Noah's Blog

Antenna Theory Analysis and Design

2026-05-21T00:00:00-05:00

Why I Read This Many of my learning excursions come from long waits in the airport, because as an engineer, it is impossible to be bored at an airport. My desire to learn more about antennas came from reading about the Instrument Landing System (ILS) antennas after seeing them on the runway. After my reading, I wanted to learn more about how antennas worked and how they were designed.

BitArmyKnife - The Electronics Multitool

2025-09-29T01:00:00-05:00

Description:

The BitArmyKnife is an all-in-one multitool for electronic communication. It is essentially a USB to everything converter with applications from serial communication debugging, flashing FPGA’s, dumping flash chips, and interpreting low speed digital logic. Specifically, this chip combines an FTDI2232HL and a CypressFX2 based logic analyzer on one PCB, creating two separate devices on the same board.

Some features include:

Level shifting to support a wide range of logic levels
Clearly labeled 1.5 mm pin headers for each communication interface
Indicator LEDs for data transmission and communication status

Project Motivation

While designing flight computer PCBs for Princeton Rocketry and testing peripherals like GPS modules, compasses, and radios (all of which use serial communication), I found the need for a device independent of the system under test. This would let me validate communication between hardware peripherals without relying on unfinished flight computer firmware.

One specific example was when I implemented satellite communication on a high altitude glider using a RockBlock satellite modem. I needed to confirm that my ground station code was transmitting commands correctly over the Iridium satellite network, so I connected the RockBlock to my computer via the BitArmyKnife and was able to verify I was receiving data on the UART that would eventually be connected to the flight computer.

Very useful! I did not have to configure the flight computer or halt the development of other parts of the project at all – I was able to develop this system in isolation and integrate it quickly into the rest of the glider.

Design Process

The BitArmyKnife began with a few goals in mind: interpreting various types of serial communication, and having the capability to decipher logic of an unknown signal. After some quick googling, I stumbled upon the Tigard project, an open source FT2232H USB to Serial board, which I took much inspiration from. I examined the Tigard Design and picked a few key decisions that I wanted to implement:

Level Shifting: BitArmyKnife supports 1.8V, 3.3V, and 5V logic levels, and I made sure to pick regulators large enough to power a connected device. There is even an option to use the voltage of the target source, which is often safer. This is accomplished by a SP4T switch that can select the logic level that powers the output side of a SN74LVC8T245PWR Bus Transceiver. So, the 3.3V logic from the FTDI chip enters one side of the buffer, and exits shifted to whatever logic level is selected by the switch.

The Mode Switch: The magic of this device is its ability to accommodate so many different communication protocols. This is done through a “Mode Switch” that simply connects and disconnects certain nets coming from the FT2232H to protocol specific header pins on the BitArmyKnife. To fully understand how it works, you need to know some things about the FTDI2232HL:

There are two channels on the chip (A and B). BitArmyKnife uses channel A exclusively as a UART. Channel B is where things get weird. Channel B is configured to use FTDI’s MPSSE or Multi-Protocol Synchronous Serial Engine. According to AN_135 Application Note, “the MPSSE allows communication with many different types of synchronous devices, the most popular being SPI, I2C and JTAG.”

\ \

It is important to know that SWD and I2C function differently than these other synchronous protocols because they have bidirectional data lines (SWDIO and SDA). This is solved by tying DO and DI together with a 330 ohm resistor when the Mode Switch is in SWD/I2C mode. The resistor enables the FTDI2232H and the device to read and write to the same line while limiting current into whichever pin (DO or DI) is not being used. Note, the 0 ohm jumper is not used normally; it can be used for fast signals and lengthy wires where signal integrity is more important than protection of the device.

I also knew that I wanted a primitive logic analyzer to decipher other kinds of digital signals, and after watching a few youtube videos about cheap FX2 LA’s, I decided to add one to the PCB.

Finished Result

The BitArmyKnife is working flawlessly so far! I have tested UART, SPI, and I2C communication, verifying that the FT2232HL was hooked up correctly and the mode switching works. I am very happy with the PCB aesthetically as well; my routing skills have matured and I put a lot of effort into planning out clean and optimal traces. I successfully used this tool to communicate with the RockBlock modem, a uBlox GPS, ht16k33 based 7 segment display, and a Teensy microcontroller.

I have not tested the logic analyzer (the semester started), but I can verify that the Cypress FX2 powers on and is recognized by my computer. I will continue to post about this device after I verify more of its features.

