# Using a spectrum analyzer to find LC resonance?

1. Apr 8, 2013

### hobbs125

Hi everyone,

I am using a signal generator in series with a RLC circuit, generating an AC signal at 1kHz.
I have my probes connected across the resistor

When I change the timescale the Fres bandwidth changes on the spectrum analyzer.
At 1mS the resonant bandwidth is from 25kHz-50kHz
At 2mS the resonant bandwidth is 13kHz-24kHz
At 5mS the resonant bandwidth is 4.9-9kHz

Can anyone here explain why the resonant bandwidth is changing when I change the timescale on the spectrum analyzer? And how I can find the actual resonant frequency of the circuit?

Am I doing something wrong here?

2. Apr 8, 2013

### Integral

Staff Emeritus
RTFM?

You need to give us some information about the spectrum analyzer you are using. Brand and model number at a minimum. Then maybe someone here could read the manual for you.

3. Apr 9, 2013

### davenn

you sure its a spectrum analyser ?? not an oscilloscope ?
or is it a combo and you are using the O'scope function?

spectrum analysers have freq on the X axis ( Not Time) and power on the Y axis

As Intergral said .... what make and model is it ?

Dave

4. Apr 9, 2013

### hobbs125

It's a propscope usb oscilloscope with a spectrum analyzer.

Last edited: Apr 9, 2013
5. Apr 9, 2013

### Staff: Mentor

What values of R, L, and C are you using?

6. Apr 9, 2013

### vk6kro

If you connect the R, L and C in series, with the L connected to ground at one end, resonance is achieved when maximum voltage occurs across the inductor.

So, you vary the frequency to get maximum output on the spectrum analyser connected across the inductor.. Then read the frequency of the signal generator.

If you can't find resonance, connect a 50 ohm resistor (or similar) across the signal generator output.

7. Apr 9, 2013

### hobbs125

My main question:
Why is the resonant bandwidth changing when I change the timescale?

Is there a way to calculate fres based on the what the timescale is set at?

8. Apr 9, 2013

### vk6kro

It sounds like you are using the device in oscilloscope mode.
A spectrum analyser has frequency as the horizontal scale and amplitude as the vertical scale.

Try feeding the signal generator output directly into the spectrum analyser.
You should see a vertical line moving horizontally depending on the frequency.

If you can't see this, check the user manual to find out how to use the device.

9. Apr 9, 2013

### sophiecentaur

Spectrum analysers need to be read with care.
When you change the timescale on a spectrum analyser ( the scan rate would be a better term) the bandwidth of the receive filter is often changed automatically (the detector needs to be able to follow changes fast enough as the scan progresses). If the filter you are measuring has a narrow bandwidth, it is quite possible that the width of the peak you are seeing is a product of this plus the receiver bandwidth. If you use a very slow scan rate, the receiver filter will be at its narrowest and the narrow peak you can see is more likely to be representative of the filter under test.
If you have enough controls on your analyser, you should be able to adjust the receiver bandwidth manually (a warning flag "uncal" may show up) and see the same effect with a constant scan speed. It's worth a try, in any case.

Having read this, I realise my comments are aimed primarily at an analogue analyser (my main experience is with UHF and above) but there may well be an equivalent problem with sample rates and the resulting FFT .

10. Apr 9, 2013

### hobbs125

Ok,

I connected the spectrum analyzer directly to my signal gen as vk6kro suggested. Now I can clearly see that the spectrum analyzer is working. Thank you vk6kro and others for you comments and help thus far.

This also helped me realize all the signals I have been seeing are just noise:( which is extremely frustrating.

In the attached picture you can see the 1kHz peak from the signal generator, and then a ton of noise from 25kHz to nearly 50kHz?

So, can anyone tell me what might be causing all this noise and how I might get rid of it.
Also, this noise is what is changing frequency range when I chane the timescale....I'm wonding if my oscope is just screwed up?

#### Attached Files:

• ###### scope1.png
File size:
42.3 KB
Views:
214
11. Apr 9, 2013

### davenn

and what is worse is that all that noise is substantially stronger in level that your 1kHz signal
which should be the strongest signal there

OK you need to determine the source of the noise, either where its being generated or getting into the circuit from an external source.

disconnect the sig gen and LC circuit from the scope/analyser ... is the noise still there ?
If it is, it could be artifacts generated by the scope hardware or software OR it could be noise generated by the computer getting into the scope hardware ( computers are VERY noisey electronics)
if there is no noise spikes visible, then proceed ....

