resistors used are say R1, R2, say in voltager divider bias, then the input impedence = (R1// R2// Rin), where Rin is the impedance of the active device. So if either R1, R2, are small, then the input impedance of the amplifier reduces, and is limited by the low value.

 

[unplebeian]

You have to look at the small signal model of a transistor and I think the input resistance is h_fe x R_base.

Also too much base current into the base can kill the transistor. So use k-ohms.


For opamps I think you have to look at the input offset voltages.

 

[berkeman]

It's often a tradeoff between quiescent power consumption and noise. If the bias resistors are small, then you are burning a lot of power in the bias circuit, which is unnecessary and wasteful. If the resistors are too big (100s of kOhm or MOhm), then you will generally get more crosstalk into your bias circuit from other noise sources, and thermal Johnson noise starts to degrade the fidelity of your analog signal.

 

[chroot]

In reality, almost no one uses discrete resistors for biasing anyway. Maybe for garage-built audio amplifiers, but that's about it. Even marginally high-performance circuits do not use them.

- Warren

 

[capnahab]

You sound like a newbe, in every electronic circuit, current is the defining parameter. First thing you do is make sure that you don't burn it up.

 

[berkeman]

You sound like a newbe, in every electronic circuit, current is the defining parameter. First thing you do is make sure that you don't burn it up.

It doesn't matter if they are new to electronics, it's a perfectly good question, and shows that they are trying to learn some of the intuition behind practical circuit design.

BTW, I thought of one more factor that goes into resistor value choices in circuits -- bandwidth. You will generally choose the largest resistor values (to minimize the quiescent current consumption that I mentioned before) that does not cause too much of a lowpass filter distortion of your signal. The lowpass filter is formed by the resistance and the parasitic capacitance in the circuit.

For example (and as a Quiz Question), if you have a 200kHz signal that you want to pass through an opamp circuit of some kind, and the input capacitance at the V+ and V- inputs of the opamp is about 20pF each, then what kind of impedances can you use in the opamp circuit without running into parasitic lowpass rolloff issues?

 

[capnahab]

I was only trying to get him interested and make him appreciate the good work he will do. Many times this work goes unappreciated and unrewarded but there is still the self-satifaction of doing the best you can by not tilting the pin ball machine, rocking the boat or jumping into a volkswagon bus full of crazed hippes, etc. Been there,done that.

 

[seang]

In reality, almost no one uses discrete resistors for biasing anyway. Maybe for garage-built audio amplifiers, but that's about it. Even marginally high-performance circuits do not use them.

- Warren

Why is this?
Just curious.

 

[seang]

Oh yeah, one more comment I have on picking biasing resistors. This past summer we were making some high frequency transistor amplifiers, and we really realized the importance of choosing good resistors. In addition to a lot of the topics already covered itt, like bandwidth, power, noise--you have to make sure that the resistors don't throw off your matching network too much! This is really only a high frequency consideration, but its important nonetheless.

 

[Corneo]

If the resistors are too big (100s of kOhm or MOhm), then you will generally get more crosstalk into your bias circuit from other noise sources, and thermal Johnson noise starts to degrade the fidelity of your analog signal.

Hi berkeman, can you elaborate on this crosstalk issue? When you say crosstalk, I am thinking in terms of RF.

 

[berkeman]

Hi berkeman, can you elaborate on this crosstalk issue? When you say crosstalk, I am thinking in terms of RF.

Sure. Consider two PCB traces on the top layer, separated by a few mm and running parallel for a few cm. Say it's a 4-layer PCB with ground plane as layer 2 (the inner layer adjacent to the top/outer signal layer). There is capacitance between the two traces that increases as the two traces are spaced closer together. The capacitive crosstalk increases as the traces are spaced closer together, and decreases as the spacing from the outer layer traces to the inner ground layer decreases.

The higher the impedance of the second trace, the higher the crosstalk voltage will be. It's an RC effect, so if the HPF crosstalk pole is at 1kOhm*1pF --> 167MHz, there won't be much crosstalk voltage. But 1MOhm*1pF --> let's lots more crosstalk noise get coupled across from trace-to-trace.

 

[deakie]

some very nice replies...also, to add...dont know why because berkman has done a good job covering the bases...

from a simple perspective then...

if your resistor valuers are low in the bias network, then the current is high and that current is enough to drown the incoming signal.
if the resistor network is too high, then the current is easily influenced by nearby fields which will induce odd flow on your bias current...hence your noise...

so at the end of the day, you must know the parameters of your incoming signal in order to design properly for its reception.
aim so that everything is a one to one relationship so that you get maximum power transfer...

idealy, it would be nice to have an high impedance fet in follower mode at the front end (for voltage signals)...building a filter into the back end of it should knock out most unwanted signals...
i tend to utilise + and - supplies to erradicate bias networks as much as possible...

 

[ranger]

Hi deakie,

Please pardon my ignorance. But I'm an undergrad and I've only done biasing where needed by using resistors for BJTs. Could you elaborate on the statement "i tend to utilise + and - supplies to erradicate bias networks as much as possible.", please.