Clock/Counter Card

November 25, 2006 4:04 PM

Timer and Control Circuit. I need to measure the rotation of a spindle and the time during which a circuit is thrown open during that rotation. I think I need a clock, a counter and a software interface. Where do I start?

My thinking is towards a PCI card at low cost - to put in a cheap PC, which would allow very accurate time measurement by providing a clock pulse which would be run through a number of circuits, circuits which would simply open and close depending on the rotation of a spindle. I'd like to be able to take the results and use Visual Basic or Java to display the results in real-time. I think I need something like this but before I sink $400, any points to look for, things to avoid?

posted by grahamwell to technology (4 comments total)

What do you mean by very accurate ?

There are plenty of USB solutions in the USD$100 range.
posted by b1tr0t at 4:16 PM on November 25, 2006

A PCI card is one approach. I kind of like Keithley hardware and tools.... http://www.keithley.com/ They acquired Metrabyte years ago and their hardware is kind of industry standard. I've been using USB dongle type pods with combined analog IO, digital IO and some counter/timer capability. (These things sort of come together frequently because they usually have IC's like the old 8254 programmable interval timer chip which has a few bytes of quasi-bidirectional IO.

A UK card based on this chip for about 40 pounds :

http://www.decisioncards.com/io/index.php?style=lab8254


Key questions to ask about your application... how fast are the signals and what precision requirements. The 8254 stuff usually has a 10 MHz clock limit, meaning it could be used to get resolution of about .1 microsecond.

The architecture of your solution will dictate the how much confidence you can have in your measurements. If you plan to use a polling approach to measure intervals, you'll run into a lot of factors that can screw with repeatability of the measurements.

My approach to this kind of stuff is to observe the signals and characterize them with a scope first. Failing that, do a thorough analysis of the timing involved theoretically, then conceive a measurement approach, THEN go out and find hardware to support your concept. What you are doing sounds pretty garden variety, though, and there are scads of approaches available cheaply, especially if you have slower signals and modest accuracy requirements.

Good luck.
posted by FauxScot at 8:22 PM on November 25, 2006

Thanks. My question could have been more precise. I think 10MHz will be more than enough. I was thinking of speeds of more like 50Khz. I'll go explore and specify some more.
posted by grahamwell at 1:23 AM on November 26, 2006

"A few hundred thousand" is my brief - so at the upper end say 1MHz would do nicely. My concern really is cost and the programming interface. I'll explore the suggestions above.
posted by grahamwell at 11:24 AM on November 26, 2006