Build an op amp with three discrete transistors
This simple ac-coupled design inverts, runs off a single supply, and is self-biasing.
Lyle Russell Williams, St Charles, MO; Edited by Paul Rako and Fran Granville -- EDN, December 1, 2011
Resistor R3 and capacitor C1 form a
compensation network that prevents
the circuit from oscillating. The values
in the figure still provide a good square-wave
response. The ratio of R2 to R1
determines the inverting gain, which is
−10 in this example.You can configure this op amp as an
active filter or as an oscillator. It drives a
load of 1 kΩ. The square-wave response
is good at 10 kHz, and the output reduces
by 3 dB at 50 kHz. Set the 50-Hz low-frequency
response with the values of
the input and the output capacitors. You
can raise the high-frequency response
by using faster transistors and doing
careful layout.
Talkback
-
To calculate open loop gain needs to use transistors transconductance values multiplying by resistors load for each stage, not use Beta.
So you will see that about 120dB open loop gain is avaliable.
Vladimir Doubovis - 2012-10-2 20:21:34 PST -
In this case, for a CL gain of -10, you only need OL gain of 990 to achieve 1% accuracy. The very high OL gain is needed if you want accurate CL gains that are much higher.
David Albean - 2011-19-12 10:14:06 PST -
i guess 120dB is defined by how well the -10 gain is obtained in respect to R2/R1 ratio....
jkl jkl - 2011-18-12 13:55:47 PST -
For fun I set out to see what it might take, without a drastically altered topology, to get to the advertized 120dB gain. I decided, since power is being squandered anyway, to put a good deal more current through the first two devices. And figure that the source impedance could be a lot lower, and selected a feedback R of 10k based on a 100mV drop due to input Q base current. And in addition to assuming heavily bypassed zeners in the emitters of the first two devices to allow for internal loads to be larger resistance for given collector currents, I changed the output device to a DMOS enhancement mode part (BS170), with its on threshold again providing a local bias voltage.
These strategies give one a lot of help. But to get to a midband gain of a million or so, I applied positive feedback from the output to the collector of the first device. Things get really shaky when the gain is about 2 million. And I'm not going to show a schematic as these things have a way of taking on a life of their own, and I don't need a liability suit from someone who may claim I recommended this :)
Now these results are from driving the contraption from a voltage source with 1 farad in series, and a 10k feedback R for bias. The -3dBr point at high frequencies is about 24kHz, a nominal GBW product of 48 GHz. Needless to say this will not be stable at low gains!
Brad Wood - 2011-9-12 14:36:11 PST -
Yes - that agrees a lot better with my calc of OL gain = -4300. Certainly not 1 million - as stated.
Dave Albean - 2011-7-12 11:42:33 PST
