Common-source BJT amplifier with fixed gain despite HFE changes?

[allanvv]

Is there a simple way to get a fixed DC gain in the range of 5-8 with just a single transistor or maybe two? I'm trying to convert a 0-3.3V DC analog signal into approximately 0-12V with as few components as possible, and I don't need high accuracy.

I tried adding a feedback resistor from the collector output back to the base but it doesn't help any. The gain seems to be proportional to the hfe of the transistor which apparently can have huge variation.

 

[ninju]

DC gain using transistors. I don't think it is possible in the way, because a DC biases a transistor. Changing it would change the quiscent point.

You can use transistors, in a boost converter for DC-DC conversion.

 

[leo_o2]

Pls try charge pump for fixed gain.

 

[KerimF]

Is there a simple way to get a fixed DC gain in the range of 5-8 with just a single transistor or maybe two? I'm trying to convert a 0-3.3V DC analog signal into approximately 0-12V with as few components as possible, and I don't need high accuracy.

May I ask you what your supply voltage is?
Usually by using transistors, there will be two non linear regions; one nears the ground and another near Vcc.
It seems your input signal varies from 0-3.3V. If I understood you well, you like to amplify the DC signal to 0-12V.
That is the DC gain required is 12/3.3 = 3.64.
But you said also that you look for a gain between 5 and 8. So which gain is the right one for you?

In any case, I attached a possible solution just to give you an example. Perhaps you will like to change it as convenient for your application.

Kerim

 

[allanvv]

Hi, my supply voltage is 12V. I don't actually care about the precise gain, but I would like to guarantee hitting the cutoff and saturation regions. My purpose is just to extend the linear region, since normally with a common-emitter amplifier it has a very sharp characteristic. In the end I'm going to have a slow outer digital control loop (that I get for free) acting as a very crappy linear voltage regulator.

I'm trying to do it in only one or two transistors. With only one in common-emitter configuration, I can design it to work for HFE's varying in a wide range, with the linear region proportional to the hfe, but I was wondering if there was a way to get less dependence on HFE by adding another transistor.

Maybe an NMOS is better, because then threshold voltage variations only shift the voltage transfer curve left or right, but the gain and linear region size stays fixed.

 

[KerimF]

By using one transistor, the output will be inverted. I added Q2, Q3 and Q4 as a current mirror so that the load can be grounded and the output signal will have the same polarity of the input one. As you know, using a rail-to-rail opamp will be a good solution, but you like to do it with transistors. You are right, perhaps using NMOS is better but also the output will be inverted if using one transistor... right?

 

[keith1200rs]

Be careful with circuits which rely on matching for current mirrors when implemented with discrete devices. You circuit will simulate ok because Q2 & Q3 are perfectly matched but in real life they won't be.

Keith

 

[KerimF]

you are right Keith... He likes to use transistors and he said

and I don't need high accuracy

so it was just an example which is far from being perfect since its voltage gain is non-linear close to ground and the ouput doesn't reach 12V.

 

[keith1200rs]

My comment was really about the fact that a current mirror with discrete devices will be very poor - try simulating with a 10mV offset in the mirror.

Keith

 

[KerimF]

Please don't worry... no one will build it for real ;-)

 

[keith1200rs]

I thought the whole point of the question was for a circuit that can be built? What is the point in posting circuits that won't work?

Keith

 

[KerimF]

Please forgive me Keith... I won't do it again...
It seems all what you do in life is perfect... I don't...

After your third posts, I got your point, my place shouldn't be with a perfect *** as you are.
Unfortunately, I am just a human being, so I can't avoid doing mistakes once a while.

And thank you for not insulting me too since you are the boss here.

I promise you, you will not see me here doing such a crime again, meanwhile keep doing well as you used too.

Kerim

 

[keith1200rs]

I am not the boss here and my posts are as a user, not a moderator, but I always do try not to misguide people.

Keith

 

[ServoAmp]

With as few components as possible!
how about 3 components.
2 resistors and 1 opa544
the opa544 only has 2 more pins than the BJT and they connect directly to power.
the two resistors set the gain.

https://focus.ti.com/lit/ds/symlink/opa544.pdf
/