LM8261 - Filter design & Noise

Hello,

I am Jagdish, I need certain help in understanding the opamp filter behavior. I tried simulate to get the answer and could find good reason to justify.


I am using filter with an opamp LM8261, on slide#1 i have input coming ultra sonic transducer (which has some considerable amount noise present) which i am trying to filter using active filter stages. On Slide#2, i am trying to measure output and what i see is much noisy signal and amplitude levels are increased.

On slide#3, I tried to change R16 value and i could able to see much cleaner signal. I want to understand this whether something is wrong in my filter design or what difference is that i am making by changing R16 value. (Vref=0.44)

I'll appreciate some help on this to understand whether do i need to change the opamp which can work better in reducing the noise. thx

Regards,JG


False Alarm.pptx

5 Replies

  • Hello Jagdish,

    Can you please show the circuit following the 1.1k/6.8k parallel resistor?. This is important. The full schematic and/or your simulation would be helpful.

    Is Vref always 440mV? What is channel 2 showing in the Pg3, bottom right scope photo?

    Regards,

    Regards,

    Paul Grohe

    Low Power Amplifiers and Comparators (LPAC) Applications

  • In reply to Paul Grohe:

    Hello Paul,


    Please see my schematics. thx

  • In reply to Jagdish Gundu:

    Hi Jagdish,

    Can you explain the last scope photo? The channels and positions have changed. Is the 8261 ouput now Ch4 (green)? I assume CH 2 is now the comparator output.

    Is there a reason you have bypassed the internal stepped-gain amplifier stage? This internal stage gradually increases the gain by 32 steps (0 to +35dB) after the TX pulse. This is done to prevent overloading of the proceeding processing stages during TX. It is common to use a log amp or time-changing gain amplifier to compress the initial TX pulse and increase the gain for the echos.

    Adding a fixed gain stage (x16 in your case), could overload the bandpass and comparator stages by the TX pulse.

    Changing R16 changes the low pass filter passband. It should not have any effect on the external amplifier.

    Where exactly are you measuring on page 3, bottom right? The output of the LM8261 is a low impedance output (very low - as it is a high-current amp), so increasing a 1k resistor on the other side of a 1k resistor should not make any change (assuming you are looking at pin 1 of U3).

    IF you are looking at the input of the low pass filter (pin 5), then the amplitude can change.

    Your "noise" has a very periodic pulse pattern to it, about 100us, or 10KHz. That looks like capacitive noise pickup. Do you have any other systems running at 10kHz, or a multiple of 10kHz (Switchers? Clocks?).

    Piezo senors are high impedance (in RX), and can easily pick up environmental noise (capacitive AND sonic). This noise could be limiting your measurement (causing early false trip). The "stepped" gain also reduces the impulse noise so that it will not false-trigger soon after the TX pulse - but you will still false-trigger at some point.

    The sensor leads should be short and preferably twisted. But avoid long coax as it can add extra capacitance. The sensor and receive circuit should be close to each other and within a shielded enclosure.

    It can also be ground loops. Even though the transmit pulse voltage is >100V, the receive side is a small signal circuit. Check for ground loops, wire routing, bypass cap grounding, etc to make sure you are not picking up local noise.

    What happens if you power just the sonic circuit from a *linear* power supply?

    Regards,

    Regards,

    Paul Grohe

    Low Power Amplifiers and Comparators (LPAC) Applications

  • In reply to Paul Grohe:

    Hello Paul, thanks for the Reply.

    Yes you are correct the last photo and scope channel assignment.

    Yes we have not opted internal gain amplifier as application requirement is something different than what actual IC does.

    Is there any mitigation that we can avoid overload on bandpass & Comparator stages? I have done some calculations on filter calculations and i need your help in understanding on how its actually working.

    "Where exactly are you measuring on page 3, bottom right? "  I am measuring Echo pulse on U3 PIN. I tried having isolated power supply and also circuit network analysis to clear all ground loops.

    No such sources of periodic noise are captured. I have 200Khhz of transducer primary side switching circuit and 65kHz of SMPS frequency (AC-DC Fly back converter).

    I need your support to understand what difference that it is making in the circuit that if i replace R16 from 1.1K to 6.8K value.

    Regards,

    JG

  • In reply to Jagdish Gundu:

    Hello Jagdish,

    The stepped-gain amplifier is what avoids the overloads - and it's final gain (35dB) is higher than your external amp. You would be better off using that - as intended.

    Are you changing R16 or R17, or both?

    Changing either of these changes the filter poles. Increasing R17 or R16 to 6.8k moves the LP frequency below 80kHz. You are cutting off your 200kHz signal! It looks 'cleaner" because you have reduced the entire signal.

    The filter resistors and capacitors need to be changed in pairs. You cannot just change one value.

    The proper way to "clean-up' the signal is to narrow the passband bandwidth around 200kHz. You can add external filter stages to add more poles.

    Filters are a semester course in themselves, and are beyond the scope of this support forum.

    However, there are many tools available to help. One of these is Filter-Pro:

     http://www.ti.com/filterpro-dt

    And our on-line filter designer:

     http://www.ti.com/lsds/ti/analog/webench/webench-filters.page

    This is Thomas Kuhels' favorite on-line filter calculator:

     http://sim.okawa-denshi.jp/en/OPseikiLowkeisan.htm

     http://sim.okawa-denshi.jp/en/OPseikiHikeisan.htm

    Of course, your favorite simulator, like TINA-TI, can help you fine-tune the circuit.

    The internal LP and HP amps look to be simple buffers, so you can use an op-amp in a follower configuration to represent the internal amplifiers.

    In filter-pro, select a low pass or high pass , 2 pole, gain of 1 Sallen Key filter (where the amp is a buffer - which mirrors the buffer amps in the controller). Then you can tweak the circuit from there.

    The LM8261 is not the cause of the problem. In fact, you may want to reduce the 47pF feedback cap to more like 10pF - as you are starting to roll-off before 200kHz.

    I think you would be better off using the internal stepped-gain amp and adding additional bandpass filtering stages to narrow the bandwidth around 200kHz.

    Regards,

    Regards,

    Paul Grohe

    Low Power Amplifiers and Comparators (LPAC) Applications