Pulling a mV signal out of the Noise

Thread Starter

SamR

Joined Mar 19, 2019
5,031
Background: Working with a common emitter amplifier exercise that injects a mV signal. My Siglent SDS 1102CML+ is swamped with noise! How do I eliminate the noise to pull out the signal??
Capture.GIF
Scope without probes attached and measured noise signal.
1.gif2.gif3.gif

Scope with probes attached and 60hZ noise signal.
4.gif5.gif6.gif

How would I elmininate the noise to capture the 10mV injected AC signal in and out of the amp?
 

Thread Starter

SamR

Joined Mar 19, 2019
5,031
Do you see that noise when you attach the pobes and ground to the circuit?
Yes! The cordset to the scope has GND. I tried attaching probe GND to the GND lug for adjusting the probe capacitance and to the External Input's BNC outer shield GND of the connector. No change.
 

Thread Starter

SamR

Joined Mar 19, 2019
5,031
This is the 10mV 10kHz signal taken from the point it goes into the base cap. Looking at it at the base across the 1uF cap 8.gifis the same.
 

ronsimpson

Joined Oct 7, 2019
2,985
How about a LC across Rc. This will kill the gain at all but the frequency you want to see.
Sorry about the bad inductor drawing. My paint agility is crude.
1571786700018.png
 

Thread Starter

SamR

Joined Mar 19, 2019
5,031
How about a LC across Rc.

Thx. At this point, I'm not looking for a workaround but to resolve the noise problem. I can turn up the AC signal amplitude but would like to get to the root of my problem first.
 

crutschow

Joined Mar 14, 2008
34,280
Yes! The cordset to the scope has GND. I tried attaching probe GND to the GND lug for adjusting the probe capacitance and to the External Input's BNC outer shield GND of the connector. No change.
Is that with it connected to the operating circuit?
A floating probe will pick up a lot of AC main's noise.
 

Thread Starter

SamR

Joined Mar 19, 2019
5,031
The last screen-shot is connected to the circuit. The is some non-60Hz hash on it that looks digital to me shown at 17.26kHz or 98.9060kHz? That interferes with the AUTO function of the scope to capture and display the input signal. I have tried to manually adjust the scope to display the AC input signal but am not able to capture it. I have tried filtering and signal averaging w/o success. I have increased the signal input amplitude up to 1Vpp and dropped the frequency to 1kHz without any real success. I have tried 3 different NPN transistors. I am also not getting any amplifiation on the collector output signal. Very frustrating. The baseline signal without any probes connected has more than 10mV of noise on it. When I change the channel input to GND it is a perfect flat line. On the DC or AC inputs there is baseline noise of 60mVpp with a mean of 0.0mV? So how can I get a good 10mVpp input signal or is it even possible?
 

ronsimpson

Joined Oct 7, 2019
2,985
Looking at the noise I think there is 100khz switching power supply noise.
There is always 60hz noise. (50hz in some places)
Depending on where you live you might see 15750hz.
There is much gain in that one transistor. With any type of antenna you should be able to pick up AM radio + 60hz.
Keep the loops small, and the grounds large. + shielding.
 

tautech

Joined Oct 8, 2019
383
Sam, what are your probe and input settings ? 10x ?
Have you tried 1x or will this load down your circuit with probe additional capacitance ?

Been a while since I used a CML so I can't remember if the have an Eres mode that will clean up the signal further and better than averaging.
 

Thread Starter

SamR

Joined Mar 19, 2019
5,031
I am using 1X. No idea what Eres Mode is? I worked with it quite a while last night. After applying some filtering and signal averaging (16) I was able to get a very unstable, but fairly clean, signal. It is not possible to obtain a dual input/output trace. I still need to work with it some more today and do some further circuit analysis as the output gain is not much at all. Will post update after some further work.
 

