Mixed Power Voltages in unipolar linear transistor amp?

BobTPH

Joined Jun 5, 2013
3,126
You can make a decent small signal AC amplifier with one transistor. But a DC amplifier is an entirely different thing.

Bob
 

Thread Starter

johnyradio

Joined Oct 26, 2012
282
Before rejecting op amps as noisy, you might want to consider how much noise you get from a 1000 ohm resistor.
Fantastic observation!

For my purpose, tbh, I don't care about noise. I'm doing a switching application, so I think noise is irrelevant. More important is cost, parts count, and number of solder points. And getting close to the metal, ie all things being roughly equal, prefer discrete component to black box.

Thx
 
Last edited:

crutschow

Joined Mar 14, 2008
27,017
What's op? o_O
For my purpose, tbh, I don't care about noise.
Then why did you mention it earlier if it's a red herring? :confused:
I'm doing a switching application, so I think noise is irrelevant. More important is cost, parts count, and number of solder points. And getting close to the metal, ie all things being roughly equal, prefer discrete component to black box.
Depends completely on the application and switching requirements.
 

BobTPH

Joined Jun 5, 2013
3,126
Surprising, I would have thought AC would be harder.
AC is easier because transistor amplifiers need a bias voltage to operate. You just stick capacitors at the input and output, and this removes the constant bias voltages. For a DC amplifier, they must operate with a bias of zero, which you cannot do with a single transistor circuit.

Bob
 

Thread Starter

johnyradio

Joined Oct 26, 2012
282
AC is easier because transistor amplifiers need a bias voltage to operate. You just stick capacitors at the input and output, and this removes the constant bias voltages. For a DC amplifier, they must operate with a bias of zero, which you cannot do with a single transistor circuit.

Bob
ok, here's a thought. My signal is a positive-only staircase. My desired output is a positive-only staircase of exactly twice the amplitude. Perhaps i could use a coupling cap at the input, allowing me to do AC amplification. Then, restore the DC offset at the output, giving me a the positive-only signal i want.
 

cmartinez

Joined Jan 17, 2007
7,313
You won't do that with a single transistor. Use a friggin' op amp, despite what you learned on someone's youtube video.

Bob
I agree. I too like to understand the "discrete logic" of circuits, but there comes a point in which using an ic is not just practical but realistic too.
 

Thread Starter

johnyradio

Joined Oct 26, 2012
282
Getting a bit closer. I added a clamper to this JFET common-source amp. I'm assuming the slight amount of negative swing is due to the diode voltage-drop?
http://www.circuitstoday.com/diode-clamping-circuits
https://en.wikipedia.org/wiki/Clamper_(electronics)


(play with sim here) http://tinyurl.com/y95j8cw6

adding a pullup resistor brings us closer yet. Actually, this bring the whole wave positive, so i'm wondering what are the drawbacks?


(play with sim here) http://tinyurl.com/y6ufn728

Alternatively, let's say i don't need any gain. Let's say i want a 1-transistor circuit that will clamp or offset an AC signal positive. How would that be done? Surely that's a simple affair. (Note, my input is a fixed 5V peak-to-peak, no need to handle larger or smaller signals.)

This config is simpler, and gives a positive-only output. Seems JFET gives positive-only output by default! That could be useful, since positive-only is my goal. The output peak-to-peak is 5v to 10V, while the input is -2.5 to + 2.5. So, not quite there-- want 0V trough. But getting closer.



Thx!
 
Last edited:

ebp

Joined Feb 8, 2018
2,332
" ... adding a pullup resistor brings us closer yet. Actually, this bring the whole wave positive, so i'm wondering what are the drawbacks?"
What happens to your circuit when you load the output with finite resistance?

"Alternatively, let's say i don't need any gain."
You mean you want a gain of +1? How close to unity does it have to be? You aren't going to get 1.000. Time after time you have tossed out requests and never once actually attached any sort of reasonable specification to requirements.

You will never get precise DC stability from any single transistor circuit that actually does anything. There will always be some measure of unit-to-unit variation. There will always be variation with temperature. There will always be a degree of sensitivity to supply voltage. These things may be tolerable.

Design of transistor amplifiers that must deliver any reasonable precision is all about taking most of the characteristics of the transistor(s) "out" of the circuit. Even Bordodynov's multi transistor circuit, which does much of that, still requires precisely matched transistors in the current mirrors to produce good DC performance. They must not only be very well matched initially, but they must maintain the matching with variation in temperature, which means they must be very tightly thermally coupled to each other. (Simulators assume any number of transistors of the same type number behave identically unless the simulator is told to include the variations - if they exist in the models. They assume the transistors are at the same temperature unless told otherwise.) Neither of those things is possible with individual discrete transistors. Multiple monolithic transistors can be pretty good, but the reality is that they may still require additional resistors in their emitter circuits to partially compensate for the mismatch. This improves performance in some ways and degrades it in others. What is acceptable depends on the requirements - which is something of a tautology. Designers of (not with) op amps go to great lengths to produce amps that have highly predictable behavior, and they succeed - depending on requirements. They succeed by using a whole lot of transistors to do a superficially simple task. With really good op amps, designers with them must be very careful not to degrade performance though sloppy design and component selection.
 

Thread Starter

johnyradio

Joined Oct 26, 2012
282
@ebp when I said "what if I don't require gain?", I didn't mean "i require a gain of exactly 1."

I simply meant, "what circuit will give me positive DC offset, gain not required." Some gain or loss is acceptable.
 
Top