This has to be one of the more interesting hacks I've seen

An A/D made from a 555 timer:

http://www.bobblick.com/techref/projects/a2d555/a2d555.html

I found this in the course of looking for stuff related to DIY digital data acquisition. I don't know if it's useful, but it's interesting.

This is one of MY favorite hacks-

That is not me on the left. . .
Mike

It isn't A/D alone. It converts voltage to pulse-width. Time is still an analog quality.

Time happens to be something that CPUs can digitize pretty well.

The real hasty-hack "D/A" was the joystick port on the IBM PC 5150. IIRC, it put a charge on a cap and let the joystick pot discharge it. Cruder even than a 555, but effective for the purpose.

Mike: the image you linked, the server objects to delivering it within a page on another site (referrer URL). The way around that is to snag the actual file and host it elsewhere.

Ah, I think I see now. I'm still curious why it can't measure a voltage lower than the rail voltage, but I'm sure there's a good reason for it. Like I said, I was searching around for A/Ds for general purpose data acquisition for a lab, something that I could plug a variety of sensors into. This is one of the cheaper options, if you don't need too high a sample rate:

http://www.mtmscientific.com/atod.html

Simple and cheap, and it's being used by radio astronomers, so it works. The main disadvantage it you have to re-flash the chip with new software if you need a new data collection rate. There are ways around that, though. In the end, though, I'll probably end up with one of those 12-bit 1.8Vpp converters from TI or some such maker that are made for scientific applications.

As PRR said, this is basically a gated clock and a counter. The limitation is in the 555 voltage to interval convertor.

If you want to get a little more clever, I did a variation on this (clock/counter) for a 100 segment Peak/VU meter back in the '70s, Since this was before microprocessors were so common, I used a TTL decade counter to get my tens and units output, to drive my display. Then I gated the clock feeding the counter with a time interval generated by the exponential decay of a simple RxC  (e^-t/rc) to perform my log conversion. I won't swear to the tracking accuracy down 100 dB from full scale, but it was fine for audio metering and who is going to quibble about down around -80 VU for line metering?

JR