DIY Life

We've reported on Bug Labs before. They make a sort of gadget LEGO system so you can easily build stuff like a GPS-enabled alarm clock. Not useful? Ask the guy who kept missing his train stop by falling asleep!

I spoke to Bug Labs founder Peter Semmelhack about what the company is up to, and how the future of DIY is shaping up.

Iron Man gear is all the rage these days, thanks to the popularity of the new hit movie. We recently went to my nephew's fourth birthday party, and as a gift, he received an Iron Man mask and an arm attachment that shoots discs. Ah, the weaponry a four-year-old apparently needs....

If you've got an Iron Man buff in your household, why spend the money on a store-bought mask and arc reactor when you can build one yourself? Details after the break.

Salvaging electronic parts is a must for any DIY'er. You save big on shipping charges, and recycling is good for the environment. Hack-a-day dissected an old computer mouse and found some useful components.

What useful parts are inside an outdated PC mouse? There are a bunch of sensors and buttons, including rotary encoders that can be used to measure movement in robotics projects. My favorite find is the microchip that glues the sensors to a computer. Learn about all these parts in detail, in the how-to.

Hack-a-day doesn't delve into the proper techniques for removing parts, but there are a ton of tutorials that can help you get started. It's possible to remove many parts with a simple soldering iron and an absorbent copper braid, called solder wick. Many use a solder sucker to vacuum solder away from parts, or a special desoldering iron. Surface mount chips can also be removed, but they may require the gentle embrace of 400 degree air from a hot air rework station, or the crude gust of a heat gun.

This article continues a series about building a DIY digital audio recorder. Inspired by this microcontroller audio project [via], I set out to build a simple digital recording device. I chose Texas Instrument's MSP430 microcontroller for this project because it's fast (16 MHz), it's cheap ($1), and it's very low power. Read the first part, and the second part.

This week we'll progress towards a digital audio recorder by playing audio files from a SD memory card. First, we'll convert an audio file to a raw format and image it directly to a SD card. Then, we'll interface the SD card with the MSP430 and play an audio file. See it in the video:


Next time we'll extend this basic circuit to include a microphone and audio recording capabilities.

Read on to learn more about generating audio with a microcontroller.

Overview
Read the previous article in this series for the fundamentals of microcontroller audio. Learn about basic MSP430 connections and programming in the MSP430 primer. This week we'll use the tiny, inexpensive MSP430 microcontroller to play audio from a SD card.



Playing an audio file
In the previous project, I used the MSP430 pulse-width modulator (PWM) to generate simple tones. The PWM duty cycle changed the frequency, and thus tone, of the generated signal. The signal is cleaned by a simple audio filter, and amplified by powered PC speakers. Hear it in this video clip from the previous article:




Each value generates a different frequency. If we change the values about 8000 times a second (8KHz) we can reproduce telephone quality audio.

8 bit audio (each sample requires one byte) at 8KHz requires 8000 bytes of data per second. 16 bit audio, the quality associated with CD players and PCs, requires twice as much data per sample (16 bits or 2 bytes) -- 16 kilobytes/second.

With 2K of internal flash, the MSP430 can store about one-quarter second of audio. External memory is needed to store any meaningful amount of audio. There's no better source of cheap storage than flash memory chips. A raw flash memory chip would be a special order item, but you probably already have some flash memory -- SD cards! These cards are ubiquitous in digital cameras and low-end MP3 players.

Working with SD cards
NOTE!!! We are going to use the SD card as flash memory. It will not have a full (FAT) file system, and can not be read on a PC by the usual methods. It will not be possible to drag and drop a raw audio file to the card and read it with the MSP430, nor will it be possible see recordings from the file browser on a Mac or PC. Special disk tools provide access to the raw disk, but a full FAT file system implementation is too big for the tiny microcontroller. The F2013 has 128 bytes of RAM and 2 kilobytes of flash. A workable FAT implementation requires 1024 bytes of RAM and several kilobytes of flash program space.

