Finding loop gain using breaking the loop method.

break the loop method

Does anyone have any comprehensive material for finding loop gain using breaking the loop method.

Actually I am trying to find the loop gain, phase and stability of a Transimpedance amplifier (input current - output voltage) using this method. It consist of only a single transistor with a feedback resistor from collector to base.

I am using the cadence, spectre and try to simulate it but I do not know how to set up the circuit and where to input the test source, and where to probe the signal to get the relationship.

Re: break the loop method

QwErTzY said:

Does anyone have any comprehensive material for finding loop gain using breaking the loop method.

Actually I am trying to find the loop gain, phase and stability of a Transimpedance amplifier (input current - output voltage) using this method. It consist of only a single transistor with a feedback resistor from collector to base.

I am using the cadence, spectre and try to simulate it but I do not know how to set up the circuit and where to input the test source, and where to probe the signal to get the relationship.

Click to expand...

in razavi's analog book, there's chapter talking about feedback and how to break feedback to calculate the loop gain


in ur case, if the feedback reverse gain is not large, which is the general case (otherwise it's a lousy feedback network), u can use the following steps:

output voltage ==> sense voltage ==> shunt
input current ==> return current ==> shunt
==> Y-model
==> short feedback input (feedforward output) terminal to gnd
==> short feedback output (feedforward input) terminal to gnd
so loop breaked and u can calculate the open loop gain Ao with loading effect
feedback factor beta = i2/v1 when v2=0, where v2 is the feedback output terminal, v1 is the feedback input terminal and i2 is the feedback output current flowing into the feedback network

so loop gain = beta*Ao

regards,
c.m.

break the loop method

i THINK break at high impedance node then add a ac=1 signal , then you will find the gain.

break the loop method

in cadence u can do stb analysis and dont break the loop

Re: break the loop method

thx for all the help. I went to read up. i think understand the shunt shunt network.

let say, if we use stb in cadence, how will the set up be?
i just probe the output voltage?
what does the information means?

Re: break the loop method

from analog lib, take an iprobe and break the loop anywhere, insert that iprobe and perform stb analysis. i have seen results not accurate if u have more than one loop.

Re: break the loop method

designtech said:

from analog lib, take an iprobe and break the loop anywhere, insert that iprobe and perform stb analysis. i have seen results not accurate if u have more than one loop.

Click to expand...

Hi,
My bandgap voltage reference has two loops both positive and negative feedback. How can I measure loop gain in Cadence if iprobe is not accurate for more than one loop?

Re: break the loop method

anhnha said:

Hi,
My bandgap voltage reference has two loops both positive and negative feedback. How can I measure loop gain in Cadence if iprobe is not accurate for more than one loop?

Click to expand...

anhnha, I have answered the question in your "voltage reference thread"

- - - Updated - - -

QwErTzY said:

Does anyone have any comprehensive material for finding loop gain using breaking the loop method.
Actually I am trying to find the loop gain, phase and stability of a Transimpedance amplifier (input current - output voltage) using this method. It consist of only a single transistor with a feedback resistor from collector to base.
I am using the cadence, spectre and try to simulate it but I do not know how to set up the circuit and where to input the test source, and where to probe the signal to get the relationship.

Click to expand...

I cannot comment STB analysis because I have no experience with spectre.
Therefore, a general comment:
In opamp circuits, in most cases we have a low impedance output connected with an impedance which is much larger. Therefore, without causing loading errors we can open the loop for test voltage injection at this node (alternatively, at the high impedance input).
In most cases, this principle can be used also - within acceptable error limits - if a BJT feedback loop is closed using a resistor RB between collector and base - as long as the value of RB is at least (20...50) times larger than the collector resistor RC. (see remark below)
However, a more accurate method is described by MIDDLEBROOK (google this name). According to this "double-injection method" you perform TWO ac simulations: The first one with an inline ac voltage source BETWEEN the opening nodes and a second one with an injected ac current. Then, a combined formula is used to correct the error introduced by the loading error caused by opening the loop.

Remark: For this method, it is important to have the normal operating conditions, which means: The normal signal input source must be connected and set to zero volts (Vac=0) .