Hey Geezer! Yep- I got the picture and info saved from the other site- thanks gobs!
KENtucky... yep- I've got a box of BEI (and other) encoders here, and I've made a few Absolute Position Encoders (APEs) for special projects.
In my stuff, I really don't go through the trouble of employing an encoder with precision beyond what I need... for example, if I was actually calculating the crankshaft angle for spark or fuel management, I'd wanna know, if not the entire crank revolution, at least when I'm within say... 50 degrees of TDC, up to about 10 degrees after... but with just wanting RPM, a single count-per-turn would be fine... or a count for every other turn... that's what an ignition-operated tachometer does.
The issue of counting teeth, etc., from the encoder aspect really isn't a major concern- it's really the processor's ability to count. A PIC only works so fast, and you can't easily count teeth and both measure voltages and do the ADC count processing at the same time... so like Relativity, you can't know both things in synchronicity. By using a 1-event-per-revolution device, it cuts down on the amount of counting operations (and resultant wait-states) that the processor has to digest before doing a voltage-read operation, and then an ADC count. The PIC's ADC system is really kinda hokey in some ways- it's got an internal RC circuit, and it measures Analog by taking a 'sample', and counting the number of operations it takes to see the voltage fall by 63.2%, and then applies that to T=RC. In my telemetry processors, I don't run into this time-intensity for processing demand, so it allows me to run an on-board oscillator clock at very slow speed (like, in the 100's of Khz), and my software guru and I developed a slick little GPIO interface protocol that allows a 500% variation tolerance in both the PIC clock and host processor (an AMD Elan @ 133mhz) speed without having any noticeable effect. Tracking the counts of a tone wheel would be really tough, unless you coupled the tone directly to a sound-card and used the sound-card's DSP to do a simple frequency analysis... you could probably calculate primary domain with FFT and make the math really short, but most guys don't want to bury their heads in DSP algorithms, or worse yet, figure out how to actually 'load' the algorithms into the sound-card, since most drivers don't take kindly to being given the boot.
But send me a copy of that tach circuit- it's entirely possible that I could derange it to operate on a 16F876, or possibly use three simpler PICs (one for tach, one for voltage/ADC, and one for RS232 or even Ethernet commo) on a cheap board...
As for finding supplies to make your own, you might wanna try taking a test-step into the 21st century... go to
www.ExpressPCB.com, download a copy of their software and play with it... it's got a built in price quoter, and double-sided boards with plated vias is standard... great quality, all automated ordering, you can get fast service 'n stuff, and it's drop-dead gorgeous...