Tunable thresholder

Hi Danilo,

Due to the speed, this doesn't belong in the precision amps. E2E realm (our devices are < 50 MHz).  But just for my understanding - this is basically like a "one-shot" circuit with a programmable threshold, correct?  It seems you are looking for a high-speed comparator, along with high speed logic for the pulse generation.  I think these would come from two different product lines, if they exist.

Do you agree with this assessment?

Best Regards,
Mike

Hi Mike,

Thank you for your response. We have received further inquiries from our customer below

Forgetting about the data rate, do you think such a configuration will still help?

What about If I use a single comparator that has a latch enable port as a single module, then whenever the latch is off the output will maintain the previous state for a duration.

Regards,

Danilo

A "peak-detector" might be a good start. It is essentially an amplifier that drives an RC timer through a diode. There are several high speed amplifiers with >1GHz bandwidth to choose from. It could be tuned to decay within 1ns, or, to get the 90° falling edge, you could set it for much longer and then short the capacitor with a parallel switch every nanosecond. What is the required gain? I think you could try to do this with the OPA858.

Hi Sean,

Thank you for your suggestion. According to our customer,

1) Are these types of amplifiers limited to a maximum of 1 GHz?
2) If I fed 10 GHz will it still work and only samples every 1 GHz or not working at all?
3) Would you please elaborate more on your explanation on how can use it as a thresholder?
specifically:
3. a) What should I feed in+ as my input signal and in- as the reference signal with which I want a comparison with?
3. b) How can I short the capacitor with a parallel switch every nanosecond
3. c) will the output be a just detected peak or it will be also a pulse longer than one bit depending on my configuration?
3. d) If yes to the previous part how can I adjust this pulse duration?
4) The output of this one can go even up to 27 dBm electrical power and higher is better.

Regards,

Danilo

1. The highest closed loop bandwidth I have seen is around 1GHz. You might be able to use an open loop buffer for up to 3GHz.

2. A 10GHz signal would be filtered out by a 1GHz amplifier. 

3. I am trying to help you understand the tools we have available in order to determine whether or not a possible solution exists, but I don't know if it will work for sure.

a. I was suggesting using the peak detector to extend the duration of all peaks, without thresholds. However, the diode will be a constant 0.7V threshold to charging the capacitor. It's theoretically possible to adjust this "threshold" several 100mV based on the diode's I-V curve and varying the R value.

b. You could use a high-frequency, discrete MOSFET (open-drain configuration).

c/d. You can design the RC constant either way. You just have to adjust the RC values for the desired time constant.

4. Higher voltages are going to burn more power, and it will be harder to alternate quickly.

What is this signal coming from? What is the selectivity needed by the threshold? Why is the frequency so high?

Hi Sean,

We have received this feedback from our customer,

What is this signal coming from?
It's coming from an optical data and after passing through PD it will convert to electrical signal to be thresholded

What is the selectivity needed by the threshold?
The input is a 3 level electrical signal and the output should be just 2 level and this threshold level used for suppressing the middle amplitude levels should be adjustable

Why is the frequency so high?
It's coming from optic and even 10 GH is low for such a signal

Regards,

Danilo