+ Post New Thread
Results 1 to 14 of 14
  1. #1
    Advanced Member level 5
    Points: 11,117, Level: 25
    Achievements:
    7 years registered
    tony_lth's Avatar
    Join Date
    Jul 2010
    Location
    Beijing
    Posts
    1,967
    Helped
    376 / 376
    Points
    11,117
    Level
    25

    How to detect this impulse?

    HI, Gurus,
    I need to detect an impulse:
    rising edge=5ns, down edge=200ns, amp=10uV, just once in a day, and I need to catch it, how can I do?

    I suppose to config as the following: filter1+LNA+filter+detector+video amp+LPF+ADC/FPGA.
    But there are so many specs need to determine:
    1. Filter1 center freq, BW. Say filter1 BW=100M, then noise floor is -174+80=-94dBm. SNR=10, so signal energy=-84dBm at least. But 10uV is -87dBm. So how can I determine the center freq and BW?
    2. Why is the detector purpose in this link? I guess ADC can directly sample the 100M(Or other BW?) signal.
    3. For detect such low signal, how to do the power supply? And how to select the video amp or op-amp?
    Many thanks.
    Tony Liu

  2. #2
    Advanced Member level 5
    Points: 31,617, Level: 43
    BigBoss's Avatar
    Join Date
    Nov 2001
    Location
    Turkey
    Posts
    4,644
    Helped
    1407 / 1407
    Points
    31,617
    Level
    43

    Re: How to detect this impulse?

    How about Pulse Duration ? If it's sufficiently long, it's possible to catch it with RF Detector ( see RF Power Detectors ) with an appropriate Low Noise Amplifier ( NF should be as low as possible and depends on the frequency )


    1 members found this post helpful.

  3. #3
    Member level 3
    Points: 681, Level: 5

    Join Date
    Mar 2018
    Posts
    60
    Helped
    11 / 11
    Points
    681
    Level
    5

    Re: How to detect this impulse?

    Do you need to just detect or also measure the pulse ?

    Is the pulse base line a stable known value ? Thinking of using a comparator
    with 2 control loops, one for DC baseline drift, and one for relative detection
    using a fast comparator.


    Regards, Dana.


    1 members found this post helpful.

    •   AltAdvertisement

        
       

  4. #4
    Advanced Member level 5
    Points: 11,117, Level: 25
    Achievements:
    7 years registered
    tony_lth's Avatar
    Join Date
    Jul 2010
    Location
    Beijing
    Posts
    1,967
    Helped
    376 / 376
    Points
    11,117
    Level
    25

    Re: How to detect this impulse?

    Hi, Bigboss,
    The duration is almost zero, just rising and down, it is a discharge procedure.

    Hi, Dana,
    Just want to detect it, and measure roughly, no need to very accurate.
    So do you have some example about the 2 loops? I have no idea about it at all.

    Many Thanks.
    Tony Liu

    - - - Updated - - -

    PS:
    Dana, the pulse base line should be known value. I can measure it all the time except the pulse happened.

    - - - Updated - - -

    Hi, Dana,
    The signal is very low, about 10uV, but comparator need high voltage.
    So I need to amp the signal.
    Which type amp should I use? The RF LNA, or low noise op-amp? How to determine the BW?
    Many thanks.
    Best,
    Tony Liu



  5. #5
    Member level 3
    Points: 681, Level: 5

    Join Date
    Mar 2018
    Posts
    60
    Helped
    11 / 11
    Points
    681
    Level
    5

    Re: How to detect this impulse?

    The general concept of using a fast AC amp with a slow precision OpAmp
    to form a compound best of both worlds OpAmp in this -

    http://www.ti.com/lit/an/snoa818/sno...=1590192863075


    Regards, Dana.


    1 members found this post helpful.

    •   AltAdvertisement

        
       

  6. #6
    Advanced Member level 5
    Points: 31,617, Level: 43
    BigBoss's Avatar
    Join Date
    Nov 2001
    Location
    Turkey
    Posts
    4,644
    Helped
    1407 / 1407
    Points
    31,617
    Level
    43

    Re: How to detect this impulse?

