[time-nuts] Zero-Crossing Detector Design?

ct1dmk ct1dmk at gmail.com
Mon Jul 23 11:25:42 UTC 2012


For the specific application of driving an FPGA clock pin (that has an
enormous input bandwidth) many things can go wrong.
All fine about the advantages and disadvantages of
the gate with resistor feedback, all I can say is that over here I it 
was not the best solution
we found over many FPGA board versions with external clock.
The latest design and the one we keep using standard now, uses a 
differential pair of PNP
transistors (BFT91), this will have a moderate gain of x10 or so and the 
resistors set
for making a limited wave of 0 to 3.0V (simple change of resistors can 
make it 0-2.0V
or else required). we use a 100MHz narrow band limited bandwidth sin 
signal so no filter
added, but we could add if we need one.
The output signal into the FPGA looks very clean and has a few ns rise 
and fall times
(not super steep, but the fpga input does the rest. It does depend also 
on the resistor values and current used in the transistor pair).
This was the way we could get the very minimum clock jitter in the FPGA 
and a
simple circuit quite tolerant to input levels and make a very clean and 
well defined signal
into the FPGA. Way better than 74F or 74LV gates etc.
The only inconvenience is that it needs +5V for the circuit to work (the 
VCCIO of +3.3V is not enough).

My 2 euro cents ;-)

Luis Cupido
ct1dmk.


On 7/22/2012 9:32 PM, Bill Fuqua wrote:
> Wow, I have not checked this list for some time. But there is a lot said
> about zero crossing detectors.
> Lots and lots of replies, so many that I have not looked at all of them.
> 1. Do not use CMOS inverters. Even though so much has been published on
> using these in linear mode by
> adding a feedback resistor, they can be a nightmare. The fast ones
> (74HC, 74AC, etc) have so much high frequency gain they are
> likely to take off into oscillation on their own.
> 2. The first thing you can do to get a good clean zero crossing is to
> reduce the noise. This means to pass it
> thru a narrow band pass filter such as a crystal filter. The narrower
> this filter is the closer to a pure sinewave it becomes
> and the less noise you have.
> 3. In research when we want a precise trigger we use what is called a
> constant fraction discriminator.
> This may not be needed if you have a very clean signal and its amplitude
> does not vary and you are wanting to
> trigger exactly at zero. But a constant fraction discriminator triggers
> on a point that is a constant fraction of the
> amplitude of the signal. They require a delay so that a fraction of the
> peak of the cycle can be compared with the rising edge
> of that cycle. This is mostly used with triggering on pulses of varying
> heights and when subnanosecond
> timing is required.
>
> My suggestion is to clean up your signal as much as possible and reduce
> noise bandwidth using a bandpass filter and
> then use a low noise amplifier for the front end of your zero-crossing
> detector.
>
> 73
> Bill wa4lav
>
>
>
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to
> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
>



More information about the time-nuts mailing list