I keep hearing that you shouldn't design a circuit around a transistor's β (BJT) or K (FET) value because they are often imprecise. But the equations I have for the circuit analysis require the use of β (BJT) or K (FET) to calculate the exact current.

So if not those values, what properties do you consider when designing a transistor circuit (other than maximum ratings)?

Thanks

 

Imprecise isn't an accurate description. A properly designed circuit will take beta range into consideration and will usually assume the minimum value.

Post a schematic for more specific answers.

 

Beta/hFE can vary wildly from transistor to transistor, say something simple like a 2n3904 will vary from a minimum hFE of 40 to 300, and on top of that vary wildly by collector-emitter current/voltage.

How I select a transistor depends highly on what kind of circuit I am building, if I just need a switching transistor I usually go by maximum current/voltage rating and peek at hFE to check it's not lower than I like(so I don't have to drive it with massive base current).

If it's for an amplifier it then depends on where in the amplifier, some stages need very high hFE, something like BC547C might be suitable(400-800 hFE) or if we need high voltage(say above 40v rails) then 2n5401(160v, but low hFE 30-240). If I need stages that need to be somewhat similar I try to test the hFE of the transitors at some fixed current and select similar valued ones, it's not perfect but weeds out those way off.

So the real answer (TL;DR): It depends.

 

What about a simple mosfet switching circuit? If i want an exact current to be passed, I need this equation to find the exact gate voltage required: \(i_d=k(V_{gs}-V_{th})^2\)

 

I'd limit the current with a drain resistor and drive the transistor into saturation. As you cannot reliably count on transistor to transistor variations, or even variations in temperature. It also simplifies the equation down to ohms law.

If you are trying to build a current sink which have to be somewhat accurate I'd run the current past a sense resistor to ground and feed that voltage to an opamp which then drives the mosfet(with matching voltage reference on the other input).

 

I'd limit the current with a drain resistor and drive the transistor into saturation. As you cannot reliably count on transistor to transistor variations, or even variations in temperature. It also simplifies the equation down to ohms law.

Doesn't the current through a saturated FET still depend on the gate voltage?

 

What about a simple mosfet switching circuit? If i want an exact current to be passed, I need this equation to find the exact gate voltage required: \(i_d=k(V_{gs}-V_{th})^2\)

For a switch you don't adjust the current with the Vgs, you use some external impedance to control that.
Normally you don't use Vth, for a switch, you just overdrive the MOSFET so that it is fully on, with a low impedance.
The Vgs for this is 10V for a normal MOSFET and 3v to 5V for a Logic-level type MOSFEWT.

 

Doesn't the current through a saturated FET still depend on the gate voltage?

To saturate a transistor more or less boils down to driving the gate/base to the point(and beyond) where increasing the voltage/current does not change the switched current through the transistor any more and the transistor can be said to be fully on.

 

At one stage I was compelled to use power transistors in parallel (with emitter resistors). I found that sometimes despite the emitter resistors, they still did not share the current equally. It was then that I took my transistor tester with me to the shop every time I needed to use transistors in parallel. The people were kind and allowed me to select the transistors using my transistor tester and the problem was solved, as I then could use transistors with equal gain together in parallel.
The published hFE of transistors vary from batch to batch and even in the same batch (this was revealed during those tests).

Bottom line: use your calculations, but get the REAL value for hFE.

Hope this makes sense!

Johann

 

Yep, thanks for all the help everyone!

One last question. As I understand it, a BJT has 3 operating regions where it is either closed, linearly amplifying, or open. A FET also has three operating regions that are closed, linearly resistive (ohmic/linear), and saturated. But that saturated mode still depends upon the gate voltage to regulate the Ids current (even though Vds is innefective).

Is there a point when a FET, like a BJT, acts like a completely open circuit?

 

Well th BJT can operate in cut-off (BJT is OFF, no current flow), active region and in saturation (BJT as a switch).
As for the FET we have two main types of a FET a MOSFET and a JFET.
And MOSFET can work in Cut-off, in a active region (saturation) as a amplifier and as switch in triode/linear region.

Also notice the terminology is different and quite confusing.
The "constant current" region is usually called the "active" or "linear" region for a BJT but is called the "saturation" region for the MOSFET.
The region where Vds/Vce in too low to sustain constant current operation is usually referred to as the "saturation" region in the BJT but is usually called either the "linear" or "triode" region in a MOSFET .

PS. Also you might find this interesting
http://forum.allaboutcircuits.com/threads/bjt-switch.91524/#post-666871
http://forum.allaboutcircuits.com/threads/transistor-circuit-design.77553/#post-547392
http://forum.allaboutcircuits.com/threads/bjt-saturation-question-s.81150/#post-577296
http://forum.allaboutcircuits.com/threads/fixed-gain-mosfet-help.69397/#post-481403