bandwidth drop with cascaded limiting amplifier

[Ipanema]

Hi,

Can anyone explain to me why the bandwidth of limiting amplifier(LA) get reduced when cascaded? e.g. single stage LA, BW=10GHz, but with 5 stages of the same LA, BW=2GHz.

Thanks.

 

[zigzagfish]

of cos when you cascade the 5 the same amps, whose main poles would be the same and aroud 10G as well as other coresponding poles, the overlapping of the poles would decrease the bandwidth significanly, you can see it clearly from the frequency response graphics

 

[Ipanema]

Hi,

From my simple understanding, if there exist 5 poles exactly at 10Ghz, I should get 100dB/dec at 10Ghz compared to 1 pole at 10Ghz, which will give only 20dB/dec. The -3dB points will reduce a little bit, but not as much as 2Ghz. I would like to know what make the pole shifted in this case.

Thanks.

 

[bmalp98]

Hi,

Bandwith is point where gain drops by 3dB. Cascading 5 stages makes 15dB..

Or, more mathematically, take (A/(wc+jw))^n, bandwidth is where imaginary and real part of the denominator are equal. Develop in Taylor series for the dominant pole, you get
wc^n + n*wc^(n-1) * jw. Equating real and imaginary part:
w = wc/n

Thats what you see.

bmalp98

 

[surianova]

my opinion is when a few limiting amplifier cascade together, the gain will increase from first stage to the last stage. As gain increase, the miller capacitance also increase according to formula cgd(1+A). This will form a even lower frequency pole as incresing stages, f=/2pi. RC. The output of the last stage will be
the dominant pole or the first pole.

Suria

 

[bmalp98]

Hi,

I don't think so. What you call Miller capacitance has to be connected from the (inverting) output to the input of the amplifier. Physically, this is the base-collector or drain-gate capacitance. Since these capacitances "see" only the local gain of a single stage, there is no increase, right?

Best,
bmalp98