    This signal must be amplified first, it's obvious otherwise no active device can detect this level..
    But your calculation is very rough so you cannot interpret this pulse as ordinary sinusoidal signal.10uV can be detected by an Low Noise Amplifier but the bandwidth should be as narrow as possible so that amplifier does not add extra noise.Bandwidth will be roughly 0.35/tr=70MHz may be more or less.So, A DC-100MHz LNA will I believe serve your purpose..Then you can easily detect this amplified signal.
    Of course practical try out will absolutely give you a reasonable result.


    1 members found this post helpful.

    •   AltAdvertisement

        
       

  7. #7
    Advanced Member level 5
    Points: 11,117, Level: 25
    Achievements:
    7 years registered
    tony_lth's Avatar
    Join Date
    Jul 2010
    Location
    Beijing
    Posts
    1,967
    Helped
    376 / 376
    Points
    11,117
    Level
    25

    Re: How to detect this impulse?

    Hi, Danadakk,
    Your name "Dana" in Chinese is meaning "a great Guru", interesting. Many thanks.
    Best,
    Tony Liu



  8. #8
    Advanced Member level 5
    Points: 11,117, Level: 25
    Achievements:
    7 years registered
    tony_lth's Avatar
    Join Date
    Jul 2010
    Location
    Beijing
    Posts
    1,967
    Helped
    376 / 376
    Points
    11,117
    Level
    25

    Re: How to detect this impulse?

    Quote Originally Posted by BigBoss View Post
    How about Pulse Duration ? If it's sufficiently long, it's possible to catch it with RF Detector ( see RF Power Detectors ) with an appropriate Low Noise Amplifier ( NF should be as low as possible and depends on the frequency )
    Hi, Bigboss,
    The pulse oscillate in the cable, so its duration is about 10uS or more.
    Can I detect it with RF detector? Can the RF detector detect the pulse waveform? And need more about the pulse, such as phase vs freq etc, can you comment it?
    Best,
    Tony Liu



  9. #9
    Advanced Member level 5
    Points: 18,913, Level: 33
    Achievements:
    7 years registered

    Join Date
    Nov 2012
    Posts
    3,361
    Helped
    841 / 841
    Points
    18,913
    Level
    33

    Re: How to detect this impulse?

    rising edge=5ns, down edge=200ns, amp=10uV, just once in a day, and I need to catch it, how can I do?
    Your waveform appears to be similar to sawtooth. The fundamental is therefore at 100MHz. But if there is only one pulse, the fourier transformation is a sinc function. For that the fundamental is at 0 Hz!!

    Please pay attention to details. The impulse is not repeating. Hence a 100MHz BP filter is not suitable. Please do a FT of the pulse and take the first lobe on the right side. It should be more than 200MHz.

    Next you need to estimate the noise floor. What will be approx S/N ratio under your real life conditions?

    If SNR is 10 or so, you are on stable ground.


    1 members found this post helpful.

  10. #10
    Advanced Member level 5
    Points: 11,117, Level: 25
    Achievements:
    7 years registered
    tony_lth's Avatar
    Join Date
    Jul 2010
    Location
    Beijing
    Posts
    1,967
    Helped
    376 / 376
    Points
    11,117
    Level
    25

    Re: How to detect this impulse?

    HI, C_mitra,
    Thanks for reply.
    My purpose is to detect very high voltage cable leaking current pulse.
    This object has been studied for about 100 years, begining from UK.
    The pulse is about 5ns-rising/200ns-falling, but some text book says that this pulse will OSC in the cable for about 10us or more.
    So I am going to follow bigboss comment:
    Design a DC-50M pre-filter,+LNA, then input to ADC. So noise floor is about -90dBm, and with SNR say 6dB, I expect to detect -84dBm signal.
    It is said that main signal is in 3-30M, but it is truly very wide, some use 1-1.5GHz.
    Frankly, I don't understand how to detect 1-1.5G signal for this pulse.
    Could you give me some help?
    Many thanks.
    Best,
    Tony Liu



  11. #11
    Super Moderator
    Points: 266,396, Level: 100
    Awards:
    1st Helpful Member

    Join Date
    Jan 2008
    Location
    Bochum, Germany
    Posts
    46,553
    Helped
    14167 / 14167
    Points
    266,396
    Level
    100

    Re: How to detect this impulse?

