Believe it or not, it’s actually quite difficult to generate random numbers on computers and microcontrollers. I’m talking about true random. A lot of the randomness computers use isn’t random at all, it’s predictive (or pseudo random). Difficult, but still predictive. It’s important when doing cryptographic functions to use a good random source.
So f4grx decided to build a random circuit and do some analysis on it. The circuit design is built around the idea of using an open collector on a transistor. What’s essentially going to happen is tiny electrical noise in the air is going to cause the circuit to produce 1′s and 0′s which then can be used to plot random.
Not happy with your cable modem performance? Crack it open and fix it! That’s what forum user Xymox did. After cracking open his DPC3012 he noticed there were a few spots on the main PCB that were not populated with parts. The theory is they saved on budget and didn’t populate all the parts.
Xymox stuffed a huge amount of tantalum capacitors all along the power rail. He also added large capacitor on the power supply. It appears that the modem also runs pretty hot so he addressed the temperature issues by adding Arctic Silver to the heat sink and a fan. After the modifications are all done to the modem, he measured a 1000:1 reduction in power noise.
Having a spare unmodified modem to compare against, Xymox states a gain of 1db to the signal to noise ratio. The upstream power decreased by 1.5db meaning the modem has to try less as hard to maintain a good signal. There is also a side by side plot of ping times captured over a week. The average ping time has improved as well.
The guys over at SDR for Mariners put together a nice writeup on reducing the electrical interference on the good ol’ cheap RTL-SDR dongle. When the signal you’re looking for is very faint, you need to remove all of unwanted noise or else your signal gets difficult to separate. In their case the unwanted signal is in the form of FM radio.
Even if you aren’t a big RF nerd you know some of these remedies already like ferrite beads (you know those giant blocks on the ends of the expensive cables!). But I’m actually surprised at some of the other ones. I would have thought that the metal connector around a USB cable would help prevent noise, but according to their tests it was acting like a collector for their intrusive FM signal. By removing the shield and just connecting the data pins they dropped the noise by 10db. Not using the extension cable and just plugging straight in is even better.
So the winning combo appears to be all of their methods tested. A stick wrapped in aluminum foil, touching the metal of the USB plug, connected by a USB extension cable, inside a metal housing.
I’d love to see a picture of the final setup. Good tips to know when chasing down a faint signal.
Joo Won Park is doing some pretty creepy noise experiments. He runs a site called 100 strange sounds and has some pretty awesome audio experiments. In this one he is using a telephone pickup device, that is converting the EMF noise into sound. Then he adds a little bit of post processing that alters pitch and envelope. The result is some pretty creepy stuff honestly.
Another cool experiment of his (that I’ve seen before) is the no-input mixer. Which is an audio mixer that is in crazy feedback mode and actually doesn’t have any input going to it. The inefficiencies of the mixer are fed back into its self looping really really fast creating noise. Then altering the mixer and knobs / sliders cause it to alter into weirder noise. I recommend checking out more of his videos, this guy has some really cool experiments going on like playing a pool noodle.