The Perils Of Return Path Gaps

March 15, 2024 by 9 Comments

The radio frequency world is full of mysteries, some of which seem to take a lifetime to master. And even then, it seems like there’s always something more to learn, and some new subtlety that can turn a good design on paper into a nightmare of unwanted interference and unexpected consequences in the real world.

As [Ken Wyatt] aptly demonstrates in the video below, where you put gaps in return paths on a PCB is one way to really screw things up. His demo system is simple: a pair of insulated wires running from the center pins on BNC jacks and running along the surface of a piece of copper-clad board to simulate a PCB trace. The end of each wire is connected to the board’s ground plane through a 50 ohm resistor, with one wire running over a narrow slot cut into the board. A harmonics-rich signal is fed into each trace while an H-field EMC probe connected to a spectrum analyzer is run along the length of the trace.

With the trace running over the solid ground plane, the harmonics are plentiful, as expected, but they fall off very quickly away from the trace. But over on the trace with the gapped return trace it’s a far different story. The harmonics are still there, but they’re about 5 dBmV higher in the vicinity of the gap. [Ken] also uses the probe to show just how far from the signal trace the return path extends to get around the gap. And even worse, the gap makes it so that harmonics are detectable on the unpowered trace. He also uses a current probe to show how common-mode current will radiate from a long conductor attached to the backplane, and that it’s about 20 dB higher with the gapped trace.

Hats off to [Ken] for this simple explanation and vivid reminder to watch return paths on clock traces and other high-frequency signals. Need an EMC probe to check your work? A bit of rigid coax and an SDR are all you need.  Continue reading “The Perils Of Return Path Gaps” →

This Week In Security: Apple Backdoors Curl, Tor’s New Bridge, And GhostRace

March 15, 2024 by 8 Comments

OK, that headline is a bit of a cheap shot. But if you run the curl binary that Apple ships, you’re in for a surprise if you happen to use the --cacert flag. That flag specifies that TLS verification is only to be done using the certificate file specified. That’s useful to solve certificate mysteries, or to make absolutely sure that you’re connecting to the server you expect.

What’s weird here is that on a MacOS, using the Apple provided curl binary, --cacert doesn’t limit the program to the single certificate file. On an Apple system, the verification falls back to the system’s certificate store. This is an intentional choice by Apple, but not one that’s aimed particularly at curl. The real magic is in Apple’s SSL library, which forces the use of the system keychain.

The current state of things is that this option is simply not going to do the right thing in the Apple provided binary. It’s documented with the note that “this option is supported for backward compatibility with other SSL engines, but it should not be set.” It’s an unfortunate situation, and we’re hopeful that a workaround can be found to restore the documented function of this option. Continue reading “This Week In Security: Apple Backdoors Curl, Tor’s New Bridge, And GhostRace” →

CATS mobile transceiver in a 3d-printed case

CATS: A New Communication And Telemetry System

March 15, 2024 by 27 Comments

CATS is a new communication and telemetry standard intended to surpass the current Automatic Packet Reporting System (APRS) standard by leveraging modern, super-cheap Frequency Shift Keying (FSK) transceivers rather than standard FM units. The project is in the early stages, but as of this writing, there is a full open source software stack and reference hardware for both Raspberry Pi-based gateway devices and an STM32-based mobile device.

CATS packets are called ‘whiskers!’

From a radio perspective, CATS uses raw FSK rather than the inefficient AFSK used by APRS. A real killer for channel utilization is the PTT time; this is the dead time around a packet APRS requires for ‘keying up’ and ‘keying down.’ The CATS standard is aggressive with PTT timing, enabling the channel to get going on sending the data sooner.

Additionally, compared to APRS, the packet baud rate increases from 1200 baud to 9600 baud. Other key points are using LDPC encoding for forward error correction and data whitening (a useful PDF guide from Ti) to smooth over any burst errors.

One of the neat concepts of APRS is the APRS-IS (APRS Internet service). This enables amateur radio services to be connected over the Internet, vastly improving range. The CATS equivalent is called FELINET (if you’re not spotting all the ‘cat’ references by now, go and get another coffee). Together with the I-gate hardware, FELINET bridges the CATS radio side with the current APRS network. As FELINET expands to more than the current few dozen nodes, APRS services will no longer be required, and FELINET may well replace it. Interestingly, all software for FELINET, the APRS relay, and the I-Gate firmware are written in Rust. We told you learning Rust was going to be worth the effort!

On the reference hardware side of things, the CATS project has delivered a Raspberry Pi hat, which uses a 1 watt RF4463 transceiver and supporting passives. The design is about as simple as it can be. A mobile transceiver version uses an STM32 micro to drive the same RF4463 but with supporting power supplies intended to run from a typical automotive outlet. Both designs are complete KiCAD projects. Finally, once you’ve got some hardware in place and the software installed, you will want to be able to debug it. CATS has you covered with an RTL-SDR I-Gate module, giving you an independent packet log.

APRS is quite mature, and we’ve seen many hacks on these pages. Here’s an earlier APRS IGate build using a Raspberry Pi. Need to hook up your PC to a cheap Chinese transceiver? You need the all-in-one cable. As with many things amateur-radio-oriented, you can get playing cheaply.

A Look Inside A 70-GHz Electromechanical Attenuator

March 15, 2024 by 4 Comments

It might not count as “DC to daylight,” but an electromechanical attenuator that covers up to 70 GHz is pretty close, and getting a guided tour of its insides is quite a treat.

