🚀 go-pugleaf

***One full page of noise floor & feedback loop calculations, at the end.***

In article <1992Apr9.184230.13126@acuson.com>, kumar@acuson.com (Kumar Neelakantan) writes:
> I am looking for a design or names of companies/consultants who could design
> an RF preamplifier with the following specs:
> 
> Gain: 30 dB
> Input Signal Dynamic Range: 2 mV - 1.2 V pk-pk ( ~55dB )
> Max Output Voltage: 36 V pk-pk
> Frequency Response: Flat (100 KHz - 10 MHz) with +- 0.1 dB passband ripple
> 			- 3dB Frequency: 20 MHz
> Impedance of input load connected to preamp: 0.5 - 1 Mohm, 3-5pF
> Output Impedance: 50 ohms	
==>  [ 18volts peak/50 Ohms = 300 milliamps ]
> Output Spectral noise density: 0.1 microVolts/sqrt. Hz with input
> of preamp terminated in 50 ohm load  [
==> The thermal noise in one Hertz Bandwidth is -174dBm. 
  Given 30dB gain, we have -144dBm noise power per Hertz at output.
  Since 0dBm is 223millivolts RMS @50 Ohm the required 0.1microvolts is a
  factor of 2.2 million weaker, or -127dBm. We are thus allowed an amplifier
  noise figure of -144 to -127, or 17dB, at 50 Ohm source resistance.

  You provide no phase response requirements, or distortion requirements,
  or short-circuit requirements, or what the reactance of the load will be.
  There is a conflict between the 3DB point of  20MHz, and only being
  down 0.1dB at 10MHz.

  To get a response accurate to 0.1dB, given 30dB closed loop gain,
  (assuming 0.1dB is about 1%, in absence of remote node 'rnews' 
  allowing separate SHELL to run LISP to compute 0.1dB)
  we need open-loop gain of 30dB + 20*log10( 1/1%) = 30 + 40 = 70dB, or 3100.
  Given a 10MHz max freq to hold the 0.1dB, ignoring any benefits of
  phase shift in generating error signals, we need GainBandwidth of
  31GigaHertz. If done by one feedback loop.

  Probably should use two loops. Try closed-loop gains of X10 and X3.
  The X10 only outputs 6 volts peak-peak. The X3 outputs the 18volts PP,
  and must include the time-delay/phase-shift of a 300milliamp driver.

> If you have any info call me at (415)694-5722 or send me e-mail to
> kumar@acuson.com
> kumar neelakantan 


Allen Sullivan
Designing RF/video circuits for Go-Video, Scottsdale AZ
CS PhD student Arizona State Univ
Developing LISP methods for analog circuit synthesis