GBPPR Non-Linear Junction Detector | Hacker News Reader

GBPPR Non-Linear Junction Detector

web.archive.org

27 pointslainon20 comments

One of the great stories of using "Non-Linear Junction" devices occurred circa 1985, during the Cold War between the US and the Soviet Union.

The US was building a new embassy in Moscow. Naively, they allowed the Soviets to do most of the work. The result was a building impossible to secure. Eventually the top 4 stories needed to be completely replaced.[1]

The relevant part to this discussion is that the Soviets mixed thousands of discrete diodes[2] into the precast concrete. What is a diode? A nonlinear junction. What does it cost? A few pennies. What does it do? It creates false positives, so it keeps you from using a device such as the one discussed in this article to detect actual bugs.

[1] https://en.wikipedia.org/wiki/Embassy_of_the_United_States,_... [2]https://www.washingtonpost.com/archive/politics/1990/06/18/u...

The USSR was extremely skilled at creating listening devices that didn't have nonlinear junctions, or indeed any active electronics at all:

https://en.wikipedia.org/wiki/The_Thing_(listening_device)

Yes, that's a famous one. I got a chuckle out of the photos of Pelosi and Schumer leaving the White House after meeting with Trump a while back ( https://twitter.com/cbouzy/status/1072563584796909568 ). It's as if the Russians aren't even trying to hide their bugs anymore.

What's going on in that picture?

I don't know if you're up to speed with US politics, but at the time the photo was taken, they were the Congressional minority leaders. They had just told the President that they wouldn't support funding for his border wall project:

https://www.realclearpolitics.com/video/2018/12/11/trump_mee...

The President insisted on debating the issue with them with the press in attendance, which is somewhat unusual, and they embarrassed him rather badly as a result. They are strutting away more or less triumphantly. Whether that was a good or bad thing isn't a good thing to get into here, but that's basically the gist of it.

The joke is that the wooden plaque above the front door to the White House resembles the 'Great Seal' bug that userbinator mentioned. This was an advanced listening device that the Soviets presented as a gift to the US ambassador at the end of WWII. It was advanced in the sense that it was entirely passive, requiring no power or vacuum tubes, and was very hard to detect with the technology of the day. It could be said to be among the first applications of modern RFID technology.

Ah, the plaque. Thanks.

That WaPo article is great. I'd never heard of "Batwing" before. Sounds like it may have been a radiothermal generator reminiscent of those used for long-duration space missions http://large.stanford.edu/courses/2012/ph240/belanger2/ . The Soviets were known to be liberal in their use of small-scale atomic power sources elsewhere.

Any theory on why PN junctions emit even harmonics and natural nonlinear junctions emit odd harmonics?

A PN junction (ie, a diode) only passes current one way, so a sine wave applied across a PN junction gets turned into only half a sine wave, with either the positive or negative half clipped depending on the polarity. The fourier series of a rectified sine wave is mostly the second harmonic.

A natural nonlinear junction usually consists of a thin layer of insulating oxide between two pieces of metal. It only conducts when the voltage exceeds some threshold where the electrons tunnel through the insulator. When a sine wave is applied, both the tops and bottoms of the waveform are clipped. Because it's symmetrical around zero, the energy it re-radiates is mostly at the third harmonic.

Makes sense, thanks.

does this in only detect junctions connected to an [explicit or implicit] antenna or does it also detect any junction without the junction needing to be connected to an antenna?

my intuition is this method would require the detected junction to be connected to at least an implicit antenna to which the detector can couple, but I could be wrong?

Anything that's more or less conductive can be an antenna if the frequency is high enough. That's one reason to design these instruments to operate in the 2.4 GHz ISM band, where the wavelength is around 12 centimeters. Junctions associated with metallic objects over 1 cm in size should be readily detectable.

Another way to think about it:

If the voltage-to-current transfer characteristic of whatever you're working with is symmetric about the y axis, then taking the maclaurin series gives you a polynomial with only odd terms. If you apply a sinusoid to that, you get odd multiple harmonics.

Working through the same reasoning with an asymmetric transfer characteristic means you get even polynomials as well, which means you get both even and odd harmonics.

I wonder how much easier/cheaper this would be with modern technology/parts (archive.org shows that the link to this project on the overview page already existed in 2008, which is their first capture, and the PCB style looks quite old).

It should be noted that this is for US use-only, since it transmits on the 900 MHz band (FCC part 15).

In Europe this frequency falls into the GSM band and will get authorities to your doorstep in a hurry.

Kind of a bummer, doesn't look like we ever find out if his project worked or not.

I read this as “GDPR non-linear function detector” for neural network regulation

same

Yeah haha I read GBPRR and thought "Oh no is this another law the EU wants to introduce?"

I vouched for this, because I thought exactly the same thing! And wherever killed it needs to get a sense of humor.

And, I'm giving a talk at CircleCityCon about signals intelligence. One area I haven't gotten into is NLJDs.... one can only diversify into so many areas :)

Edit: Hmm. an immediate -1. Someone's got an axe to grind.

Reply on news.ycombinator.com