How to filter the noise from oscilloscope's probes

[oliverlin09]

Dear all,

My circuit is a receiver. (Please see the attached file)
The input connects to a sensor, and output connects to a oscilloscope.
However, I found the oscilloscope's probe has its own noise.
The noise is about 90mV (Pk-Pk)(The biggest one) and its frequency is between 3MHz to 500MHz.

The problem is my signal from the input is about 100mV@10MHz.
For some reason, the noise from the oscilloscope's probe will be amplified as well.
My circuit's filter should be able to filter the signals over 20MHz.

I am wondering if there are any ways to remove the noise.

 

[karesz]

Hi,
An oscilloscope probe is passive and hase not so much noise as you asumed it.
If you have active probe_may have some more noise, but not so an irreale amplitude pls., for usability it is usual max. i.e. 0.1-0.3 div on 5mV position...

How did you measured pls these 90mVpp noise from your probe?
It sems to be irrealistic_ you have <60 Ohm input impedance...
Is it possible, that you do solder a coaxial cable to the input directly?
I assume that your probe hase had a longer GND clips_is it true pls?
Or is your sensor the source of the high noise...
You can check it if you put an 50 or 62 ohm resistor (or a simple short) on the same probe and your noise must be change to be very small peak-peak value, maybe less as 2-10mVpp, but you must it apply with shortest leads & GND loop to the probe clip.
K.

 

[frankliner]

If your signal is periodic and you have a stable trigger use averaging on the oscilloscope. If it is not periodic then a moving average filter may help.

 

[oliverlin09]

Hiya,

How did you measured pls these 90mVpp noise from your probe?

I used oscilloscope's probe, and it's x20.

Is it possible, that you do solder a coaxial cable to the input directly?

I did use a proper coaxial cable, but I think my circuit is on a breadboard.

I assume that your probe hase had a longer GND clips_is it true pls?

The probe's GND is short, but the input connector's signal and GND I soldered by wires and put the connector on my breadboard. I think perhaps this is the problem.

I used your suggestion and put a small value resistor on input or output. It did reduce some noise.[/quote]

 

[FvM]

From OP data, I calculated an amplifier output related noise of about 30 mVpp. Several 10 mVpp (probe input related)
are surely contributed by the oscilloscope input channel, if it's using a x20 probe and has 500 MHz bandwidth. An
assembly technique unsuitable for high frequency circuits (breadboard) possibly adds RF interferences and gain peaking.
All in all, it's normal operation.

Switching the oscilloscope to reduced bandwidth would be the first step to reduce the noise, suitable coupling of the
oscillospe to amplifier output the next. If done correctly, only the OP noise should remain.

If you want to determine the oscilloscope and probe's own noise, short the probe input without connceting to the OP output.

 

[karesz]

Hi,
But if we check the datasheet: I see with these "Au" a possible bandwidth of only ca. 10MHz, you dont need a 500MHz scope to check it, or you can problemles use hes "20MHz bandwidth limiter."...
My Gefühl sagt, dass 30mVpp sei doch einiges zuviel-oder?
I must tell, I have lot of RF breadboards used in my life, you can do it (i.e. with mini-mounts), but you must work "clean": but with RF-technique you need overall a clean style of (RF)working...
K.

 

[FvM]

The amplifier noise is estimated from AD8042 input en of 16 nV/sqrt(Hz), noise bandwidth of about 10 MHz, gain of 120,
factor 5 Vpp/Vrms for noise. Just to know the order of magnitude.

 

[oliverlin09]

Hi FvM,

16 nV/sqrt(Hz), noise bandwidth of about 10 MHz, gain of 120, factor 5 Vpp/Vrms for noise.

Sorry, this part is difficult for me. Please cold you explain it??

Oliver

 

[karesz]

Hi,
If your bandwidth is 10`000`000 Hz calculate a square route from it, in relation to your input is en=equivalent noise signal(at the inp.)= squarert of 10MHz multiplyed with 16nanoV(unic Hz en at the input) & multiplayed with gain factor of 120= possible U noise at your output, measured with 10MHz system bandwith...
K.

 

[FvM]

I should mention, that the gain calculation is only correct for a low impedance source, you didn't tell about this point.
I didn't want to focus on the exact noise calculation but clarify, that amplifier noise is a parameter that shouldn't be
neglected in wideband circuits.

 

[karesz]

Hi oliverlin09,
For better methodes in RF- testing I sourched for my favorit scripts for you...
Good lecture!
Check pls i.e. for the second half of Appl.Note70...
K.