Square Wave 80 - 120 MHz with adjustable duty cycle

Hello,

I need to design a signal that is a square wave between 80-120 MHz that can have its duty cycle vary from 10-90%.

Initially, I was going with the LTC6905 to generate a 50% duty cycle, however, adjusting the duty cycle in a subsequent stage proved to be harder than just generating the appropriate waveform. I entertained the idea of using a 5th order chebyshev LP filter to get a sine wave and use a certain trigger level to set the duty cycle, but I can't find comparator IC's that work at such high frequencies.

I would like to adjust the frequency continuously (i.e. with a resistor) and have relatively coarse duty cycle steps (10% increments). I would also be fine with having discrete steps from a PLL. However, I have never used a PLL with a duty cycle other than 50%.

If anyone has some tips, or a suggestion for IC's that I can look into, it would be highly appreciated. Thank you for your time.

Kind Regards,
-Michael

Here's a trick: LVDS receivers are basically comparators. And they're very fast. So you might be able to use the comparator approach.

But, there definitely ARE high speed comparators available. They're more expensive, but more accurate. Look at linear technology, for example.

10 % duty cycle at 100 MHz means 1 ns pulse width and respectively shorter rise and fall times. Definitely beyond CMOS or TTL speed range. What's the intended signal level and output impedance?

LVDS or ECL are suitable IO-standards.

At those frequencies, having a perfect square wave is very difficult.
As a fact if you have ever seen an LVDS signals, you will see that they are trapezoidal.

At those frequencies board parasitics and transmission line effects also play a significant role.

schmitt trigger said:

At those frequencies, having a perfect square wave is very difficult.
As a fact if you have ever seen an LVDS signals, you will see that they are trapezoidal.

At those frequencies board parasitics and transmission line effects also play a significant role.

Click to expand...

Yes, but with proper design he can get an imperfect square wave. The output rise/fall time of a 55LVDS32, for example, is about 0.6nS. I guess it depends on what the OP's end goal is...