To answer the last question first: I'm not sure what video you saw of the "Singing Balloon". The sound level output of the balloon is satisfactory; I wanted the higher voltage to get more headroom. In order to get the volume you hear I had to severely clip the (original) waveform. With a higher voltage (more headroom) I could have clipped the sound less and it would have sounded much (much) better. As with any disposable product, sub-fractions of a penny add up fast.

I didn't listen to the videos you attached but I think I get it; the card is programmed with DTMF tones (http://en.wikipedia.org/wiki/Dual-tone_multi-frequency) that are recorded (or generated via code) and reproduced into the mouthpiece of the phone. Phone systems recognize this and dial the number. Lot's of products have been on the shelves over the years that do this. The business card is cute but I'm not so sure of the practicality (my opinion - but we're not talking about that). You didn't say if this is exactly what you want to do. If it is, I'd offer the following...

The links I posted show how to connect the inductor for operation. I don't know that this would help anyway. When using an inductor in an oscillator you are tuning the components for the target frequency. With DTMF you are generating a couple of different frequencies so (if) you get it to oscillate it will not be very efficient and the voltage generated will be all over the place. Frankly, I'd abandon that approach.

You *could* however design a small high voltage step-up circuit. I'm going straight from memory but I think both Maxim and Linear Technology make IC's to do this. I have no idea of the Vmin values... You could then use this with a transistor to power just the piezo. I think these devices are current hungry and you won't get far with a lithium coin cell (they probably have an enable of some sort to turn them on and off, but...). And then there is the noise on the supply rail to deal with...

However.

You don't need high volumes to make this work, the XOR'd output from two pins driving the piezo should be fine; with 3V you get 6Vp-p like this (5 gets you 10). You're not going for resonant frequency so you can pretty much ignore anything to do with that other than holding the piezo solidly.

Keep in mind that waveform is not going to be a perfect square-wave when driving the piezo, nor will you get all the way to the rails.

I'd look for the piezo devices that have a crystal element on both sides of the brass coin. These are more efficient than than the single sided and only cost a few pennies more.

The thing you need to keep in mind doing this is the stability and accuracy of the frequencies being generated and that the tones are sine wave, not square. As I recall from doing a DTMF decoder years ago was that you have a very small window of deviation from the center tones to make it work.

PBP has a DTMFout routine that needs a 20Mhz to get it right. It's PWM from one pin and needs a filter to get it "sine-wavy". You'll loose voltage through the filter so don't think it'll be anywhere near the rail (its meant for going to into an amp). I've seen assembly code years ago to generate DTMF and used a few but that was a long time ago.

Good luck.

P.S. When preparing a lengthy post, before posting it copy it to the clipboard. If the session has timed out you can just paste it back in.