The prototype design accesses the SD card using the simple SPI protocol. SPI is a simple three wire interface that allows a microcontroller to read and write data from external chips. If you want the dirty details on how the SD card works in SPI mode, read some of these tutorials [Tutorial 1, Tutorial 2]. A source code library from TI provides easy access to the disk without writing our own initialization and read/write routines (mmc.c/mmc.h in the project archive).



SD card storage is divided into 512 byte chunks called sectors. A full 512 byte sector must be read or written at once. The MSP430F2013 has only 128 bytes of RAM -- not enough to hold a whole sector for random access. We can still read/write each byte in order, as it's needed. This method will suffice as long as each byte is read or written sequentially.

Playing an audio file is as simple as reading a byte from the SD card and writing it to the PWM duty cycle register.

Converting audio to a raw format
My 'hello world' audio clip comes from the intro to my videos. I converted the clip to an 8KHz .wav file with Audacity and then saved it to raw text file with the free version of Switch, as described here.

1. First, install Switch.
2. Start Switch, and add your desired files to the queue by clicking "add files".
3. Now set the output format to ".raw", and then click "Encoder Options". Set the encoder format to "8 bit unsigned", sample rate to "8000", and channels to "mono".
4. Now click convert and the raw version of the audio files are created.

The raw text version of the audio file is included in the project archive.

Imaging an SD card
The MSP430 isn't powerful enough to deal with a file system, so we need the audio to start exactly at the first sector the of SD card. A simple copy->paste operation would put the file on the SD card according to the rules of the PC file system. To avoid this, we image the file directly onto the card. This will be familiar if you've ever 'imaged' a CD .iso file, say, for a bootable Linux live CD. I used DiskImage 0.9 from Durban.

Imaging the audio file onto a flash card was pretty easy with DiskImage. I put the SD card in a USB flash card reader and attached it to the USB port. I opened DiskImage and it detected the card as a logical and physical volume.

1. It's crucial to deal with the disk directly, rather than through the file system. All operations should be done to the listings under "Physical drives - raw drive, independent of partitions" box.
2. Find the physical partition that represents your SD card - it's easy to identify the correct physical drive because it should be type "Removable" and the correct size (in my case 14MB). Be sure you don't overwrite your actual hard drive, this utility can easily do that!!!
3. Click on the correct drive to highlight it.
4. In the "With selected item do" box, click the "Import from file" button. Click yes, and type "i agree" after you verify that the correct disk is selected. Click yes again.
5. Next, DiskImage prompts for the file that will be imaged onto the disk. Choose the raw audio .txt file outputted earlier.

With the raw audio imaged onto the SD card, we're ready to play it from the MSP430. Take the SD card out of the computer and put it into the card holder on the prototype.

Firmware
The example program is included in the project archive. The example software is written using the demo TI/IAR Kickstart C compiler.



SD card access
The example firmware starts reading the SD card at the first sector. It reads one byte at a time over the SPI interface and places the byte in the PWM duty cycle register. When the end of a sector (byte 511) is reached, the next sector is immediately loaded and initialized so that the next byte is always ready when needed.

The example firmware will copy bytes from the SD card to the PWM register until it reaches the sector defined by the variable flashDisk.lastSector. At the end of this sector, the program begins again at the first sector.

The value to use here is determined by the number of sectors consumed by the audio file. The example audio file consumes 58,447 bytes. The SD card is arranged into 512 byte sectors, so the file ends at sector 115 (sector numbering starts with 0). Update this value if you are working with a custom audio file:

flashDisk.lastSector=115; //last sector (512byte block) where file is stored on flash...

Playback rate
The playback must match the sampling rate of the audio file, or the audio will sound too fast or too slow. Since I sampled audio at 8KHz, the PWM duty cycle register should be updated with a new value 8000 times each second. I appropriated the watchdog timer (WDT) from the MSP430 to sound an alarm (an interrupt) 8000 times per second. When the timer interrupts, a bit of code copies the next byte from the SD card to the PWM duty cycle register.

