double range pulse amplifier

[sdouble]

Hi all,
I do need to amplify and filter photomultiplier pulses. Frequencies involved in the pulse are in the [1;170] Mhz range. Repetition rate is moderate (say, 10 kHz).
From 1 photomultiplier output I would like to generate 2 voltage signals.
1)copy of the input, only low-pass-filtered. Opamp based 2nd order  Sallen-key topology, Bessel filter foreseen.
2)amplified output : X10 gain. Same frequency range. This is not straight forward I think due to the high gain bandwidth requirement.

Last but not the least : the amplified output could  get saturated. I planed to use -5 V and 5 V rails. Input signal could be of 1 volts. In case of saturation of the amplified output, I need to avoid the saturation of the other (non amplifed) output.
Do you have any schematics in mind which could fulfill my needs ?
best

[David Hess]

The requirements are not impractical but I suspect there will be a problem with the x10 amplifier recovering from saturation unless this is taken into account in the design.  If it is a problem, then the input signal level is high enough that a current biased diode bridge at the input of the amplifier can operate as a very fast and accurate clamp to prevent this.

[Neomys Sapiens]

For actual suggestions, even if not detailed, you left out some (rather many) parameters.
Like impedance present at the outputs, need to be short-circuit proof or not, and also some information related to the time-domain behaviour, such as total delay of the circuit and if a delay difference between the outputs is allowed.

[sdouble]

thanks for your help
here are some clues

For actual suggestions, even if not detailed, you left out some (rather many) parameters.
Like impedance present at the outputs,

The photomultiplier is 50 ohms  adapted.

need to be short-circuit proof or not,

that is not needed.

and also some information related to the time-domain behaviour, such as total delay of the circuit and if a delay difference between the outputs is allowed.

This is not crucial too. timing is anyway calibrated based on actual data. Total delay will be of a few ns I guess which is clearly not an issue.

[sdouble]

here is what I had in mind


http://]