    It is said that main signal is in 3-30M, but it is truly very wide, some use 1-1.5GHz.
    Frankly, I don't understand how to detect 1-1.5G signal for this pulse.
    Either your assumptions about 5 ns pulse risetime are wrong or they are using unnecessary large bandwidth.

    What's the objective of your measurement system? Detect pulses reliably and e.g. classify their magnitude/energy or acquire the exact waveform? In the latter case, a higher bandwidth than 50 or 100 MHz may be wanted, but consider that a 50 ohm resistor involves a noise density of 1 nV/Hz^0.5 respectively 10 µV noise voltage at 100 MHz noise bandwidth. Thus a smaller measurement bandwidth can be necessary to detect anything. You'll also think about signal source impedance and noise matching.


    1 members found this post helpful.

  12. #12
    Advanced Member level 5
    Points: 18,913, Level: 33
    Achievements:
    7 years registered

    Join Date
    Nov 2012
    Posts
    3,361
    Helped
    841 / 841
    Points
    18,913
    Level
    33

    Re: How to detect this impulse?

    but some text book says that this pulse will OSC in the cable for about 10us or more...
    Are you using a coaxial cable? What is the nominal impedance? Are the impedances (freq bandwidths) nicely matched?

    Perhaps I am not getting the problem right.


    1 members found this post helpful.

    •   AltAdvertisement

        
       

  13. #13
    Advanced Member level 5
    Points: 11,117, Level: 25
    Achievements:
    7 years registered
    tony_lth's Avatar
    Join Date
    Jul 2010
    Location
    Beijing
    Posts
    1,967
    Helped
    376 / 376
    Points
    11,117
    Level
    25

    Re: How to detect this impulse?

    Hi, FvM,
    The rising is about 1ns, so it's truly that the signal can reach 3G.
    In many electric power standard, for such sensor, the detected freq band is 300M-1.5G. The sensor uses antenna to detect the discharge inside the high-voltage box. So the impedance for my sensor is 50 ohms.
    But they can detect 1pC min, that is about 10uV.
    Yes, the final design should give PRPS chart and PRPD chart. That is some phase, freq, amplitude, etc., chart.
    I guess I need to do sth like a spectrum analysis, right?
    Best,
    Tony Liu

    - - - Updated - - -

    PS:
    The discharge pulse emerge TEM wave, which osc in the power cable, from one insulated point to the other. Then its duration maybe more than 10uS.
    And the sensor using antenna to detect it from some windows in the power cable.
    That is two systems.
    Best,



  14. #14
    Advanced Member level 5
    Points: 11,117, Level: 25
    Achievements:
    7 years registered
    tony_lth's Avatar
    Join Date
    Jul 2010
    Location
    Beijing
    Posts
    1,967
    Helped
    376 / 376
    Points
    11,117
    Level
    25

    Re: How to detect this impulse?

    Hi. Gurus,
    I found some examples. The following is one example of the pulse on time domain and spectrum.

    - - - Updated - - -

    My object is to get the following three charts, for example of the corona:

    - - - Updated - - -

    Shengxi Station Example.PNG is one example of the live station result.

    - - - Updated - - -

    "Corona-PRPD.PNG" is the corona PRPD (Phase Resolved Partial Discharge), that is the superposition of PRPS.
    "Corona-Peak-Three-Phase.PNG" is the peak detector of the three phase.
    "Corona-Amp-Cycle-Phase.PNG" is PRPS (Phase Resolved Pluse Sequence), it's three axis are amplitude, cycle #, and phase.
    "Pulse Current vs time.PNG" is the pulse on time domain.
    "Pulse Spectrum example.PNG" is the pulse spectrum.

    - - - Updated - - -

    1.My design now is DC-50MHz, LNA+Driver Amp, ADC, FPGA, etc.
    Could you pls comment on my design?
    2. If design from 0.3-1.5G, how can I do? Should I design sth like a spectrum?
    many thanks.
    Best,
    Tony Liu



--[[ ]]--