Spectrum Analyzer on the Cheap

Provided you have an NTSC-compatible TV you can build yourself a really inexpensive spectrum analyzer. From there you just need one trivial piece of hardware to complete this build. [Bruce Land] has come up with a spectrum analyzer that shouldn’t cost any more than $5, if that’s what’s been keeping you from adding this tool to your workbench!

The spectrum analyzer is based on a PIC32 microcontroller which was previously proven in his Oscilloscope project. [Bruce] has managed to squeeze quite a bit out of this robust chip; the spectrum analyzer has 450 kHz bandwidth and runs a 256 Hz TV display and can output over 30 updates per second. The microcontroller runs the Fast Fourier Transform (FFT) to do calculations, with great results.

[Bruce] notes that the project was based on TV framework from another project, and that the FFT was added on top of that. Be sure to check out the source code on the project site if you’ve been on the hunt for an inexpensive spectrum analyzer, and if you need something with more processing power but only slightly more money, check out the FFT that runs on the Raspberry Pi’s GPU.

27 thoughts on “Spectrum Analyzer on the Cheap

    1. yes, it has been corrected now. Things like this were seriously looked down upon during my education.
      Just like specifying memory sizes in milli bits, while they mean Megabytes.
      It was also punishable by death to call physical quantities by their unit names, like Voltage, Amperage, Wattage, etc..

      1. ok, so apparently in English, the physical quantity is also called Voltage, this is not the case in dutch, french or german for instance. I guess you could use ‘electrical potential’, but Voltage seems to be the norm in English. However it is still wrong imo to call Current Amperage, or to call Power Wattage, etc. Someone calling them by their unit names shows that they had no formal education in the field. Any EE doing this deserves to be mocked.

          1. Well, VA is the correct wat of rating a transformer. Sinse the manufacturer doesn’t kwon what the cos phi of your load will be. So specifying the maximum apperent power it van deliver is logical. At least that:s what I’ve been thought and find logical.

          2. nobody is arguing that the units used are not correct, but that you don’t use the unit name to specify the quantity.
            kVA is a unit of Work, so call it that, “Work (W)”, not kVA.
            Ampere is the unit of electrical current, so it is called “(electrical) current”, not Amperage.

          3. *** correction ***

            kVA is a unit of Work, so call it that, “Work (W)”, not kVA.

            should of course say

            kWh is a unit of Work, so call it that, “Work (W)”, not kWh

            kVA is of course the unit of “Apparent Power”

          4. What I’m saying is that, “Unit Analysis”, is a very important tool in both science and engineering. There are constants, and there are variables. When you say,

            “Someone calling them by their unit names shows that they had no formal education in the field. Any EE doing this deserves to be mocked.” -[bthy],

            is absurd. I can tell you have never worked in the industry by that statement. I want to say that you should be mocked for demeaning the electricians of the world that make our designs real. I won’t, because you are just a troll that replies to their replies once you realize you are mistaken.

        1. Some people write for readers and some write for critics. The critics don’t need to read it. I’m often critical – constructively I hope. The nightly news is chock full of “a car drove off the road and hit a tree”, “The uptown trolley is celebrating an anniversary today”, “warm temperatures” etc. Complete gibberish. Certainly “amperage” is much less ambiguous than “current”, which can mean right now, or up to date or swift flowing water or air drafts or a nice jelly preserve depending on context. We should strictly use eV and dq/dt perhaps. Maybe dq/dt with a little c at the bottom to show Coulombs, like showing the reference value in dBV. Oops. dBeV

          1. (Hit the limit on comment depth). The potential difference, or voltage, is energy per coulomb. The electron volt or eV is energy per electron (or single unit of charge regardless of mass). You can use eV if you use the charge of the electron instead of the coulomb.

      2. So they were looked down upon wherever you’re from. Me too. So fucking what? This is a blog. It isn’t graded, there’s no instructor. Only asshats like yourself that see only see errors and apparently need some perspective. You knew what was meant. I knew what was meant. So did EVERYONE ELSE, and apparently that isn’t good enough for you.

    1. Forgive the newbie question. My only previous exposure to a spectrum analyzer is that of providing visualizations for music :)

      Lately I’ve been reading about the topic of emissions testing necessary for selling consumer electronics.

      Is a device like this suitable for performing some basic emissions testing when working with microcontroller projects clocked at 16Mhz or less?

      I know this would never meet the standards of a regulatory agency, but I’m wondering if this could be a practical way to identify red flags. While I know that many startups essentially ignore the formal certifications (and refuse to comment on that decision for legal reasons), I think it would be beneficial to have some peace of mind and/or relative level of certainty that a project is not actually causing excessive levels of interference.

      I’ve searched on these terms a little bit but haven’t found anything that speaks to my level of understanding.

      1. I did find this talk from DEFCON and it appears that it would be possible to do this with SDR, that these particular dongles might be limited when it comes to lower frequencies.

  1. I can’t tell what frequency the spectrum is. I want to see a line not dots. 450 kHz bandwidth seems rather wide.
    I can make a spectrum display out of a radio, cellphone, or TV tuner with it’s native bandwidth or with different filters. Using just a 555 for the timebase and scope for display rounds out the whole thing. This is able to show small traces of interference hiding between many strong signals.

  2. What about hack the TV’s tuner into a spectrum analyzer front end? If it’s a varactor based unit, it could just be a matter of tying that into the horizontal deflection, then feeding the detector output into the vertical deflection driver.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.