DIY Spectrum Analyzer 0 to 1750MHz (1999) | Hacker News Reader

DIY Spectrum Analyzer 0 to 1750MHz (1999)

lea.hamradio.si

101 pointsdschuetz29 comments

For what it's worth, this is a 2007 web translation of a 1999 article on hand-building a spectrum analyzer. To me it looks very analog; I can't imagine anybody building this today except as a sort of electronics historical reenactment project. If you're wondering how it all fits together, I think this is the better place to start:

http://lea.hamradio.si/~s53mv/spectana/lcd.html

But you'll find the whole series of articles in this directory: http://lea.hamradio.si/~s53mv/spectana

If you crack open a real spectrum analyzer (not some hacked together SDR toy) you’ll still find plenty of distributed and analog circuitry. That’s what it takes to get spur-free performance.

Speaking of hacked together RF toys, I picked up an IM-Me toy a few years ago just so I could play around with Michael Ossmann's "$16 spectrum analyzer". I was able to see (poorly) the transmissions from my wireless weather station with it.

http://ossmann.blogspot.com/2010/03/16-pocket-spectrum-analy...

There is also a five part series on Youtube talking about what else you can do with the IM-Me. It is pretty amazing and well worth watching if you have some time.

https://www.youtube.com/watch?v=WGU30mF_dgM

SDRs are total spur-fests, yes, but at 0% the cost of a real SA (rounded) the economics sometimes work out :-)

That said, at the high end of the SDR range, the new AD937x chips ($325 in qty 1k!) make impressive claims wrt automagic spur suppression. They still don't beat high-IF or YIG preselected SAs, but now they claim to be in the ballpark under non-overload conditions. Has anyone here dug deep enough to comment?

The pre-selection is a big deal. The AD936x needs good front end filtering to reject n*LO. For instance 5x146 MHz = 730 MHz, which translates that 730 MHz nation-wide LTE signal smack-dab into the 2m amateur band. Unless they have a new front-end architecture in the AD937x, all the I/Q calibration won't help them.

Of course as you said, for $325 it's not a consumer chip, so some $ will be spent on external passives.

Ah, right, I/Q calibration can't suppress anything out-of-band. Makes sense.

I really like the simplicity of this design. So, any pointers to how one would build this today?

The biggest difference between building this and a more modern style analyzer is that you would cut out the log detector and replace it with a digitizer.

Thank you, that is the kind of answer that I was looking for.

Hmm, well, it depends on where your knowledge base is :) Studying electrical engineering wouldn't hurt.

(not complete, but it's a start!) Basics: Math: Complex Numbers, Integration Physics: Wave Mechanics, Oscillators, Energy, Impedance Electrical: Currents, Voltages, Capacitors, Coils Electronics: Semiconductors, Resistors

Semi-Advanced: Electrical: Compensation, Impedance Conversion, Complex Values, Frequencies Electronics: Transistors, Amplification

Advanced: Electrical: Magnetic/Electric Fields, EM Fields, Flux, Conductors/Waveguides Electronics: Filters, Operation Amplifiers, Oscillator Circuits

and further.

When you are able to build and test your own filters and oscillators, then you might pull off building a simple/small SA yourself.

!! But remember: always work inside a shielded environment! Active analog HF circuitry interferes with everything else around you !!

I did not mean 'where do I go to pick up the basics', I think I have those covered. But since there is a critique of this design and I actually rather like it I am wondering what specifically you would do different 'today' because of advances in technology. Obviously the whole thing would be mostly using SMD rather than hole through components and the display would be integrated rather than using an external scope but the design itself is what interests me.

> I can't imagine anybody building this today except as a sort of electronics historical reenactment project.

FWIW SAs using this topology are current products with most RF test/meas manufacturers.

Sorry, maybe I was incorrectly dismissive. I don't mean to imply that one wouldn't do something similar in many ways. I was thinking mainly of the output, which was to a 128x64 LCD panel with knob-and-switch controls. I can see why in 1999 that was a reasonable choice. But 20 years later I expect the much more effective choice is to send the data to a phone or tablet for display.

