A 74xx-Defined Radio (2021) | Hacker News Reader

A 74xx-Defined Radio (2021)

acidbourbon.wordpress.com

181 points_Microft30 comments

I knew there was going to be a 74HC4046 in there before clicking the link. It is a venerable old workhorse.

When I was studying EE, 40 years ago, one of our professors always wore a lab coat with a pocket full of 4046s. He used to roam the labs and offer one to students struggling with designs.

It is such a versatile chip, with a VCO and three different phase comparators. If you have to do anything related to modulation and/or demodulation of signals below a few 10s of MHz, it may just solve your problem.

Refreshing my memory on the '4046, I came across this thread, which may be of interest to others curious about this part:

https://eevblog.com/forum/projects/basic-74hc4046-pll-questi...

W2AEW's videos (one of which is embedded in the link above) are all fantastic for anyone wanting to learn about radio and analog electronics.

I built a frequency multiplier with one, a few weeks ago. Among all of the chips available, the 4046 had the largest quantity of easily implemented design info. Just for the sheer audacity of it, I went looking for the oldest design that I could find, to reference in my documentation.

I liked that too but my first thought was "I'd replace that 4046 with a digital oscillator for better stability" and lo and behold the author did just that as an option at the end.

Of course the price for that stability is those lovely classic AM tuning whistle sounds you get with a pot connected to a 4046...

When I had my "I want to build my own SDR" phase ;) ca. 15 years ago, I also used analog switches as a switching mixer - with a PLL running at 4x the desired frequency so I could utilize dividers to get nice 0°, 90°, 180°, 270° LO signals for IQ mixing. Brings back memories...

In the 70's the company I worked for made a receiver for fixed frequencies within 10KHz-13KHz. The architecture roughly was discrete time analog, and an analog switch was available that took a digital selection and routed the indicated analog input to the analog output. That block was used to control the sampling for a filter by cycling through the selections. The rest of the design was over my head in those days, since I was the software guy and the software wasn't controlling that.

This is called a Tayloe Quadrature Product Detector [1].

1: http://www.norcalqrp.org/files/Tayloe_mixer_x3a.pdf

I've always heard this technique referred to as a "Tayloe converter" (or "Tayloe mixer"), and it is commonly seen in SDR radios.

I believe this is the original article from the 90's which popularized the idea: https://norcalqrp.org/files/Tayloe_mixer_x3a.pdf [PDF]

The formal name used nowadays (one of them, anyhow) is Quadrature Detector. It's a scheme so elegant that grasping the simple mechanics of it demystifies some of the core concepts of Signal Processing.

The QEX series "A Software-Defined Radio for the Masses"[1] is usually offered as the work that popularized this approach. The Motorola patent[2] that cites Dan Tayloe as inventor expired in 2018 [3].

[1] https://sites.google.com/site/thesdrinstitute/A-Software-Def... [2] https://patents.google.com/patent/US6230000 [3] https://www.eevblog.com/forum/rf-microwave/patent-expiration...

It's quite the rabbit hole and if you dive in you'll end up with a HAM license and a workshop full of instruments and soldering tools.

This is basically what the SoftRock SDR kit is.

The build instructions' schematic is basically a color-coded block diagram of what all is needed for an SDR. A super fun, informative build.

http://www.wb5rvz.org/softrock_lite_ii/index/projectId=8

How do you learn something like this? I'm ok at digital electronics, but doing all this weird radio analog stuff is something I don't even know where to start to learn about

Of course there's EE university courses, but generally, experimentation. You can get really far in electronics with learning by doing.

If you don't know where to start, ham radio is one traditional gateway.

Whether or not you get the license is secondary (I think most interested people should get it; especially in the US where the lowest license class is ridiculously easy - if you know basic electronics already and cram today with https://hamstudy.org/, you could probably have a Technician class license by end of tomorrow - and then you can legally build your own transmitters or buy some cheap radios and transmit with those. Transmitting is more fun than only receiving.).

The real value is all the structure around it (e.g. 1. there are books that start from zero electronics knowledge and end at several types of complete transmitters/receivers. The ARRL publishes some; 2. the license is a tangible learning milestone to work towards; 3. all basic transmitters/receivers have been done thousands of times so you can find something online and copy it to learn from it. The ham community also has many DIY kits you can just buy and build and experiment with.)

