• State
  • Locked Locked
  • Replies 6 replies
  • Subscribers 47 subscribers
  • Views 302 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

THS4541: Using 0.5x gain, the input is low, but up to 20 times more noise is measured on both the feedback and output ends

ting zhou
Prodigy 30 points
Part Number: THS4541


Tool/software:

1. The static test is shown in Figure 1, using 0.5 times the gain, 3V power supply, the input noise at the negative phase end is 200mV with 3uV fluctuating DC voltage, and the input voltage at the positive phase end is 1V ldo voltage (filtered noise), the voltage difference is measured at the output end is about 100mV, but the noise of the output positive and negative single-ended voltage is as high as 100~200uV noise, and the feedback end is the same, which is very different from the low noise claimed by the datasheet, is it because the DC voltage does not meet your low noise application? So how do I calculate the noise of a static signal, or how do I optimize the parameters in the graph to reduce the noise? How far can it be reduced in the best case?
2. Dynamic test as shown in Figure 2, if I use a sinusoidal input signal of 150kHz (>100kHz), which conforms to the input voltage noise mentioned in the datasheet: 2.2nV/√Hz (f > 100kHz), my adc sampling rate is 1.5M, and I still use 0.5x gain, then how do I calculate the signal noise output from the THS4541? Please assume that the built-in noise of the input source is 1, so that I can reasonably calculate the noise increased by using the THS4541 so that I can understand the real output signal noise
datasheet：



  • 0
    TI__Mastermind 24079 points

    Hello Ting,

    I am traveling for work today/tomorrow, so I can look into supporting your analysis and provide feedback later this week.

    In the interim, you can view this document on analyzing noise in high-speed amplifiers: Noise Analysis for High Speed Op Amps

    Best,

    Alec

  • 0 in reply to Alec Saebeler
    Prodigy 30 points

    okk，I'll wait for you to come back and talk about these issues together, and by the way, whether it's Figure 1 or Figure 2, I've removed the load resistor R160, otherwise I won't be able to receive the correct signal at the output, and the 500 ohm load that this datasheet repeatedly mentions seems to limit the use

  • 0 in reply to ting zhou
    TI__Mastermind 24079 points

    Hello Ting,

    Firstly I can answer your question regarding the THS4541 Bessel filter:

    Secondly, the 2.2nV/sqrt(Hz) noise specification is the input-referred noise.  This noise value is affected by the noise gain of the system, as well as the input-referred current noise and voltage noise.  Larger resistor values can also contribute their own Nyquist Thermal Noise.  The equations you have highlighted in RED and the attached noise analysis file do properly describe the process for calculating output noise.

    I can provide an example calculation, but it may be more beneficial to take your measured output noise and work backwards using the known values of e_ni and i_n.

    If we assume correct operation of the FDA and that the datasheet noise values are correct for this measurement, the resulting math does not add up.  The Rf value is either on a magnitude of 10^11, or the temperature is hotter than the device is rated for.

    I would ask you to again revisit your noise measurement; I do not think the measurement is accurately capturing the THS4541 noise; the values are too high for the selected resistor values, noise gain, and temperature (I assumed 25C / room temp).

    Best,

    Alec

  • 0 in reply to Alec Saebeler
    Prodigy 30 points
    Thank you for your reply, I haven't been able to reply to you in time due to my busy work in the past week, let's continue the discussion on the application of THS4541, I will give my design ideas, I hope you can correct:
    1) First of all, my application scenario for THS4541 is a 14bits SAR_ADC signal drive buffer, which realizes the conversion from a single-ended signal to a differential signal from 33522B, and my input signal will not exceed 1M, which is not a high-speed application, so the typical application of 10M input in the datasheet cannot be applied to my circuit;
    2) Since my lsb is 190uV, and the input range of 33522B is plus or minus 5V, 16bits, I infer that his lsb is around 152uV, which is not a small noise; So I used a 0.5x gain to try to reduce the impact of the noise;
    3) The graph in the datasheet shows that it has excellent 2nd and 3rd harmonic rejection capabilities, including the input frequency of about 15kHz for me, HD2(3)=-90dBc, which is a good data, but in fact I used 1Vcom in the application, 33522B output 500mVpp, offset is 0V, My THS4541 uses a single 3.3V power supply, yes, so the input of the THS4541 will have a certain negative voltage, I'm not sure if this will affect its ability to suppress harmonics normally, but the harmonics I measured are very large;
    4) And when we discussed the DC coupling application last week, the problem of high noise, I used a uV precision measuring instrument to measure, observe the fluctuation of the quiescent voltage, and the performance of the 33522 is very good. After using 7.5Ohm load voltage division noise reduction, the observation of its fluctuation is basically about 10uV, so it should not be a problem for me to measure, as for the resistive thermal noise, my Rf is the same as the typical application, Rg is only 2 times that of Rf, I think there should be no need to doubt the resistance here, I think the problem may be in balun, such as figure8-5, when the DC coupling application, a third-order Bessel filter is used at the input end of the THS4541 for a very steep low-pass noise filter, this filter is used very frequently, but I haven't found a way to know how to calculate the poles through the datasheet, but this filter seems to be indispensable in high-performance ADC applications, and it would be nice if you could tell me how to calculate it. My current DC application is to use a low-noise 33522B with a 7.5Ohm load voltage division noise reduction as the input negative end, and the forward terminal is connected to an LDO, which is not low noise, but simply uses 22uF, 2.2uF capacitors for filtering to provide input common mode, both of which are direct input THS4541, and there is no high-order low-pass noise filtering
  • 0 in reply to ting zhou
    TI__Mastermind 24079 points

    Hello Ting,

    Thank you for the response.  I have been on travel for a few days, so I will take time to read and respond soon.

    Thank you!

    Best,

    Alec