Silcon VLSI Technology

2025-09-29T00:00:00-05:00

Why I Read This

After finishing up my summer internship researching the deposition of ZnO films with Atomic Layer Deposition, I was fascinated with semiconductor manufacturing technologies; how could we make these incredible modern computer chips? After a long conversation with my advisor, he pointed me to this book to answer all of my questions.

What I Learned

Key Developments that make Today's ICs Possible - Invention of the transistor at Bell Labs. (This started from a search to switch telephone signals with a solid state device. So cool!) - Development of the planar process by Jean Hoerni of Fairchild Semiconductor. This marked the transition from Ge to Si, which was crucial for the industry. SiO2 is easy to deposit on Si and can insulate N and P type regions, allowing more control over electrical properties and smaller gate lengths. - The development of the IC. This really began the transistor revolution and paved the way for Moore's Law.

Semiconductor Physics - "The key to building semiconductor devices and integrated circuits lies in the ability to control the local doping and hence the local electronic properties of a semiconductor crystal." - Nonzero thermal energy causes Si-Si bonds to break in a Si crystal, creating a mobile electron and a "hole" in the crystal structure. These are considered charge carriers becuase they carry electrical currents through the device. - Doping enables us to control and engineer carrier concentrations to tune the performance of a device - Fermi level, probability distribution of electron energy states

PN Junctions - PN Junctions are a result of an equiliobrium between diffusion and electromagentic forces. - Depletion Region: At the junction, there is a large gradient of holes and electrons going from one side to the other. As a reult, they both diffuse across the junction. As they diffuse across the junction however, they leave behind donor and acceptor ions in the Si lattice that essentially try to pull them back through an electromagnetic force. So, the holes and electrons eventually reach a steady state where they are balanced across an electrically neutral depletion region.

Read in Progress! I am actively reading this book when I have time outside of class. I am currently learning many of these semiconduconductor manufacturing techniques through hands-on work experience, and I am excited to continue reading this book and deeply understand the theoretical principles that enable the creation of some of the most advanced technology in the world.

Resurrecting an old ADC - Motorola MC10319

2025-06-25T02:00:00-05:00

Description:

The Motorola MC10319 was a high speed parallel 8-bit flash ADC developed in the 1990s. I could not find direct examples of what this chip was used for online, but the datasheet claims the chip was designed for video broadcast, radar processing, and video display. In my case, this chip was taken out of a broken oscilloscope. A PhD student gave me this chip and told me that I would learn a lot about electronics on my journey to making it function again. He was right.

Project Overview

I broke this project down into three main systems I needed to get working: the main supplies VCC(D) and VCC(A), resistor ladder supplies VRT and VRB, and the parallel data output.

Power Supplies

For the +5V rails, I used my bench supply for simple smooth power. However, I needed -5V to power the comparators and -1.25V for the bottom of the resistor ladder to accommodate negative signal inputs. I settled on using a buck-boost converter to convert my +5V bench supply to -5V.

The datasheet provided this example and I decided to mostly follow it. It uses an MC34063A switching IC to invert the voltage. I tested this circuit with a 330 ohm resistor to simulate a load while staying under the 20mA maximum output current. However, the output voltage never went below -3V. I adjusted the feedback resistor values (1k and 3k) but could only get -5V with a very small load. I determined that there was some sort of load regulation problem due to a parasitic source inductance caused by terrible breadboard wiring. I re-routed the circuit using less wires and shorter loops, fixing the problem.

At this point the VRT and VRB supplies were trivial; I just used +1.25V and -1.25V linear regulators respectively to create the stable voltages required by the resistor ladder.

Data Output

To process the output of the ADC, I settled on a Teensy 3.2 since it is what I had around. The teensy had to process the 8bit output and also provide the clock signal, since I wanted to keep my oscilloscope free to measure waveforms (it has an integrated signal generator).

To get the fastest read possible, I connected pins D0-D7 on the MC10319 to the first 8 pins on PORT C on the Teensy. This will enable me to read the port register all at once instead of reading each pin sequentially. To find the first 8 pins of PORT C, I used the core_pins.h file located in Arduino15/packages/teensy/hardware/avr/1.59.0/cores/teensy3.

Using this, I connected D0-D7 to Teensy Pins 15, 22, 23, 9, 10, 13, 11, 12 in order. Then I could use uint8_t adc_value = GPIOC_PDIR & 0xFFto get the first byte of the PORT C register corresponding to the adc output value.

