Optocoupler "Bandwidth"

[PinheadBE]

Hi,

I have a frequency meter of which one of the inputs is a 100 MHz TTL input, with no protection at all.

All protection circuits I find imply a low-pass filter, so I decided to try with an optocoupler with TTL logic outputs.

I found online the TLP2361.
In its datasheet https://toshiba.semicon-storage.com/info/TLP2361_datasheet_en_20151102.pdf?did=14229&prodName=TLP2361 ,
it is stated that its rise time as well as its fall time are 3 ns, at most.

May I assume that its bandwith is then 1/(3ns + 3ns) = 166 MHz, and that it will then be OK for my application, if I only want to measure the frequency ?
(And not the length of the pos or neg pulses) ?

(Note that the propagation delay and/or any distortion in the ouput signal has no importance in my case)

Thanks for your ideas

[Kleinstein]

The linked optocoupler has a digital output. The rise/fall time is not reflecting the bandwidth or speed of the actual coupler. So one can not use the rise-time to BW curve as for a scope. The speed is given with 15 Mbaud - so at most some 7.5 MHz.

The usual protection is with some series impedance to limit the input current and some clamping (e.g. diodes). One does not need much low pass charateristics, as the series impedance can have a capacitive compenent, letting more of the high frequencies through.

[Manul]

I have never seen of the shelf optical couplers with such bandwidth. There are capacitively and magnetically coupled digital isolators which would pass 100MHz, so it's not hard to make, but they are somewhat expensive.

[TimFox]

Toshiba makes some very fast optocouplers, but I think they top out at 50 M bit/sec.
https://www.mouser.com/datasheet/2/408/TLP2767_datasheet_en_20160422-1075464.pdf

[PinheadBE]

OK, thanks to all !

It appears I will have to search something else to protect my TTL input.

Of course there is this (already mentionned in other topics):
https://www.digikey.com.au/en/articles/protecting-inputs-in-digital-electronics

I read it before posting, but the low-pass filter was a problem to me.   

May I simply omit it ?   

Inserting a 1k resistor followed by two schottky's (to 5V and to GND) is enough ?

[Kleinstein]

1 K and 2 schottky diodes are a first step. One should also have some TVS diode at the 5 V supply.

To get more of the fast signal through one may want some capacitance (e.g. 10-100 pF range) in parallel to the resistor. One can than usually also in crease the resistance. The details depend on the actual input gate (e.g. 74ACT... or 74AC... or  74ALS... ) and the input capacitance / levels.

[PinheadBE]

This is the input I would like to protect.
It's from the FC8000 frequency meter made by ELV (nice bit of equipment, btw  )

From the datasheet, a 74LVC can only accept an absolute maximum input voltage of 6.5 V if the current in the pin is limited (therefor the 1k resistor in series)

I would like to clamp at 5.5 V at most, to be on the safe side, so Schottky's to the power rails should do.   
But I'm not sure if the capacitance of those diodes will not reduce the bandwith

[Kleinstein]

The capacitance of the input and the diodes will reduce the BW a little. With a capacitor equal to that capacitance one would only get a reduction in the gain by a factor of 2. With a larger capacitance in parallel to the resistor one would get even less reduction in the sensitivity. It is still a balance with the protection against fast transients and ESD.

[timeandfrequency]

Hello PinheadBE,

Putting a resistor (a few K perhaps ?) in series with IC301 input n°5 and some good TVS to GND and VDD should protect your counter input circuit.

[PGPG]

This is the input I would like to protect.
It's from the FC8000 frequency meter made by ELV

Isn't it a simplified schematic and protection that you want to add externally is already there inside?