Quieter

Clara received the Disney Little People Princess Castle as a Christmas gift. She’s really enjoyed it—she’s the perfect age to understand that the different buttons around it make different sounds. Despite that we have Cinderella, Snow White, and all seven dwarves, she knows that Snow White and Cinderella are the ones who trigger special sounds when pressed on the pedestal. Not only that, but she presses them repeatedly, cycling through spoken slogans to find the little songs, which have come to be her preference.

It’s really adorable.

It was also really loud: the toy being played with in the living room could be clearly heard through the entire house. For ourselves, and other parents in the position of having a noisy electronic toy, a number of options exist:

1. Disable all electronic behaviours by removing batteries (the “nuclear option”).
2. Disable audio by disconnecting speaker (for toys with lights, motion, etc).
3. Muffle audio by adding tape, foam, or some other physical barrier to the speaker.
4. Attenuate the audio by modifying electronic circuit.

Of course, for me, it had to be that last option. Some months ago, I had performed a similar modification on a noisy activity table, so I was pretty confident I’d be following the same basic path. Here’s the bottom of the castle with the screws removed:

Here’s the main circuit board. Whatever the brains of it are, they’re potted under that black blob—the only external components are a few capacitors, so that’s everything in there—the logic, the memory, and the amplifier:

Unfortunately, I had assumed the speaker would be an 8 ohm one, but it was clearly marked as being 32 ohm. Both are standard impedance values for speakers, but I think of 32 as being more typical for headphones. Anyhow, the idea of the modification is to add two extra resistors like so:

The resistance in parallel with the speaker (6.8 ohms) is about 1/5 of the value of the speaker itself. Whatever current flows through the overall system, 4/5ths of it will go through my new bypass resistor, and only 1/5 of it will go through the speaker itself. The series resistance (27 ohms) is there to maintain the overall resistance of about 32 ohms—it’s best not to stress the amplifier with a load drastically different from what it was designed for.

I hadn’t wanted to open up the castle twice, so the resistors I specially ordered for this were assuming it would be an 8 ohm load. When it turned out to be 32, I had to improvise with some junk drawer stuff, which made the whole thing a bit messier than I was hoping for:

The castle now plays its sounds clearly to those in its immediate vicinity, but no longer beyond. Success!

Amplifier Upgrade II: Baby Steps

As I’ve been working on the schematic for my amplifier refurb project, I’ve realised how much stuff I am depending on to “just work”. Hardware and electronics are not like software—a mistake on a printed circuit board means cutting traces and soldering flywires, or even money spent getting a corrected PCB made.

We were on vacation this past week with some other families from our church. I had hoped to have a PCB finished so that I could build it up and be working on software for it. I didn’t get it done in time (not even close), but I did order in a few key parts from Digi-Key anyway, to prove out some stuff I was unsure of, and begin building up the software. I took some breadboard, an Arduino, my laptop, and a soldering iron borrowed from work.

~

The first item I wanted to play with in advance was the IR receiver. I decided to use the TSOP382, a convenient through-hole package which includes the necessary filter and demodulation circuitry onboard. The internal processing circuit gives an output of V_S (the supply voltage) when there’s nothing, and GND (zero voltage) when the 38kHz IR tone from the remote control is detected—very simple and easy for reading from a microcontroller.

The TSOP382 datasheet recommends a simple lowpass filter on the power supply to the device, for ESD protection. So I built up that circuit on a piece of breadboard, and plugged it into the Arduino:

Now, there’s lots of existing code out there for interfacing Arduino with remote controls, but I’ve had mixed results with libraries for Arduino, and I was interested see how far I could get just figuring it out from the cottage we were at—with no internet access, no scope, and not even a multimeter.

Turns out it wasn’t too hard.

I wrote a simple program to capture the rising and falling edges on the INT0 pin, and then set up the high-speed 16-bit timer to time the gaps between them. I then dumped the captured timing information to the PC over the serial port, and plotted it in python using matplotlib. This is the pulse sequence sent by our Sony Blu-Ray player’s remote when the power button is pressed:

Logic low is the tone, logic high is a gap. It looked at first like it might be some weird binary thing, but it’s nothing so devious as that—the gaps are simply pulses, grouped by their digit in the sequence. In the example above, the coded command is 1224212411. Once I had this figured out, I changed my logic to only need half as many interrupts: I fire an initial one on the low level of INT0, and then reconfigure it to fire on rising edges. After a timeout with no more rising edges, I process the received command and restore the original low level triggering to await the next command.

Timing between edges like this may seem more complex than just polling, but it’s very robust to variations in the clock between the remote control and receiver. It also allows the AVR microcontroller to idle in its lowest-possible power consumption mode, as a low signal on INT0 is capable of waking the chip up from that state.

~

Another item I wanted to play with was my selected display. I was tempted to go with a fancy dot-matrix device (like the adafruit negative LCD). Ultimately, though, all I really need is to display the audio source (out of four), and the volume level. The most significant criteria is actually just that the display be legible from across the room. Thus, I ended up selecting a simple seven-segment LED display, the LDD-A812. I will show a number from 0..64 for the volume level, and C1, C2, C3, C4 for which channel is active.

I wanted to try this device in advance for a couple reasons. The big one was brightness. The manufacturer’s figure of 3.9mcd means nothing to me; I want to be able to put it—illuminated—behind the smoked plexiglas that will be the final new front to the amp, and make sure that it’s bright enough to be visible. Second, I wanted to prove to myself that I could PWM the voltage supply to this device and avoid having to use current-limiting resistors. Finally, I want to use a MOSFET as the switching device for the supply to the common anode, and I wanted to check that I wasn’t completely off my rocker about how that was going to work. Here’s the prototype I mocked up:

I haven’t got a picture of it in operation, but I lit up a few segments, and it did indeed seem to work as intended. It would be a lot of unnecessary work to hook up all the segments to the Arduino board, so I probably won’t bother with that, but I was glad to confirm the basic operation.

~

The final item I was curious about was the hardware knob that will provide physical interface to the device as an alternative to the remote control. I’ve pretty much decided on the PEC11-series encoder from Bourns, but I was less sure about what knob I’d use, and whether I wanted the encoder with detents or smooth operation.

All but a very few of the knobs on Digi-key and Mouser are oriented toward use with potentiometers—that is, they have a position indicator on them, which makes no sense when the position is displayed on the screen, and the rotational input is purely relative. They’re also really expensive.

In the picture is a heavy steel knob I ordered in from McMaster-Carr (on a potentiometer I had lying around from a previous project), and a knock-off Telecaster guitar knob from an sketchy eBay vendor (on the PEC11 encoder I ordered). I still haven’t fully decided which knob I like best, but I think I would like the encoder without the detents—if it’s anything like the pot, it will feel more like a weighty hifi device, and less like a cheap car stereo.