The timer runs on the calibrated 8 MHz internal clock. The WDT is set to trigger every 512 counts of the internal clock (8MHZ/512), or 15,625 times per second. This is about twice as fast as we need, so the interrupt routine uses a switch that updates the audio only once every other interrupt, or 7,812.5 times per second. Not exactly 8000 samples-a-second, but the internal oscillator will vary with temperature and age anyway. Using the internal crystal keeps the design simple and the part-count low.

If you're after tighter tolerances, consider using a watch crystal on the P2.6/7 pins of the MSP430 as the timer clock source.

When the MSP430 is not copying audio to the PWM, it enters a loop that continually checks if a new audio byte should be loaded into a single byte buffer. The buffer ensures that data is available when it's needed, and that audio quality isn't effected by delays in reading data from the SD card. A ton of power could be saved by entering sleep mode between interrupts, we'll look at this again in a future article.

Say Hello
In the video you can see the basic firmware functions. First, a breakpoint set at the timer shows how the audio is only updated on every other byte. If we move the breakpoint inside this switch, the debugger stops every time the audio changes. Next, execution halts each time 512 bytes are read from the SD card and the sector changes. A final breakpoint halts execution each time the sector resets to 0.

Next time
The next article will be the final functional enhancement to the digital audio recorder. We'll extend this basic circuit to include a microphone and audio recording capabilities.

Prototype Circuit
For complete details of the prototype, see the previous article in this series. This week's firmware is in the project archive.


Related links
Program a MSP430 microcontroller.
Make a singing MSP430 microcontroller.

Recently, here at DIY Life, Ian Lesnet has written some tutorial posts dealing with microcontrollers: devices that essentially act as the "brain" of a small electronics project.

Now Annie Scott, of our sister site, Styledash, has written about a microcontroller called the Lilypad Arduino. On the Arduino site, the 2" wide flower-shaped board is described as "a microcontroller board designed for wearables and e-textiles. It can be sewn to fabric and similarly mounted power supplies, sensors and actuators with conductive thread."

This is all very technical, but in practical terms, what does it mean? Among other things, it's a (relatively) simple way to make hand-washable t-shirts that light up or sing when you go into a dark room, or when you move. Read more about the Lilypad Arduino, including where to purchase it and how to use it, after the break.

My sister sent me her iPod telling me it's locked up: the hold button switch stopped working. It was stiff and felt like a piece of grit was in the switch. With the hold switch broken, all the other buttons stopped working as well, even while the screen indicated that the device was on.

I first tried the farmer method of fixing things, by adding a micro-drop of mineral oil to the switch....wrong, that didn't do anything. I tinkered a little more, and found out the actual switch on the circuit board was busted.

Hit the continue for more on this.

This is my second article about building a DIY digital audio recorder. Inspired by this microcontroller audio project [via], I set out to build a simple digital recording device. I chose Texas Instrument's MSP430 microcontroller for this project because it's fast (16 MHz), it's cheap ($1), and it's very low power. Read the first part here.

This week we'll progress towards a digital audio recorder by generating simple tones with the MSP430 microcontroller. We'll use the MSP430's pulse-width modulator to generate an audio waveform, and clean it up with a simple low-pass filter. The signal won't be strong enough to drive a speaker directly, but it'll work great with a cheap set of powered PC speakers.

Next week we'll expand on this basic circuit to play audio recordings from SD memory cards.

Read on to learn more about generating audio with a microcontroller.

Microcontroller audio

There some great resources on the web to get us started with microcontroller audio.

This series of articles on Arduino audio (part 1, part 2, DAC options) gives a fantastic introduction to the theory side. There's tons of great stuff in these pages that I won't duplicate here.

Another project feeds audio samples into the microcontroller from a PC serial port. The microcontroller is like a simple PC sound card, it's not capable of independent operation. This project, and the great video, inspired me to design this digital audio recorder.

