frequency devider

divide by two should be the spectrum of a square wave, Dividey by n>2 should be a pulse train. Like the first two diagrams shown here:
http://www.dspguide.com/ch13/4.htm

Yes, i know... In theory output spectrum mast have only 3,5,7... etc harmonics.
But this only theory... I think that real spectrum must have 2,4nd harmonics at levels at least -60dB. But may be i'm mistakes.

P.s. sorry for bad english.

Two possible sources of even harmonics exist:
-Crosstalk from input and device internal signals
-Output waveform unsymmetry

Cause both aren't specifies for the said device, you can't rely on a certain even harmonics suppression, to my opinion. Furthermore, a frequency divider is basically a digital device without specified analog signal properties. Filtering should be considered to use the output as an analog signal source.

If you have perfect 50% on and 50% off square wave with zero rise time, you would have no even harmonics. If you had 49.9% On and 50.1% off time, you woulld have even harmonics, as the equations predict.

Added after 2 hours 8 minutes:

HMC363 divide by 8, 800 MHz in,

Outputs:

-6 dBm @ 100 MHz
-64 @ 200
-15.7 @ 300
-68 @ 400
-20.2 @ 500
-68 @ 600
-21 @ 700
-36 @ 800

Rich

Thank you! This is what i'm need :)