A question about interpreting and reducing high frequency noise in a low level square wave

As a part of a circuit the output section of the circuit is as follows(C1 is actually 10u):

The input is a 1kHz 0V…5V pulse train coming from the output of a 74HC74, and the output is a square wave around 10mV pk-pk.

Here are the input signal in time domain and its FFT:

And here are the output signal in time domain and its FFT:

I have questions about interpreting and reducing noise here.

1-) The FFT of output signal has two high freq. spikes. How can we interpret it? Does that mean it has 500kHz and 750kHz high freq. noise?

2-) Is it possible to eliminate those high freq. noise by not effecting the square wave form much? The output will be sampled with an ADC less than 100kHz sampling rate. So do I still need to eliminate that high freq. noise

• The noise could be at output circuitry, analog psu, ADC noise. May 31 '17 at 11:44
• This is going into a sound card input that's designed for audio? Then the inputs should already have anti-aliasing filters designed to remove anything over around 20 kHz. Filters aren't brick walls but 500 kHz is so far off the signal range expected by the input that it should be completely removed. Assuming they haven't saved money and not put any anti-aliasing filtering on the input. May 31 '17 at 12:03
• Yes to soundcard jack or smartphone jack I see makes sense if they already take care of noise May 31 '17 at 12:07

It really depends on where the noise is coming from..it could just be scope noise.

Divide all your resistors by 10 will help.... and I'm not sure why you think you need the diodes with such a controlled signal source.

If need be you can add a cap to the output to limit the frequency @ x10 your signal frequency.

That is also a pretty high output impedance.. I would not be sending that too far. You really want it to be a low impedance output if you are sending it over to a PC or you will pick up all kinds of noise from the PC itself.

• Smartphone input impedances are around 800 ohm to 1.5kOhm. And the mic outputs some volts of bias voltage. Im wondering how would you interface it. Can you show as a diagram if possible? May 31 '17 at 12:04
• C1 is 10u actually May 31 '17 at 12:09
• @user16307 MIC? Smart-Phone? SIgh... You should really edit the question to be more specific and not make this a moving target question. May 31 '17 at 12:15
• Smart phone have TRRS jack one of them is mic input. So the 10mV pk-pk square wave signal will be coupled to that input. May 31 '17 at 12:18
• "If need be you can add a cap to the output to limit the frequency @ x10 your signal frequency." I want to try this but how do you size that cap for 1kHz freq? That cap one lead will be at the output and the other at GND right? May 31 '17 at 12:41

Almost certainly if you shorted the scope probe to the earth lead of the probe and touched them to the target system's 0 volts then you would get that noise. There is nothing especially noisy about the original signal and the circuit you use. You are just pushing the scope too far I reckon.

• My aim is I will send this 10mV square wave to an ADC of a soundcard or a smartphone. And the software will count the freq. I see this noise in scope but not in a PC software for a soundcard daq. So wasnt sure what is the freq. of this noise and would it affect the pulse freq. readings at the ADC side. May 31 '17 at 11:50
• Check your scope as i suggest. May 31 '17 at 11:52
• I shorted the scope probe to the earth lead of the probe and touched them to the target system's 0 volts, but I dont see that hf spikes in FFT in that case it is just flat noise distribution like white noise, May 31 '17 at 11:53
• Maybe i misunderstand you, could you show in a small figure the wiring for the test you suggest? I shorted scope lead to its earth May 31 '17 at 11:59
• It sounds like you did it right. Did the noise amplitude look the same as that seen on top of the square wave? May 31 '17 at 12:09