[time-nuts] Zero-Crossing Detector Design?

[Magnus Danielson]

On 07/20/2012 12:33 AM, lists at lazygranch.com wrote:
> Are you speaking of slew rate limiting in the strict sense of the word, that is a current starved input stage due to the presence of a compensation cap? Or are you using the term slew more vaguely.

I am speaking neither.

If you have a sine of a particular frequency and amplitude, then you 
have a known slew-rate, it peaks at 2*pi*f*A, where A is the amplitude 
of the sine. As you amplify this signal, the slew-rate will grow 
proportionally. Recall that the jitter of a trigger point is noise 
divided by slew-rate. This is why we want to increase the slew-rate to a 
maximum while adding minimal noise.

Now, as the amplifiers has a gain, to increase the slew-rate by say 5 
times, the bandwidth of the amplifier needs to be high enough to support 
this, but in order to minimize the added noise, we want to keep the 
bandwidth down. This may be best realized by also recalling that it is 
the wideband noise at the trigger points which this first-degree 
analysis depends on, and the RMS level. A 1 Hz amplifier bandwidth is 
nice, but it won't support a high slew-rate...

In a two amp setup the later amp will have a higher bandwidth, but the 
noise added of the first amp will also be gained up, so a tigther 
bandwidth there will keep its contribution lower. You end up with having 
high benefit for low noise amps in the beginning, but as you gain 
slew-rate the amplifier slew-rate capability becomes more important over 
it's noise properties. It's being balanced by amplifier feedback terms 
for both gain and bandwidth. Also, diode limiters will maintain the 
output as clipped sine, so we can continue to gain the output for slopes.

Cheers,
Magnus