Multiplying Frequency

Frequency multipliers and dividers require common (integer) denominators or harmonic relations.

At such low frequencies it would be more appropriate to measure the period of each cycle with a high resolution counter, do the math, and generate an output with another timer. Lots of little micros could do this easily.

BTW, analog multipliers are **not** the way to go! A possible solution with discretes is to use a PLL to get a **much** higher frequency that can be divided by a counter. Getting a PLL to stay locked onto a moving signal with a large frequency difference is a lot more difficult than it sounds.....the uC solution will certinaly be simpler!

As with so many of these sorts of problems, the best approach is probably a small micro. As an example, a PIC 12F675 is in an 8-pin DIP, and requires only 5V supply. Doesn't even need a crystal, just a cap across the supply. Of course you have to program it.

You'd be better of changing the tooth wheel if that is possible.

Regards,

Boris Mohar

Got Knock? - see: Viatrack Printed Circuit Designs (among other things)

void _-void-_ in the obvious place

But unless you are talking high volume or need real small size, going as small as the 12F675 is probably an exaggeration! Being small can make life difficult ;-)

Why ? He requires only one input and one output pin, so in this case there are even too many IOs :) Frequency precision is not an issue either since he only needs that the ratio between input and output is correct.

vic

I didnt even think about using a micro, how hard would be it to code? I have only taking one class on C AVR programing and it wasn't one of my favorites. Thanks for the replys guys. I dont mind a complex analog solution.

OK, so he gets chips and programmer for a 12F, and then next project he needs a UART.

I always have some 16F876/7's hanging around (28/40) and they do for all my little jobs ;-)

Coding in assembly for a PIC or 8051 is very easy. You just type in the instructions one above the other. Designing the software usually takes longer than coding it.

reciprocal counting, table lookup.

variables: current_count, current_output

initialize: jam something reasonable into current_output. start the output task going, flipping the output line when the output timer hits zero.

two tasks: one foreground watching the input, one generating the output (probably interrupt driven)

the input watching task loops forever, counting periods between edges of the input, save that as current_input.

take current_input and convert that (calculation, or table lookup) to the output count, save that as current_output.

the output task, probably interrupt driven, a hardware timer with a timer zero interrupt.

load the timer with current_output. When you get the timer zero interrupt, flip the state of the output line and load the timer with current_output again.

Doing it this way, current_output lags current_input by at least a measurement cycle.

Left as an exercise to the reader is what to do with whacko (jittery) readings: instead of jamming in a new value for current_output, you could average the old with the new. Or, you could window the value and limit the change per cycle to a predetermined amount. The idea is to smooth noisy data, and do that in a single-chip micro without a multiply instruction. (Add and shift right to average, but watch out for overflow). You can also smooth/window on the input side.

The idea is to try and match the physics of the system; how fast can the speed change? Well, ignoring events such as a thrown rod or a siezed bearing, in which case weird data from this gadget is the least of your problems.

Just make sure failure modes of *this* gadget are benign!

And the problem with a 4046 Phase Lock Loop is....?

low frequency? i dont see how a low divide by factor would work.

No problem at all to use a 4046 PLL. It'll just take more hardware than a micro.

If you're setup to work with and are familiar with some kind of micro that would probably be the simplest way to go, but to learn a new assembly language, although a good thing to do would, if you're like me, take a lot longer than getting the 4046 to work. This is especially true of learning the ins and outs of programming interupts.

Btw, multiplying 36.5Hz up to 43Hz is a factor of 1.17808 not 1.15, but maybe 1.15 is close enough? 1.18 would be quite a bit closer. Anyway, use the 4046 to multiply the freq by 118, then divide that by

Mike

------------------------------------------------------------------------ The odds of a single modest length protein randomly forming is approx

A fairly trivial program for a 8 pin micro.

So tell us already.

Personally I'd use gears... as someone already calculated, 139 teeth to 118 teeth ;-)

...Jim Thompson

With gears there's no averaging... the ultimate in PLL's ;-)

...Jim Thompson

Using a micro, I'd measure the period, average it a bit, and then multiply that number by 0.8488372/2 ** in fixed-point math and use that to precisely time each half of the output period. The clock is non-critical so it could be on-chip RC.

Best regards, Spehro Pefhany

Sure there is. It's called "momentum".