EEVblog #343 - Spectrum Analyser Tracking Generator Tutorial

[poptones]

we  both attended Wayne State University...

When were you in Detroit?

[JackOfVA]

we  both attended Wayne State University...

When were you in Detroit?

From birth through 1976 when I moved to the Washington DC area.

[JackOfVA]

A few more filter images and measurements.

First is Coilcraft's P7LP filter module. These are inexpensive 7th order low pass filters (under $5 each and free samples are available from Coilcraft) intended for PCB mounting. I made a small adapter board with BNC connectors and assembled a couple dozen P7LPs in various frequency ranges to use as quick and dirty filters for tests and the like. The main drawback to these filters is the ferrite core inductors used and the consequent intermodulation and harmonic distortion if strong signals pass through the filter. I'm sure my filter board could be improved as well, but it's adequate for my purposes. As the ripple in the stop band shows, this is an elliptic design filter.

Next is a 24 MHz cutoff low pass filter used by RCA for 100 watt level marine SSB transmitters. It's all  through hole parts. Looking at the response, I believe it's a Chebyshev design.

Then is an RCA 37 MHz cutoff low pass filter for 250 watt FM two-way transmitters operating in the 30-35 MHz range. No images of the inside, though, as it's soldered shut. The filter response looks to be elliptic.

Lastly is the home brew 15 MHz LP filter swept response overlaid with these three filters, 300 KHz - 1 GHz. Not unsurprisingly, the multiple stage filters show sharper transition from pass band to stop band -- that's why we use multiple section filters -- and all show spurious responses to some degree or other as the frequency increases to some multiple of the stop corner point.

Although Dave did not mention it, his 15 MHz filter and the one I built to duplicate it is a Butterworth.

[T4P]

Nice VNA you have there 

[lordnoxx]

Hi Guys, I have a short question about the Rigol DSA 815.
From its specs I can read that it has a frequency range from 9kHz to 1.5GHz. But at one or two points in the video Dave talks about a range of 0Hz to 1.5GHz. So whats the truth? Can it really go down to nearly 0Hz? And if so....why does Rigol specify a range starting with 9kHz?

I am actually not quite familiar with spectrum analyzers so maybe you guys can help me there. Thanks!

[EEVblog]

Hi Guys, I have a short question about the Rigol DSA 815.
From its specs I can read that it has a frequency range from 9kHz to 1.5GHz. But at one or two points in the video Dave talks about a range of 0Hz to 1.5GHz. So whats the truth? Can it really go down to nearly 0Hz?

No, it can't, 9KHz is the minimum.
"DC" is just an RF expression that means "low frequency", it's not actually DC.
If you want very low frequencies, what you want is a DSA (Dynamic Signal Analyser)

Dave.

[Dago]

I'd be really interested in a review of the signal hound which is a 4.4GHz spectrum analyzer you can attach to your computer. Base model is bit over 900$ and you can even buy a 4.4GHz tracking gen for around 500$. It seems unbelievably good for the price!

http://www.signalhound.com/

[olsenn]

I'd be really interested in a review of the signal hound which is a 4.4GHz spectrum analyzer you can attach to your computer. Base model is bit over 900$ and you can even buy a 4.4GHz tracking gen for around 500$. It seems unbelievably good for the price!

The problem with the signal hound is that is is extreamly slow. Some people on here mentioned how the DSA815 is slow... but compared to the SignalHound, it's the difference between a cheetah and a turtle! I don't think the noise floor/sensitivity is very good either. The real advantage to the SignalHound is bandwidth.

[kc3ase]

Hi Guys, I have a short question about the Rigol DSA 815.
From its specs I can read that it has a frequency range from 9kHz to 1.5GHz. But at one or two points in the video Dave talks about a range of 0Hz to 1.5GHz. So whats the truth? Can it really go down to nearly 0Hz?

No, it can't, 9KHz is the minimum.
"DC" is just an RF expression that means "low frequency", it's not actually DC.
If you want very low frequencies, what you want is a DSA (Dynamic Signal Analyser)


Dave.

Q: Why is the minimum 9kHz? What is the limitation?

[tree]

Q: Why is the minimum 9kHz? What is the limitation?

If you're referring to why it doesn't go down to DC, then:
There's a DC decoupling capacitor in series with the input. This prevents DC from being "seen". Only the AC is passed through.

If you're wondering why it's 9kHz and not some other non-DC frequency, then the simple answer is that was the bandwidth that was designed in.

[Christe4nM]

9kHz is the minimum because FCC has defined that RF energy is EM energy in the range 9kHz to 3000 GHz. (Source: Henry Ott, Electromagnetic Compatibility Engineering, page 7)

As a result the industry took that as the standard minimum frequency for spectrum analyzers.

[SNGLinks]

That's for RF spectrum analysers. There are AF analysers that go way below 9kHz.

[rf-loop]

If I get this kind of result with what ever spectrum analyzer (with direct coaxial cable from TG out to SA input) I'm highly alarmed and start deep investigation what is going on before I continue to use spectrum for any important measurements.

Something is badly. I'm surprised Dave did not stop show and before continue first look what hell is going on. This is NOT at all normal imho.

I have long time experience with many kind of professional grade spectrum analyzer in practice and my eyes tell with this Rigol image that least there need suspect some proble. Is it user mistake or spectrum analyzer itself problem can not know with looking just some video. But if it is my equipment I do some real tests for know more.



I did one example for just thinking what is going on there. (small box where signal trace continue)
(Image is from this Dave's video but perspective corrected some amount)

[SNGLinks]

It's 0.5 dB per division so it looks bad. It's within spec.

[rf-loop]

It's 0.5 dB per division so it looks bad. It's within spec.

Of course I know there is 0.5dB / div setting, so what, it do not matter at all if there is 0.1dB or 1dB or 10dB/div.  But these jumps and small blocks sharp edge level shifts do not look normal  (specially because shifted waveform parts match nicely to each others if you remove imagined shifts.  As I told, if it is my equipment I will start investigate what it is. (It is easy to test if know what is doing and have right equipments)
Also it is not question if it is inside specs or not and also this do not tell anything bout if it is inside specs or not. But, as told, this is not question.

It is hard to believe these "steps" are norrmal, even when spectrum is low-end entry level equipment as this is.

[rf-loop]

It is hard to believe these "steps" are norrmal, even when spectrum is low-end entry level equipment as this is.

These "steps" are the calibration data of the tracking generator.

Peter

Thank you.  This explain it. (lets think it explain it fully until someone show other)

I believe (and hope)  there is possible to run it also without cal correction for some user needs where these sharp level steps may disturb some special shape measurements because it do not interpolate between cal points. It is not nice if there is sharp step just in critical point if example doing some filter adjustments.

Least it is important that user know his equipment for avoid mistakes.