Positive sine wave to square spikes

Hi,

I want to design a circuit that can take positive sine waves as input and outputs spikes for individual pulses. I have attached a picture in which train of pulses below are my inputs and I want output exactly shown by above spikes.

Actually, I want to count events distinctively and each positive sine-like hump is one event. The frequency of inout pulses is very high (in MHz).

I hope I am clear. Thanks and waiting for your help.

Hi,
Take a comparator with some low thresold at it will do it for you!
How wide can be the output pluse, how much are the rise/tail times & amplitude range pls?
K.

The input signal's magnitude is variable and also I have to cater for the noise so I can not use simple thresholding. Also, output wave should be very narrow. As input frequency is in MHz so output pulse should be less than 1us (preferably in ns).

Input signal varies from 0 to 5 volts. Output can be 5 v spikes. Rise/fall time of the output should be as minimum as possible.

Thanks

Did you heard over "constant fraction" discriminators pls? That is what you need...
K.

Added after 35 seconds:

I asked for the times at the input pulses pls, but I wrote it falsh_sorry!...

No. Can you please provide me some information about it.

The time period of one input pulse is about 1us. The shape or time period can change so circuit should handle this and differentiate every event distinctively.

I'd take a comparator and (depending on "how square
is square") either just differentiate the edge, or feed it to a
one-shot (e.g. 74xx123).

Its so, if signal amplitude is strong variant i.e. with 40dB dynamic, and you does compare for ex. to 10-20 mV, you will become a different beginning of your output pulse, depending of amplitude. In the pulsetechnique is these effect called "walking"-mosten dont needed for precisous signal processing.
CFD (ConstFractDiscr.) can it so minimise that even for these dynamic, and rise time variancies from input pulses self, it will have only a pp value of 1-200psec (or similar)...
K.

Added after 4 minutes:

Whats your needed "doublepulse"resolution?
@dick_freebird,
A CFD has Comparator(s), maybe wit some compensatings amps too, and the output is usual a latched FF, with hes feedback can be a monostable builded and dont need some extra components + is more precise as a "123"...

You got a much better answer, than your vague specification deserves. To decide about an adequate circuit effort, you may want to achiece clarity about the actual requirements. Of course you can get CFD and pulse measurement equipment "as fast as possible". But I'm rather sure that you don't need it, fortunately.

Hallo Frank,
You can build CFD slower as 200MHz, me was interesting to tell hem what principial hes (for me clear CFD case) pulse processing needs...
If the dynamic is at 40dB, repetition rate is at MHz and their are tail pulses (its clear so from the picture), than you should take a CFD... Other question is if must or not be DC compensated as Bijalkowskys article describs.
I worked over 20 years in these metier_know relative good the subject and the detailed specs from Bijalkowskys solution too.
You can find relative good & simple realisations with normal mans comparators, I think EDN have had an Des. Idea some years ago, but the circuits basics are good presented in the attached articles, an eng. can build hes version after studying these stuffs_ otherwise is much script form web to find.
If CFD is needed or not tells us the answer: how much walking can the system accept_so simple is it
K.

I appreciate your profound answer. I had some contact with CFD or "constant-fraction-trigger" circuits, as they have been
called in my enviroment during my institute years. I also think of these circuits, when a trigger circuit with variable threshold
is required.

Refering to the original post, some basic CFD properties, particularly amplitude independant trigger delay, have not been requested.
The vague specification makes it difficult to decide which trigger circuit properties are actually requested. It reads more like a wish
list. Some points, e.g. noise sensitivity, dynamic and speed usually involve a well-considered tradeoff that can't be seriously
answered with slogans like "as fast as possible" (which most likely implies "more noisy than you believe" ).

Tnx Frank...
OK, we should wait! ...
Btw; I found the cited Design Idea, but its not so direct for these application, but developed more direct for slower rise time pulses (some usec)...
Because its an interesting idea/circuit technique_I do attach it.

Yes, the problem of time walking respectively amplitude dependant delay can arise at any time scale and the relative accuracy
requirement can be even more critical at lower speeds. But as I said, the original post doesn't mention this aspect. If e.g. "simple"
frequency measurement or pulse counting is intended, walking won't matter.