An almost-GHz active differential oscilloscope probe
What I really wanted was an active differential probe which should be able to look at a 500MHz signal at a couple of volts or so with decent fidelity. To do that I wanted a probe with about one GHz of usable bandwidth. I have an old Tektronix 11801B scope with two 20 GHz sampling heads with 50Ī© single ended inputs that can handle signals which are 1.0V peak-to-peak so thatās what I wanted the probe output to be compatible with. The probe must attenuate the signal so to not overdrive the inputs and to make things easy to calculate I decided on letting the probe have a 1:10 attenuation.
I have used Tina-TI before and am fairly familiar with it so a natural beginning was to try to simulate the circuit I had in mind in Tina.
Update: An earlier version of this schematic showed the output of U3 being grounded. That was an error and the ground connection has been removed.
As pointed out by Elliot Williams this is fairly traditional instrumentation amplifier built with separate OP-amps.
I tried a bunch of different OP-amps from the TI libraries but settled on the LMH6702, a ā1.7-GHz Ultra-Low Distortion Wideband Op Ampā. It has a high input impedance (1MĪ©) when used in a voltage follower configuration and has about twice the bandwidth i need.
The input signal is simulated by a voltage generator (VG1) which outputs a square wave which is then fed to two resistors (R1, R2) centered around a constant voltage source (VS1). This is intended to simulate a 1V LVDS clock signal biased by 1V being fed into a differential termination.
The first stage is a pair of of OP-amps (U1, U2) in voltage-follower configuration. I was a bit worried about oscillation so instead of connecting the voltage follower positive input and feedback directly Iāve added some inputs resistors (R3, R4) and feedback resistors (R5, R6) to avoid this. VM1 and VM2 are measurement points that measure the outputs of the OP-amps.
The second stage is a differencing amplifier (U3) with a less than unity gain (set by R7, R8, R9, R10) which gives a 1:5 attenuation. Iāve read that using an OP-amp to attenuate a signal is supposed to have problems with some OP-amps for reasons having to do with to noise feedback, but it worked in simulation so it should work in real life, shouldnāt it?
Finally thereās a a 50Ī© resistor (R11) for impedance matching with the output which together with the 50Ī© (R12) termination at the scope input attenuates the signal by 1:2 giving a total attenuation of 1:10 for the whole probe. VM3 measures the signal at the scope input.
VS2 and VS3 are the +5V and -5V voltage supplies for the circuit.
DC performance in simulation looks good. +/-3V at the inputs (X axis) become a +/-300mV signal at the scope (VM3). VM1 and VM2 show the signals after the first stage.
AC performance looks decent. A 1GHz signal signal will be attenuated by 30dB, but a 500MHz signal ought to still be visible on the scope. The phase error doesnāt look to bad, by the time it becomes large the signal should be so attenuated that it doesnāt matter.
A transient simulation of a 100MHz square wave at the input shows that the signal will be attenuated by 1:2 on top of the baseline 1:10 attenuation. Most of the fine details in the signal have disappeared but it should still be possible to measure the frequency and give a rough idea about the quality of the signal.
At 500MHz all detail has been lost and weāre basically left with a weak sine wave.
For More Details: An almost-GHz active differential oscilloscope probe