0

I have a number of input lines (4-16 lines) on which RF pulses occur randomly on each line. I have to combine these lines into a single signal line. The pulses on each line are about 0.5ns wide and about 1V in amplitude and occur randomly. How to sum (combine) these lines into a single signal lne? The first approach that comes to mind is a summing amplifier based on an operational amplifier. However I do not know of any opamp able to handle such narrow pulses and then able to output a signal of several gigahertz. I was than thinking of building a summing amplifier mentioned in this thread: Op-amp can add more than two voltages, while discrete transistors can't? Again, even when using some good RF transistors designed for LNA amplifiers, the performance of this circuit is not good enough. Is there any other way of summing such narrow, high frequency pulses? The goal is to count the pulses (the sum of them, the total number).

Dictador
  • 26
  • 1
  • Can the pulses on different lines occur at the same time? If so, combining them into one line could cause missed counts. – Barry Jun 22 '23 at 12:23
  • Yes, the pulses on different lines can occur at the same time. That is not the problem as long as they trigger the counter. It would be a problem if two pulses occuring at the same time would result in nothing being counted. – Dictador Jun 22 '23 at 12:42

1 Answers1

2

You haven't said what frequencies are involved (you did say "output a signal of several GHz), what losses you can tolerate, etc. But here's a generic approach.

A combination of 2:1 or 4:1 combiners and RF amplifiers would do it. You would have to analyze the architecture to make sure you have enough gain to compensate for the losses in the passive combiners. Something like this:

enter image description here

SteveSh
  • 9,672
  • 2
  • 14
  • 31
  • The pulses are 0.5ns wide with much lower rise time. They appear randomly on each signal line. I do not need to preserve the shape of the pulses, they are just pulses that have to be counted. I use LNAs with a bandwidth of 10MHz-10GHz to amplify them. 0.5ns gives 2GHz max on each line. Assuming 16 lines, we are talking Fout=32GHz (32GHz is extreme case, real world we talk probably pulses with max frequency of 10GHz). It is another problem how to count them. – Dictador Jun 22 '23 at 12:30
  • The solution with n:1 power combiners is good. Simple but I didn't think about it. – Dictador Jun 22 '23 at 12:42