Eventually you'll find that you have enough background to use more "professional" resources (e.g. data sheets, real academic textbooks / professional books, etc.) (at least if you have some math background from another field, e.g. CS), build more complicated things etc.

Note that getting a radio license requires publishing your legal name and a mailing (often home) address publicly in a government database under your callsign, and broadcasting your callsign with every transmission, effectively doxxing yourself to everyone within radio range. Additionally, some (generally older) ham radio users tend to exclude newcomers or are even racist (I once received a transmission without callsign (which is illegal) calling for killing n-words). In short, I do not consider getting or using a license to be safe for people in vulnerable situations.

To anyone considering getting a license: the address you use does NOT have to be your actual home address; just one you are able to get mail at. But yes, remember that the address will be publicly available, along with your name, so choose carefully.

I've yet to actually have a verbal conversation on amateur radio; I've been having enough fun playing with WSPR. Those folks are generally pretty nice. I've received some signals here in Germany from US callsigns that, upon looking them up on qrz.com, turned out to be older men I would absolutely cross the street to avoid, but others from old guys I would happily buy a beer or coffee to get their stories, such as one long retired Chicago-area broadcaster who runs a website of Chicago TV history, including footage of a young, carefree-looking Obama. The former usually are transmitting with waaay too much power for Weak Signal Propagation Reporter; the latter are playing by the spirit of the thing and are doing cool nerd things into their 70s and 80s.

Dig deep into understanding capacitors, inductors, and all the first order math around them, then impedance, then get a grasp of the hidden capacitors in inductors, and the hidden inductors in capacitors, and you'll be in a pretty good position, with a path that you should be able to see fairly well.

The ARRL Handbook is a wonderful place to start.

I always prefer the FFT/waterfall visualizations instead of amplitude/time scope graphs when talking about mixing and modulation. Makes it way simpler to reason about I think.

There is a high level mixer (I think it's called that) design that uses a 3157 video switch chip. I tried to build one a while ago and found some manufacturers putting the chip under the 74LVC series. I thought this article is about that chip but this is mone amazing.

Analog multiplexer employed as an RF mixer is pure genius.

Of course you can only drive it with square-wave, and mixing with square-wave, intuitively, feels like a terrible idea (think of all the harmonics!). But surprisingly, listening to the recording of the whole thing working, end result sounds A-okay!

If you think about how a diode mixer works then you're using a square wave LO, same for a Gilbert cell.

Remember, a square wave is just a sum of odd harmonics. The only downside is that a strong signal at a multiple of the signal of interest might get aliased into your IF. Hence bandpass filtering on the input is important.

I'll have to build one of these when I find the time. I wonder how hard it would be to turn it into a SSB receiver.

I don't have much circuit expertise, but in theory I believe demodulating SSB can be done by mixing (multiplying) a sine wave 2x the SSB center frequency, and summing that with the original signal (and then filtering the large frequency residues of course). This brings the opposite bands next to each other (then you can demodulate as DSB normally).

Analog multiplexers work fine for dorect conversion receivers, so you could just take the audio from the mixer output. Sadly that gives you audio from both sidebands.

If you mix in some passives or actives a Phasing receiver could be built to do away with the unwanted sideband. Or maybe one could use the Weaver method for ssb demodulation. But Weaver has it's problems with the switching frequency breaking thru. But that's less of a problem on receive.

Spock: “I am attempting to construct a Mnemonic memory circuit using stone knives and bear skins.”

The use of an analog mux for the mixer is a brilliant hack I never thought of before. It wouldn't work for "pure" applications like ring modulation because it's not a general 4-quadrant multiplier, but as an IF mixer it's fine. As the article demonstrates.

I like the little prototyping modules. One has a few pots, another the logic chips ... they seem to all take their power with wire pairs that join a narrow "power-bus module". 3D printed boxes with prototyping PCBs... Looks like fun.

>Also it is rather priceous (around 2€)

What is it with electrical engineers and an absolutely out-of-whack sense of "expensive"?

The SMD is 2€ through the big parts distributor where he can get all the other parts, and pay for shipping just once.

But he wants DIP, which he said they don't have in stock. So now he pays extra for DIP, extra for rarity, and again for shipping from another seller. It costs probably more like 10€ at this point. Why pay 5x when you can pay 0.5x?

You don't want to waste money like that on multiple parts in each project, across multiple projects each year.

When you are designing for runs of 50k - 1000k units, this is expensive.

But clearly the author of the article is not.

Reply on news.ycombinator.com