Quick Test

I hooked up a potentiometer to VRT and VRB and made the input to the ADC (VIN) the wiper of the potentiometer. This allows me to sweep the complete range of the ADC. And it works! The ADC values were 255 with VIN = VRT and 2 when VIN = VRB. I thought that I should get 0, but even with VIN shorted to VRT I still get 2. I did more testing and found that when VIN = 0 = VRM, I get a value of 130 instead of 128. So it seems there is an offset of 2. I troubleshooted this, and attributed it to the breadboard wiring. This is a very sensitive IC after all, and I am very happy with the accuracy I am getting, even with decoupling capacitors inches away from the IC.

I also tested my understanding of the clock cycle. On the rising edge, the data output latches receive old data and output it to the data bus. During the high time, the comparators follow VIN but do not latch any data. On the falling edge, the comparator outputs the data taken right before the edge, and this data is latched to the data outputs. During the low time, everything remains latched and the chip waits for the rising edge again.

I tested this by using a button as a clock signal so I could manually advance the system timing. I got clean edges with a 6.8k pull down resistor on VIN and a button connecting VIN to +5V. When the clock was low, I adjusted VIN using the potentiometer and the ADC output did not change. This is because the data output is still latched with old data, and the comparators are not free to follow VIN. When I press the button, old data is latched and I see the ADC value change. While I hold the button, I adjust the potentiometer to allow the comparators to follow the changing VIN, but my ADC value doesn’t change since the output latches still have the old data. When I release the button, the ADC value changes to a value representing the last known position of the potentiometer, latching the last value of the comparators prior to the falling edge. Makes sense!

Future Plans

Now that I really understand how a flash ADC functions, I want to process a real analog signal, not just a potentiometer. I am planning to make a simple guitar tuner, by processing the raw output from a guitar.

I also want to de-cap this chip and examine it at the silicon die level. I plan to label the resistor ladder and the comparator network, find the binary encoder, and all of the data latches.

Ultrasonic Sonar Project

2025-05-15T02:00:00-05:00

Overview

I built an ultrasonic sonar system using an ESP32 and custom circuitry. The system emits ultrasonic pulses, measures their return time, and visualizes object distance on a radar-style display. It mimics commercial sensors like the HC-SR04 but is fully built from low-level components and software.

Design Summary

Transmitter: Uses a level-shifted 40kHz square wave to excite an ultrasonic transducer. I used the MAX232 and capacitors to generate the 15V pulse required.
Receiver: A 3-stage op-amp (LM324) circuit amplifies the weak echo signal. Careful DC offset tuning ensures ESP32 compatibility.
Signal Processing: Distance is found by timing echoes using the ESP32 using micros().
Rotating Scan: A servo sweeps the sensor across an area, and the ESP32 sends angle and distance to a Processing 4 script, which visualizes detected objects in real time.

Key Challenges & Learnings

Op-Amp Sensitivity: I learned to tune analog circuits using a signal generator and potentiometers. Finding the balance between noise and sensitivity proved difficult.
Software Timing Conflicts: Running both the servo and sonar timing code on one ESP32 introduced errors. I learned the limits of precise timing on a single microcontroller and resolved it by writing a custom PWM function instead of using the Servo library. The sensor would perform much better with a dedicated microcontroller.
Noise & Artifacts: The circuit was prone to noise. Though we added a high-pass filter, a more targeted band-pass would have been ideal.

What I Took Away

This project strengthened my understanding of both analog signal design and embedded systems. From carefully creating and tuning op-amp circuits to writing low-level ESP32 code, it brought together theory and hands-on problem solving.

Detailed Writeup

This project was part of a final project for my circuits class, so here is a detailed writeup created by me and my partner: 📄 View the full PDF report

PCB Design for Real World EMI Control

2025-04-16T00:00:00-05:00

“Rather than depend on luck and many hours spent in the EMC laboratory trying various circuit combinations, up front design and consideration of the parasitic circuit elements, the possible resonances, and the overall equivalent circuits will help the designer be successful the first time” (Page 21)

Why I Read This I was designing a flight computer PCB following routing guidelines and common practices for PCB Design, but I wanted to know exactly why I was doing these things. So, I found this book that provides a theoretical backing to why these common practices exist so I can understand what's really going on to make informed design tradeoffs.

What I Learned Two Coupling mechanisms: Electric Field and magnetic Field - Electric Field: Occurs through displacement current which has a parasitic capacitance effect - Magnetic Field: Occurs through current and inductance

These get dangerous when they offer a lower impedance return path then the intended one

High frequencies radiate more than lower frequencies. Often it is not the fundamental frequency that is the problem, it is the higher harmonics.

Inductance - At high frequencies, inductance dominates - Lots of formulas to calculate inductance of different geometries