Ya’akov

Joined Jan 27, 2019
9,069
I am using 1X. No idea what Eres Mode is? I worked with it quite a while last night. After applying some filtering and signal averaging (16) I was able to get a very unstable, but fairly clean, signal. It is not possible to obtain a dual input/output trace. I still need to work with it some more today and do some further circuit analysis as the output gain is not much at all. Will post update after some further work.
Sam, in the Acquire menu, first option:

1571834836086.png
 

Thread Starter

SamR

Joined Mar 19, 2019
5,031
In my CML aquire menu there are only 3 options, Sampling, Peak Detect, and Average. For noisy signals, I use Average w/ an average of 16 or 32 counts typically. The choices are 4/16/32/64/128/256 so I am on the low end of averaging the input signal for processing speed's sake. There is also Mode, either Real Time or Equal Time. Not sure what the difference is but I use Real Time.

EDIT: I also had to forgo the AUTO detect and manually adjust the X-Y Axis parameters. The AUTO mode tunes to the higher frequency noise and eliminates any of the input filterings to do so. So apply filters, then manually adjust the signal to the optimum clean display. Even that is not real stable but at least a close approximation of what I would expect it to be from the math analysis.
 
Last edited:

Ya’akov

Joined Jan 27, 2019
9,069
In my CML aquire menu there are only 3 options, Sampling, Peak Detect, and Average. For noisy signals, I use Average w/ an average of 16 or 32 counts typically. The choices are 4/16/32/64/128/256 so I am on the low end of averaging the input signal for processing speed's sake. There is also Mode, either Real Time or Equal Time. Not sure what the difference is but I use Real Time.
Ah, plan for an upgrade? :)
 

SLK001

Joined Nov 29, 2011
1,549
I believe that you are picking up stray ambient noise. Put your entire circuit over a ground plane and most of it should go away. A piece of copper clad pcb should do the trick - just be sure to ground the ground!
 

Thread Starter

SamR

Joined Mar 19, 2019
5,031
I believe that you are picking up stray ambient noise. Put your entire circuit over a ground plane and most of it should go away. A piece of copper clad pcb should do the trick - just be sure to ground the ground!
That is what I thought. I tried shutting down the computer, fluorescents, and ceiling fan but that still leaves WiFi and a host of other noisemakers including the 3 phase Power feeder less than 100' from the house. However, the house has a metal roof and mylar foil in the siding (all of which I grounded) so I am living in a bit of a faraday cage. The wife goes outside to use her cell or stands in front of a window. I do still detect some of the neighbor's WiFis. I am in the process of relocating my desk/bench and have already removed all of my radios and antenna wiring including my dedicated ground so will have to wait until I finish the relocation project to go to a ground plane device but will keep it in mind.
 

SLK001

Joined Nov 29, 2011
1,549
What? No piece of copper clad lying around? You could also use a piece of aluminum. Try a piece of grounded aluminum foil under your circuit.
 

Thread Starter

SamR

Joined Mar 19, 2019
5,031
No dedicated ground other than the 120V panel ground ATM. I could cobble something up a try it. Going out for lunch w/ the "boys" in a bit so will give it a shot later and see. Thx for the suggestion!
 

OBW0549

Joined Mar 2, 2015
3,566
I tried shutting down the computer, fluorescents, and ceiling fan but that still leaves WiFi and a host of other noisemakers including the 3 phase Power feeder less than 100' from the house. However, the house has a metal roof and mylar foil in the siding (all of which I grounded) so I am living in a bit of a faraday cage.
Yes, but even with all that stuff turned off you've still got interference sources inside that Faraday cage. Even if everything is turned off and unplugged, the power wiring in your house is still emitting plenty of 60 Hz garbage, plus harmonics. You can see this if you attach a long piece of hookup wire to your scope probe and hang it in midair; in my case, I read over 5 volts peak-to-peak from a 6 foot wire.

I agree with @SLK001 and I'd also go two steps further: I'd completely enclose the amplifier circuit in a metallic enclosure connected to circuit ground, and connect all input signals, such as from a signal generator, via shielded cable.

I've found that's what I have to do any time I'm working with millivolt-level signals, otherwise I get tons of interference that looks a lot like some of the waveforms you showed in your top post.

Also, what are you powering this circuit with? For low-level analog work, I've found that any form of switching supply is a no-no; with a switcher, high frequency spikes such as the ones you show in post #4, are almost impossible to get rid of.
 
Top