Connect the LC circuit only, does the noise reappear ?
if so, maybe the wiring is picking up radiated noise from something around your workshop
if there is no noise spikes visible, then proceed ....

finally test the signal generator output, connect the sig gen directly to the scope.
are there noise spikes or a clean 1kHz spike ?

go through that for a start and let us know the results :)

Dave

12. Apr 9, 2013

### hobbs125

Dave,

When I turn off the signal generator the noise is still there. ...also, when I do not have the probes connected to anything the noise is still there.

I think your right about it likely being the computer. I will try to move the computer away as far as the usb cord will allow (about 2 ft) and see if the noise is still there. If it is I can work around it, but it is very frustrating.

Is there anything I could do to remove the noise if it's from the computer?

13. Apr 9, 2013

### davenn

Hopefully this is the unit you have .... http://elmicro.com/files/parallax/32220gettingstartedv10.pdf

Possibly ...
You could try a clamp-on ferrite choke on the USB lead right where it enters the computer
that may emphasis on the word MAY :) cut the noise level down.
depending on the length of that USB lead you may be able to wrap a couple of turns through the ferrite choke.

clamp-on ferrite choke ....

cheers
Dave

#### Attached Files:

• ###### choke.jpg
File size:
15.7 KB
Views:
577
14. Apr 9, 2013

### vk6kro

It might be computer noise coming down the USB cable or pickup from a noisy broadband modem, but this should be showing on the sine wave shown as part of the screen grab.

This looks quite clean.

So, it may be a faulty device, perhaps becoming unstable.

Could you take it back to where you bought it and get the supplier to demonstrate that it is not faulty?

If this isn't possible, it is still a useful device and you could just live with the noise problem. Real Spectrum Analysers cost hundreds or thousands of dollars, so you couldn't expect the same results with a USB device.

15. Apr 9, 2013

### hobbs125

OK, Thanks everyone for the help and support.

I will try the ferrite choke.

BTW, my wireless router is only 2 feet away from my setup so I'll also try moving further away from it.

If those things don't do the trick I'm not to worried. I'm still pretty happy with the oscope.

16. Apr 10, 2013

### sophiecentaur

If I understand correctly, the 1kHz signal is on the top of your graph and the result of the FFT is on the bottom. This is very suspicious because the FFT is just some maths, done on the original time-domain signal. If that signal is clean (and it appears to be), then a simple FFT will not generate anything spurious like the other picture shows. If 'noise' is visible on the frequency plot then it should also be very visible on the time plot.

But I have another problem with what you say. You seem to imply that putting a 1kHz signal into your filter will show its frequency characteristic. This is not true. If there is only energy at 1kHz then there will be no information about the filter response elsewhere. You need, either a swept frequency (tracking) input or an impulse if you want to see the response of a filter.

I agree with vk6kro and suggest that you read the manual to make sure you are doing what you think you are doing with the instrument.

17. Apr 10, 2013

### AlephZero

FWIW, two of the USB devices I use came with their own cables, which have something that looks similar to a choke built into the cable. (And it's at the device end, not at the computer end - the connectors at each end of the cable are different so you can't reverse it).

Maybe your device is supposed to use a similar cable, but it has got "lost" among the rest of the lab equipment?

18. Apr 10, 2013

### jim hardy

The truth about computer produced electrical noise is leaking out !
:tongue:

19. Apr 12, 2013

### sophiecentaur

They are a real nuisance for lots of receiving equipment. You need an antenna up in the attic or on the roof; a hand held receiver gets all sorts of old rubbish from the computers.

But is still don't see why the spectrum of that nice looking sine wave has so much apparent HF shash on it, according to the 'analyser mode'. That scope trace should, by rights, be a mush unless there's some extensive LP filtering of the signal when in 'scope mode'. We need to know more details about what those display pictures actually represent.

20. Apr 12, 2013

### eq1

21. Apr 12, 2013

### sophiecentaur

22. Apr 13, 2013

### vk6kro

That's right.

The Spectrum Analyser mode and the Oscilloscope operate at different times.

Somehow, when it operates in Spectrum Analyser mode, it generates noise.

This may be a faulty device component or just a driver bug.

It can't be externally generated noise or it would show on the sine wave in the oscilloscope window.

You could try to get it repaired or replaced or try to get updated drivers from the manufacturer.

23. Apr 13, 2013

### sophiecentaur

The only reason I can think of for the two different apparent behaviours is that the FFT is being performed at the wrong time - i.e. instead of using the string of samples that are shown on the scope mode trace (which should produce a single spike), it must be using an unsynched string of samples. That would have to be more or less software. Looks like it's a matter of approaching the manufacturer. There could be an upgrade just waiting to be installed.
I wonder if it is possible to vary the input frequency by a small amount and produce a large effect on the arrangement of those HF spurii. Aliasing can be a real problem in digital systems and this could be a form of temporal aliasing.

24. Apr 13, 2013