Choosing the right frequency dividers for pulsatile inputs at 80 and 160 MHz

Greetings,

I have two pulsatile signals whose frequency I wish to divide:

A. 8x frequency division of a pulsatile 80 MHz input. As exemplified in the following oscilloscope screenshots, its waveform is somewhat irregular, with a primary ~3 dBm peak followed by a secondary peak that can be up to -4.5 dBm high, in its average power (all measured to 50 ohm load):

B. 16x frequency division of a pulsatile 160 MHz input. As exemplified in the following oscilloscope screenshot, its waveform is clean but rather weak (-19.8 dBm in its average power to 50 ohm load, if I've calculated correctly):

In both cases, the optimal output would be a ~10 MHz sinusoidal, or clipped-sinusoidal wave, at ~3.3 Vpp to a DC-blocked 50 ohm load. Other periodic ~10 MHz output signals may also be acceptable, if they satisfy VOH of at-least 2 Volts and VOL of at-most 0.4 Volts, to a DC-blocked 50 ohm load.

I've browsed through the wide selection of available prescalers, but seemingly at these frequencies they all expect a strong square-wave input, and output a weak square wave. Can you please suggest which evaluation board of yours, or a cascade of evaluation boards, is most likely to be able to pick up these signals correctly and produce the required output?

Many thanks,

Lior

Parents
  • 0
    •  Analog Employees 
    on May 4, 2017 5:41 PM over 3 years ago

    Hi Lior,

    Dividers are essentially "dumb" devices - that is, they take whatever is provided at the input, and if the signal is of sufficient amplitude with a reasonable slew rate and falls within the operating frequency range, it will try to divide it. In Case "A" above both the peaks will be divided and passed to the output of the divider. To prevent this the signal needs to be cleaned up.  This can easily be accomplished by the addition of a non-inverting comparator circuit. This will not only eliminate the secondary peak but will also have the added benefit of providing a sharp clocking edge for the divider input. The divider IO's are CML like and Analog devices has many comparators to choose from that may simplify this task. Since you'll need a divider for the /16 case as well I would recommend the HMC394 (or HMC705) as this may help you achieve some economy of scale.

    With respect to the divide-by-16 case the HMC394 would be a good choice (alternatively HMC705) and while the signal itself appears to be o.k.  the amplitude is too low. These microwave frequency dividers require higher drive levels at these frequencies below 500MHz as well as when the slew rate decreases so some amplification of the signal will be required to get up to ~0dBm or so.  

    Best Regards,

    Marty

Reply
  • 0
    •  Analog Employees 
    on May 4, 2017 5:41 PM over 3 years ago

    Hi Lior,

    Dividers are essentially "dumb" devices - that is, they take whatever is provided at the input, and if the signal is of sufficient amplitude with a reasonable slew rate and falls within the operating frequency range, it will try to divide it. In Case "A" above both the peaks will be divided and passed to the output of the divider. To prevent this the signal needs to be cleaned up.  This can easily be accomplished by the addition of a non-inverting comparator circuit. This will not only eliminate the secondary peak but will also have the added benefit of providing a sharp clocking edge for the divider input. The divider IO's are CML like and Analog devices has many comparators to choose from that may simplify this task. Since you'll need a divider for the /16 case as well I would recommend the HMC394 (or HMC705) as this may help you achieve some economy of scale.

    With respect to the divide-by-16 case the HMC394 would be a good choice (alternatively HMC705) and while the signal itself appears to be o.k.  the amplitude is too low. These microwave frequency dividers require higher drive levels at these frequencies below 500MHz as well as when the slew rate decreases so some amplification of the signal will be required to get up to ~0dBm or so.  

    Best Regards,

    Marty

Children
No Data