Audio Output from Spartan 3 Starter Kit

The main danger is that mechanical shock to the speaker can produce a high voltage at its terminals (even where your equipment is turned off). This can damage a semiconductor device.

Does anyone here know the simplest way to avoid this? I'd guess that a 5V zener diode across the speaker should work, but maybe there's a better way.

(Some semiconductor devices have internal "clamp" diodes to their power rails and this provides some protection, but an FPGA intended for use in a "hot insertion" product probably doesn't have these diodes).

The (electrical) impedance of the speaker depends on frequency: there will be (mechanical) resonances at certain frequencies. Generally, you can expect the impedance to be less when the sound is louder. However, for a small speaker such as you might use for experiments, even the minimum impedance is unlikely to overload an FPGA output.

Mike

FPGAs do indeed have the protection diodes. You can drive a piezo directly from the FPGA if you like, however you'll probably find the audio volume less than satisfactory with the typical miniature transducers, especially with the low voltage I/O standards found on most of the modern FPGAs. It helps if you drive it in a differential mode.

I am guided by "Cyclone II Device Handbook, Volume 1" (Altera Corporation, July 2005), chapter 4 ("Hot Socketing & Power-On Reset"), page 4-1:

"There are no internal current paths from I/O pins to VCCIO or VCCINT power supplies. Signals driven in on I/O pins do not power the VCCIO or VCCINT power buses".

I understood this to indicate that there are no protection diodes in this case (except perhaps to 0V), but you indicate that there are always such diodes. Have I misunderstood this?

Mike

It's the old conflict between ESD (electro-static discharge) protection (which wants a strong diode against a well-defined voltage, like VCC) and "Hot Socketing" which cannot tolerate a diode connected to Vcc (since that would be against ground while the card is being pushed in). I cannot speak for Altera, but a traditional compromisen is a diode against a positive Zener diode level. If I were you, I would get a positive supply, a 10 kilohm series resistor, and probe the pin to see how high the voltage will go. You cannot do any harm with less than a mA, but it will clearly show you the clamp voltage. A good old multimeter is a wonderful thing, be it analog or digital :-) A scope will do also, especially when it sits already on your bench. Peter Alfke