Perhaps unsurprisingly, this one comes to us from [Shahriar] at “The Signal Path,” where high-end gear most of us never get a chance to work with goes for one last hurrah after it releases the magic smoke. And indeed, that appears to be exactly what happened to the Rohde & Schwarz 75 dB step attenuator, a part that may have lived in the front end of one of their spectrum analyzers. As one would expect from such an expensive component, the insides have some pretty special engineering. The signal is carried through the five attenuation stages on a narrow strip of copper. Each stage uses a solenoid to move the strip between either a plain conductor or a small Pi pad with a specified attenuation. The attention to detail inside the cavity is amazing, with great care taken to maintain the physical orientation of the stripline to prevent impedance mismatches and unwanted reflections.

The Pi pads themselves are fascinating, too, especially under [Shahriar]’s super-duper microscope. All of them were destructively removed from the cavity before getting to him, but it’s still pretty clear what’s going on. That’s especially true with the 5-dB pad, which bears clear signs of the overload that brought on the demise of the whole attenuator. We suppose a repair would have been feasible if it had been just the one pad that needed replacement, but with all of them broken, it’s off to the scrap bin. Or to the recycler — there appears to be plenty of gold in there.

We thought this was a fantastic look under the covers of an exquisitely engineered part. Too bad it didn’t rate the [Shahriar] X-ray treatment, as this multimeter repair or this 60-GHz phased array did. Oh, well — maybe next time.

Continue reading “A Look Inside A 70-GHz Electromechanical Attenuator” →

Bonkers Nerf Blaster Sprays Balls Everywhere

March 14, 2024 by 3 Comments

Nerf blasters are fun toys, to be sure. However, they’re limited by factors like price and safety and what Hasbro thinks parents will put up with. Few caregivers would ever countenance a build like this one from [ItllProbablyWork].

It’s a blaster designed to fire 48 darts in a second or so, or a truly ludicrous 288 Nerf balls. Like so many rapid fire blaster designs, it’s based on a pair of rotating wheels which fling darts out at rapid speed. The trick to the rapid fire ability is the delivery of ammunition. In this case, the blaster has a rotating drum of 12 barrels, which can each be loaded with 4 darts or 24 balls. As the drum rotates into position, a trigger mechanism unlatches a spring which forces the contents of the barrel out through the wheels and on to the target.

It’s mostly pretty good with darts, but with balls, it tends to send them flying everywhere, including jamming a bunch into the blaster’s internals. It is very funny to watch, though.

We’ve seen some other great blaster builds recently, too. Video after the break.

Continue reading “Bonkers Nerf Blaster Sprays Balls Everywhere” →

Making Floating Point Calculations Less Cursed When Accuracy Matters

March 14, 2024 by 27 Comments
Inverting the earlier exponentiation to reduce floating point arithmetic error. (Credit: exozy)
Inverting the earlier exponentiation to reduce floating point arithmetic error. (Credit: exozy)

An unfortunate reality of trying to represent continuous real numbers in a fixed space (e.g. with a limited number of bits) is that this comes with an inevitable loss of both precision and accuracy. Although floating point arithmetic standards – like the commonly used IEEE 754 – seek to minimize this error, it’s inevitable that across the range of a floating point variable loss of precision occurs. This is what [exozy] demonstrates, by showing just how big the error can get when performing a simple division of the exponential of an input value by the original value. This results in an amazing error of over 10%, which leads to the question of how to best fix this.

Obviously, if you have the option, you can simply increase the precision of the floating point variable, from 32-bit to 64- or even 256-bit, but this only gets you so far. The solution which [exozy] shows here involves using redundant computation by inverting the result of ex. In a demonstration using Python code (which uses IEEE 754 double precision internally), this almost eradicates the error. Other than proving that floating point arithmetic is cursed, this also raises the question of why this works.

Continue reading “Making Floating Point Calculations Less Cursed When Accuracy Matters” →

Celebrating Pi Day With A Ghostly Calculator

March 14, 2024 by No comments

For the last few years, [Cristiano Monteiro] has marked March 14th by building a device to calculate Pi. This year, he’s combined an RP2040 development board and a beam-splitting prism to create an otherworldly numerical display inspired by the classic Pepper’s Ghost illusion.

The build is straightforward thanks to the Cookie board from Melopero Electronics, which pairs the RP2040 with a 5×5 matrix of addressable RGB LEDs. Since [Cristiano] only needed 4×5 LED “pixels” to display the digits 0 through 9, this left him with an unused vertical column on the right side of the array. Looking to add a visually interesting progress indicator for when the RP2040 is really wracking its silicon brain for the next digit of Pi, he used it to show a red Larson scanner in honor of Battlestar Galactica.

With the MicroPython code written to calculate Pi and display each digit on the array, all it took to complete the illusion was the addition of a glass prism, held directly over the LED array thanks to a 3D-printed mounting plate. When the observer looks through the prism, they’ll see the reflection of the display seemingly floating in mid-air, superimposed over whatever’s behind the glass. It’s a bit like how the Heads Up Display (HUD) works on a fighter jet (or sufficiently fancy car).

Compared to his 2023 entry, which used common seven-segment LED displays to show off its fresh-baked digits of Pi, we think this new build definitely pulls ahead in terms of visual flair. However, if we had to pick just one of [Cristiano]’s devices to grace our desk, it would still have to be his portable GPS time server.

Continue reading “Celebrating Pi Day With A Ghostly Calculator” →