Noise reduction at the output

[Murugesh_89]

Hi,
In the following image


The normal value of the signal is at 5 V and it switches to 0 V on detection. Hence the number of high to low transitions taken as pulse count.

But in the steady state, the wave form oscillates between 5 to 3V and sometimes if noise is higher then voltage transition between 5 to 1.5v in the steady state. But my waveform will go to 0V on detection.

If voltage goes below 1.8 V( say 5 to 1.5V then from 1.5v to 5v) during normal state itself then controller takes this as a pulse, but it should not be happen.

So in the steady state even if the output oscillates between different levels i dont want it to go below 2.2 V. For doing this what should i do?

Thanks,
Murugesh

 

[crutschow]

You could perhaps filter the waveform. What is the frequency of the pulses and the frequency of the noise? What is the pulse width?

 

[jiripolivka]

The problem is in your signal source. The output is noisy as apparently there is no input to your device generating noise.
You may try to use a pull-up or pull-down resistor at the output but then your device might need a larger input to trigger the correct output.
Your device may have an internal level setting device to maintain the 5 V no-signal hen there is no input. Do you have the schematic?

 

[Murugesh_89]

Well thanks guys for your reply. I dont have the schematic though.
The time duration in which the output stays low is 100 micro seconds.

I got a idea but i need to check it out. If i connect the output to the base of the npn transistor with its emitter ground and connecting the one end of a 10K resistor to the collector of transistor and other end to 5V. I will take the output at the collector of transistor.

My thought is even if the output violates between 5 to 1.5V the output at the collector of transistor is 0V and when i get the proper zero signal during detection i will get 5V at the collector.

Will this work out?

 

[jiripolivka]

Well thanks guys for your reply. I dont have the schematic though.
The time duration in which the output stays low is 100 micro seconds.

I got a idea but i need to check it out. If i connect the output to the base of the npn transistor with its emitter ground and connecting the one end of a 10K resistor to the collector of transistor and other end to 5V. I will take the output at the collector of transistor.

My thought is even if the output violates between 5 to 1.5V the output at the collector of transistor is 0V and when i get the proper zero signal during detection i will get 5V at the collector.

Will this work out?

I think that this transistor inverter will only transmit the ugly noise amplified. The problem is that the device you have no schematic of is generating the noise until triggered by a specified input level.
Please try to use a 10-100 k resistor connected to ground or +5 Vcc, and try to put the other end of it to various accessible points in the device circuit. There should be a comparator which processes the AM-demodulated signal to output TTL levels; without a proper input, the output is noisy as you observed. I had a similar experience with a TX-RX pair where the RX was super-regenerative detector which is also noisy without a RF input. By setting the comparator to get rid of the noise output, the system then worked well.