Pre-amp Distortion Comparison Between Sine Wave, Square Wave and Arbitrary Wave

[The13thParish]

Hello everyone,

I'm testing a pre-amplifier that may be causing distortion to our signals being amplified from the back of a micron channel plate (MCP). Well, something is causing distortion so I may as well start with the pre-amplifier.

The pre-amplifier is a Phillips Scientific Model 6954, wide-band 100 KHz to 1.8 GHz pre-amplifier with a gain of 100.

Because I can't run the MCP at the moment (coronavirus), I'm using a Tektronix AFG310 function generator to send various signals into the amplifier. I'm using the lowest amplitude setting (50 mV) and attenuating this using a Crystek 20 dB attenuator directly attached to the function generator output. By my maths that gives an amplitude of 5 mV for all my functions. All connections and cables are 50 ohm.

With a sine wave at 100 kHz, I get the correct output on my oscilloscope without the pre-amp, all good: the 5 mV amplitude sine wave shows up at 10 mV_p-p as required. When I add the pre-amp in-line (and attach the attenuator now to the output of the amplifier) I get the correct 1 V_p-p output. 

If I switch over to a 100 kHz square wave with the same 5 mV amplitude, the function looks good on the scope, but forms some odd distortions when the pre-amp is in line, on both the rising and falling edges. What is the exact cause of this distortion?

I'm thinking that perhaps the amplifier cannot handle the sharp rising edges of a square wave.  There's a custom function previously saved into my generator that has a smooth but fast rising edge and a smooth, slow falling edge. It's a 10 kHz wave, however my assumption was that even though the amplifier specs are for frequencies between 100 kHz and 1.8 GHz, my custom input wave might not show the full gain, but won't be distorted by the amplifier. It comes out however grossly distorted by the amplifier and I also have to adjust the trigger hold-off in order to get a clean trace.

I'm now thinking that this pre-amp can only handle functions with positive and negative going features, rather than a function which is only strictly positive.

Clearly I don't know how pre-amps work, so I should probably clear this up before I troubleshoot ANYTHING. Any tips for clearing up my confusion? All the relevant function traces are attached in a somewhat crowded image.

(sending me to the appropriate reading or textbooks is a sufficient answer!)

All the best,
Ryan

[hamster_nz]

Not saying this is your problem, but here is a square wave made up of 1st to 15th harmonics, and the same harmonics with the 1st harmonic delayed by 0.7 (~ 45 degrees)

[The13thParish]

@hamster_nz,

That's certainly very interesting! It would not surprise me if this was the cause seeing your computation, but why and what caused it!

All the best,
Ryan

[hamster_nz]

@hamster_nz,

That's certainly very interesting! It would not surprise me if this was the cause seeing your computation, but why and what caused it!

All the best,
Ryan

Given that 100kHz is the bottom end of your bandwidth, I suspect a high-pass filter on the input is causing different phase change for different harmonics. See if 500kHZ squares look OK?

[David Hess]

That is a typical high pass response when the test frequency is too low.  A sine wave, being only one frequency, does not show the amplitude and phase distortion while a square wave or other waveform does.

If the square wave frequency was lower, then you could roughly calculate the high pass cutoff frequency from the transition time.

[Marco]

Why are you feeding a 100 kHz lower cut off amplifier with such low frequency signals? Shorten the function duration to microsecond range or less and up the repetition rates ... or get a DC coupled amplifier.

[Vovk_Z]

Yes, with high-pass 100 kHz input filter even 1000 kHz square wave will be a bit distorted. It have to be about fine only starts from about 20x that cut-off frequency (from about 2 MHz square wave).
Look here:  Practical Techniques of Square-Wave Testing. July 1957 Radio & TV News Article or here: Correcting Low Frequency Phase Distortion  (I googled briefly).
Or here: Square-wave Testing. Electronics World, Mar. 1966

[The13thParish]

@Vovk_Z,

This is great practical advice, many thanks!