TI and ATMEL have application notes detailing designs for simple digital audio recorders. TI's design[pdf] records 12 seconds of low quality audio to the flash program memory of an MSP430. It uses a specialized chip to create the audio signal, complicating the design. A great TI app note on speech compression[pdf] also has some interesting support circuitry for digital audio. The ATMEL digital sound recorder[pdf] uses a small external memory chip to store a few seconds of audio.

Pulse-width modulated audio synthesis

The cheapest, easiest way to generate audio on a modern mirocontroller is to use a hardware pulse-width modulator (PWM). A PWM is a circuit that generates a repeating time period (called the period), and turns on a switch during a percentage of that time period (called the duty cycle). This happens so fast that only the average value of the on and off periods is measurable. Different audio tones can be generated by varying the duty cycle.

For more about pulse-width modulation, see my previous projects: Make a USB color changing light, and Show PC stats on analog gauges.

This diagram gives an overview of the design.

MSP430

The digital audio recorder will use a MSP430 F2012 or F2013. Both have useful hardware modules, like a pulse-width modulator (PWM) to generate sound, an analog to digital converter (ADC) to capture audio from a microphone, and a hardware communication module (SPI) to communicate with external memory chips. They're also fast (16 MHz), cheap ($1), and very low power.

F2012 and F2013 target boards are available for the ez430 development kit. You can get started on this project with a full programmer, debugger, and development board for $20. You'll still need the extra parts described in this article, but the development kit can make life easier if you're worried about working with surface mount components. The both MSP430 models are also available as a through-hole parts.

If you want to know more about basic MSP430 connections and programming, read last week's primer on the MSP430.

The MSP430 pulse-width modulator

The MSP430 has a 16 bit PWM. This could be used to play 16 bit resolution audio, a vast improvement over most previous 8 bit microcontroller audio projects.

The MSP430 PWM has a ton of modes. I'm sure they're all super useful, but we just want a standard pulse without the fancy stuff. To do this on the MSP430, I set the following registers:

CCTL1 = OUTMOD_6; // Mode 6 is toggle/set
CCR0 = 0xFE; // 16 bit PWM Period, use period-1 for MC_1(up counter)
TACTL = MC_1; // Timer A MC_1 mode counts up to the value in CCR0, resets

The PWM output signal can be enabled on three different pins. This is handy, but it can also be really confusing. I designated pin P2.6 for the audio PWM output. P2.6 is also used for an optional 32.768khz watch crystal. The crystal oscillator circuitry on the pins must be disabled before using the PWM output. This isn't documented very well in the MSP430 datasheets, read more in this forum post.

BCSCTL3|=LFXT1S1+LFXT1S0; //enable PWM output on P2.6, disable oscillator.
//.........................................................//now there are limitations on VLO....

This was a real pain to figure out. I don't usually include these details in an article, but I hope this info will now be easily accessible to anyone facing a similar problem in the future.

Audio hardware

The signal from MSP430 PWM pin isn't quite ready to drive a speaker or amplifier. First, we can clean up the signal a bit by running it through a low pass filter. This clips the sharp edges from the PWM waveform to make it sound a bit more natural.

You can calculate the ideal low pass filter for a given frequency with a calculator like this. I used an 8000Hz frequency because I eventually plan to play and record about 8000 samples-per-second with the digital voice recorder. Note that this is just a fraction of the 48000 samples-per-second used by CD players and PCs. At 8000Hz my ideal low-pass filter has a 0.01uF capacitor (C1) and a2K ohms resistor (R1).



Finally, it's proper form to block any DC voltage from the signal path using an electrolytic capacitor (POL-C1). The capacitor allows only the AC waveform to pass out of the circuit. A value between 4.7uF and 47uF seems to work fine. I didn't notice a difference among the range of values that I tried for the quality of audio produced. This capacitor is usually not necessary, as most amplifiers have a similar capacitor at the audio input connection.

Generating an audio signal in software

Now that the signal is conditioned, we can connect it to an audio amplifier. A cheap set of powered PC speakers work great.

Test program 1 (test1 in the project archive) creates a 50% 'on' signal with the PWM. This generates a continuous tone from the speakers. See it in this short video clip.