Here's this week's datasheet-a-day reading...

PIC Microcontroller Compiled Tips 'n' Tricks Guide (Microchip DS01146B)

I've never used a PIC microcontroller in my life, but this is still surprisingly interesting for a beginner, because it has sections about a whole bunch of common microcontroller interfacing problems: the sections about PWM, comparators and motor control were particularly helpful.

Basic Concepts of Linear Regulator and Switching Mode Power Supplies (Analog AN140)

I've been learning quite a bit about switching regulators recently, but strangely enough, I still didn't know exactly what a linear regulator did. So the quick explanation of linear regulators in this app note was very useful for me!

The bulk of the rest of the app note is taken up with a detailed treatment of a buck converter design. Things I took from this:

  • It's surprisingly easy to calculate power losses in switched-mode power supplies from basic principles, at least the DC losses. The losses from switching are a tougher proposition.

  • Which lead me to: I need to learn (a lot) more about MOSFETS. Much of the material here on switching losses was a total mystery to me.

  • Another thing I need to learn about in detail is how to design feedback loops for amplifiers (for compensation, etc.). This is really important for designing robust circuits (or even just circuits that work!) and, again, it's something I know relatively little about. (I'm hoping some of the motor control work for the Mini-Mapper might at least get me familiar with some of the theoretical side of this.)

There are also good references to a long list of other Analog app notes in here.

Taking the Mystery out of the Infamous Formula, "SNR = 6.02N + 1.76dB," and Why You Should Care (Analog MT-001)

I asked on the Contextual Electronics forum for app note recommendations, and this was the first one someone suggested. It's really pretty superb, and there are a whole series of these "Mini-Tutorial" app notes that look equally good.

This one explains in some detail the origin of the ideal ADC SNR equation, which says that the signal-to-noise ratio for an ideal SNR (i.e. with only quantisation noise) is given by SNR = 6.02N + 1.76dB, where N is the number of bits in the output. The explanation is simplified from the detailed analysis and is based on a simple model for quantisation noise uncorrelated with the signal. But that explanation is both convincing and simple enough that after one read through of the app note, I can probably reproduce the argument in enough detail to get something close to the right answer. That's a sign of a very clear explanation!

The app note also explains a little about oversampling, and has one example that includes both oversampling and undersampling that looks like a total cheat! If I understand right, they're saying you can sample a signal modulated onto a carrier by just sampling at a carefully chosen frequency in the baseband, without doing the whole complicated demodulation process at all! I think I need to think about that one a bit more...

The last part of the app note shows some of the weird stuff that can happen if the "noise not correlated with signal" assumption doesn't hold.

Low-Cost Open Source Ultrasound-Sensing Based Navigational Support for the Visually Impaired (paper)

This isn't an app note, but a paper. Well, I called it Datasheet-a-Day, and I've been reading things that aren't datasheets, so reading things that are even more not datasheets seems reasonable...

This is a report on an interesting low-cost assistive device for partially sighted people. It's very simple: an off-the-shelf ultrasonic distance sensor connected to an Arduino Nano with a small vibration motor, all in the 3-D printed case worn as a bracelet. The code running on the Arduino is also very simple. The idea is to give "distance to obstacle" feedback to blind or partially sighted users by different vibration patterns on their wrist. Because the ultrasonic sensors are mounted in a bracelet, the user can steer the sensor beam, which makes for a much more natural "active sensing" experience.

It just goes to show how much you can do with very little! The researchers were explicitly aiming to make a cheap device (total cost was about US$24) accessible to users in developing countries. The experiments they did were all with blindfolded sighted volunteers, so not totally realistic (blind and partially sighted people are obviously much better at navigation and orientation without sight than sighted people are), but the system looks pretty practical and simple to use.

PCB Layout Considerations for Non-Isolated Switching Power Supplies (Analog AN136)

This has some good detailed stuff about PCB layout for switched mode power supplies. I think it's something I'll be coming back to as I try to start learning some of what's needed to be able to develop the kind of layouts you see in datasheets myself. There's a lot in this app note, and it seems like something to have at your side as you do the layout for one of these things. It even has checklists to help make sure you don't miss anything. (I'm a big fan of checklists!)

Two obvious things I picked up from a first read:

  1. It's important to identify steady and pulsating currents in and near the "hot loop" that switches current through the inductor in the supply. Getting the layout for those pulsating currents right is critical.

  2. Use Kelvin via connections for sense resistors. These are the funny looking connections that are made with vias between the pads of an SMD sense resistor.