Formula relating R/C to Clock Freq in BBDs ?

[Yuan Han]

Looking at the datasheet for MN3102, I see a graph relating the Capacitor/Resistor used to the frequency of the clock rate generated.
Is there any formula for this ?


Thanks

[Yuan Han]

*bump*

[Mike Burgundy]

Sheet:
http://www.cbtricks.com/database/datasheets/mn3102.pdf

Thinking out loud here, bear with me...
um, the sheet says that oscillation F is defined by the RC constant. Any odd number of inverters feeding back through a cap like that will oscillate - the question is *when* it switches (never just at "0" or "1", never exactly halfway either). Think of it like this: bridge the cap for a moment. Now, at start up (apply power to the chip) all ins and outs are at 0. The inverter immediately changes the output (OX3) to a "1" which should be more or less equal to the positive rail.
But hey, wait a minute! Now the inverter sees a "1" at its input, so it will flip it's output to "0". Hot dang! The input changed to 0! We're feeding it's output directly back into it's input, so the inverter will oscillate as fast as it can (there's always a minute amount of time needed to switch from state to state). Got that?
Now, introduce a resistor and cap: the whole business is *slowed down* by the time it takes the cap to charge. This depends on the caps value (larger caps need more juice to charge and so slow things down) and on the resistors value (larger values allow less juice to pass so  they slow things down as well).
This brings us back to the RC time constant (Ohms times Farads equals seconds, really!).
It suggests a ,linear relation: doubling R (OR C) will double the time constant - or lower the frequency by half. This is not quite the case according to the graphs, which I think is due to the inverters limitations. Anyone? Am I right in this? Digital isn't really my suit
It will get you in a pretty good ballpark though (esp. considering the awful tolerances of caps) - 1M and 400p should get you near 1kHz, 2M and 100p should do something like 3kHz.
hih

[Yuan Han]

heyo,
thanks for the explaination.
I was trying something like f = 1/2piRC or some related formula, but when i subbed in the values from the curve that they gave, it didn't really give the results as shown.

I guess the situation's the same with a 4049 phase locked loop ?
Which brings me to the point, that if er no "analytical" formula is available, doesn't that make the use of the chips limited ?

yh