> I can't imagine anybody building this today except as a sort of electronics historical reenactment project.

Hmm, or anyone who can’t justify spending $2k on a single piece of gear for their hobby :)

I think the point they were making is that if you were going to build something like this you would use a digitizer. That would probably be a cheaper build at this point.

In principle it's about how to build a custom HF SA. I'm sure you'll find better components for less money, or even use SDR, but good HF/microwave spectrum analysis until today relies on analog circuitry. It's amazing that sometimes you can't even find/buy a correct reactance, so you have to design something else yourself to make your VCOs work as intended.

The principles never change, but the implementation does. How to make good reliable HF filters with least effort? There it is.

I think the important part many people fail to realize with analog projects like this is that they are constructed, diagnosed and repaired with common off the shelf tools and components.

Sure it can be done with a cheap SDR but can you fix the SDR in 15 years when a chip dies? The modern all-in-one digital stuff for the most part is throw away.

These old analog projects can be pulled apart, tested and repaired with ease. Try that with a 5x5 mm bga package running code you probably won't be able to find easily or at all. Then hope the chips are still available and not long obsolete. Then figure out how to program the chip.

Bob Pease said it best: "My favorite programming language is ... solder"

I’m building a slightly crappier and less ambitious 0-200MHz one at the moment for HF use. Basically a step attenuator, LPF, a pair of NE602 for first and second LO+mix, an old Toko helical filter for first IF and hand built crystal filter (12MHz) for second IF resolution filter and then an AD8307. First LO is a VCO with varactor control driven from an opamp ramp generator. Doesn’t work yet as the Toko filter is broken and I can’t find a suitable replacement and can’t be bothered to make a helical filter.

The things look like rocket science but they are basically superhet receivers with a log converter instead of speakers and a much larger tuning range.

The sheet metal work in projects like this interests me. In the 1930s the new technology of the time was shortwave radio, so hobbyists built their own. Short Wave Craft was a magazine that supported this hobby- it's very much like Popular Electronics or Byte magazine. Always there were sheet metal plans included in the project. Here is an issue of it (pages 51 and 35 have typical chassis plans):

https://www.americanradiohistory.com/Archive-Short-Wave-Tele...

So did hobbyists have access to sheet metal brakes? Not really- they either use a vice and hammer, or bought pre-made chasses from a place called "Blan the Radio Man"- a store in NYC's old Radio Row:

https://en.wikipedia.org/wiki/Radio_Row

https://i.pinimg.com/originals/d5/4f/2c/d54f2c957af5cec29831...

If you're working with thin steel sheet, or some aluminums, building your own sheet break is pretty easy to do, even without access to a machine shop.

https://duckduckgo.com/?q=diy+sheet+break&t=ffab&iax=videos&...

It's a much cleaner finish than the hammer+vice combo.

Also you can get cheap brakes from Grizzly or Harbor Freight.

I recognised this guy from the format of his webpages (very fast loading, no unnecessary stuff --- a look at the source suggests it's handwritten HTML); he has also built his own GPS receiver:

http://lea.hamradio.si/~s53mv/navsats/theory.html

Are there tools to easily create xkcd style diagrams similar to these?

I'm fairly sure that's technical pen on paper which has been scanned, run through a quantize filter to clean it up, and shrunk.

For sure, although not completely impossible to replicate in software.

In this case almost certainly paper, technical pen and somewhat steady hand.

Drawing like this becomes natural on engineering paper or any generic graph paper with some practice.

Beats the heck out of any software for efficiency in communicating arbitrary ideas in particular because it is not an opinionated interface (personal opinions aside) and you can do what you imagine without limitation. When you have to scale to dozens of drawings or collaborators then the benefits of computerization show themselves, but not before.

With pen and paper the primary thing in my mind is the idea I'm trying to convey, with a computer it is how to convey it with the computer (until I've reached some level of expertise with the computer in a particular space).

Balsamiq Mockups can make diagrams that look similar.

https://xkcd.com/730/

Reply on news.ycombinator.com