Pushing the boundrys of the datasheet specification for a UPB1507GV

The datasheet for the UPB1507GV device says max 3.0GHz and min 0.5GHz operating frequency but I was hoping someone here may know if the 0.5GHz lower limit is fixed. Ideally I would like to periodically/temporarily drive it from the arduino or if need be from something like a 100/200MHz clock. The idea is it's driven from a 3GHz oscillator and on an event start I use an arduino to count the divide by 256 output until an event end. Then I freeze the oscillator and drive the UPB1507GV oscillator input from the arduino so I can count how many more clock pulses are needed to fire the next divide by 256 output. This will hopefully give me a >400ps accuracy time interval .

The block diagram shown in the datasheet doesn't show anything preventing low frequency operation (or preventing "zero frequency" operation). It just shows flip-flops connected as a frequency divider. So, I'd give it a try! Do you have access to an oscilloscope and function generator (or some kind of variable pulse generator)?

...I'm really not following what you are trying to do. I think you will still be limited by the "resolution" of the Arduino (and your firmware). You might need a stand-alone hardware-counter to count how long (the number of pulses) your event takes (conceptually like a stopwatch). Then you can read the (stopped) counter with the Arduino whenever it's convenient.

I've used a counter chip before, but the one I used only counted to 16. I assume you can find a chip that counts higher, or they can be cascaded. A counter is essentially the same as a frequency divider except you have access to the individual flip-flop outputs so you can read it's "state" (a binary count) at any time.

P.S. OK, I get it! Yes! Your idea should work as long as you can freeze & count slowly. (And, it's a lot less wiring & fewer I/O ports than using a counter chip.)

DVDdoug: The block diagram shown in the datasheet doesn't show anything preventing low frequency operation (or preventing "zero frequency" operation). It just shows flip-flops connected as a frequency divider. So, I'd give it a try! Do you have access to an oscilloscope and function generator (or some kind of variable pulse generator)? Thanks, I will order one from somewhere an try it. I have access to better equipment at work and can turn an arduino into pulse generator (If arduino cannot drive it then the problem is not easily solvable).

...I'm really not following what you are trying to do. I think you will still be limited by the "resolution" of the Arduino (and your firmware). You might need a stand-alone hardware-counter to count how long (the number of pulses) your event takes (conceptually like a stopwatch). Then you can read the (stopped) counter with the Arduino whenever it's convenient. I would use something like 2 high speed comparators to start/stop the clock. The best I can find are about 5ns but this is a constant I should be able to calculate away as the start/stop delays would match.

I've used a counter chip before, but the one I used only counted to 16. I assume you can find a chip that counts higher, or they can be cascaded. A counter is essentially the same as a frequency divider except you have access to the individual flip-flop outputs so you can read it's "state" (a binary count) at any time. The problem I'm having is finding anything to run faster than about 200MHz.

I looked at the data sheet for the UPB1507GV http://www.cel.com/pdf/datasheets/upb1506.pdf

The low frequency operation is not a good operation. The D flip flops may use "dynamic logic". That type of logic uses low power, but its capacitive storage leaks off at low frequencies. That leakage is reduced at low temperatures. You can cool the device to -40 C for lowest frequency operation. T

If dynamic logic is used, you will not be able to stop the clock. It will fail to hold the capacitive charges if the clock stops. The NAND gates and other gates inside the chip have diodes connected to the storage capacitors, so diode leakage current will cause droop and eventual loss of state information.

You can try liquid nitrogen cooling at 77 Kelvin for low leakage. ( -196 Celcius). But that lowest temperature may cause other failures in circuits, like threshold shifts and cracking of the silicon metallurgical interface at the package.

Conclusion At -40C, I expect you can make it work at 200 MHz.

I will get one to play with but I’m not holding out a lot of hope now that it will work.