• State Resolved
  • Replies 6 replies
  • Subscribers 49 subscribers
  • Views 189 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

OPA991-Q1: High-frequency signal distortion - is this normal?

Zijiang Yang
Prodigy 20 points
Part Number: OPA991-Q1
Other Parts Discussed in Thread: OPA991, TINA-TI, OPA323

Tool/software:

Hello Analog Experts,

I am new to amplifier circuits and have a question regarding high-speed signal distortion with the OPA991.

Given the OPA991’s gain-bandwidth product of 4.5MHz, I am curious about the maximum frequency of a sine wave signal it can reliably amplify when set up as a simple voltage follower (non-inverting input, gain = 1).

To test this, I conducted an experiment using a signal generator and an oscilloscope with the following setup:

  • VDD = 3V, VEE = 0V (single supply)
  • Signal biased at 1.5V DC
  • Amplitude: 1V peak (2V peak-to-peak)

I tested three different frequencies: 1kHz, 100kHz, and 1MHz. Here are the oscilloscope snapshots for each case: (yellow Being the input, where blue being the output of OPA991)

  100khz1mhz

I observed significant distortion at higher frequencies. Is this amount of distortion normal and expected? When I simulated the same setup in TINA-TI using the OPA991 SPICE model, I did not notice any significant distortion.

Thank you for your assistance!

  • 0
    Prodigy 20 points

    Here's my TINA-TI simulation result of the 1MHz sine wave, as you can see the output (brown) is not so distorted

  • 0
    Guru 259940 points

    Both GBW and slew rate can limit high-frequency signals.

    These waveforms look as if the slew rate is too small. The OPAx991 can do 21 V/μs. What is the slope of your input signal?

  • 0 in reply to Zijiang Yang
    TI__Guru*** 142926 points

    Hello Zang,

    I'm always curious when someone uses a high voltage op amp at low voltage. Yes, I know 3V is allowed, but we have so many other low voltage op amps from which to chose.

    A perfect op amp (infinite gain) would not distort. This is because the output will force the two input be be the same, virtual short. At 1MHz the OPA991 would have a open loop gain of 4.5 V/V, that a low number. The virtual short is no more. If out was 1Vp then difference input voltage is 1000mVp / 4.5 = 222mVp. So that isn't unity gain anymore. It is a bit more complicated than that because the output phase lag can become positive feedback. 

    Slew rate is another (actually bigger) concern. Peak slew rate needed is  Vpp*pi*f = 2V*3.14*1MHz = 6.28V/us ; how much input difference is needed to get that? There as app note that has that data. [APP] Ramping Up on Slew Rate Per figure 3-4, almost 200mV is needed (almost 20% of your input signal). 

    Therefore OPA991 is not fast enough to buffer the 1MHz input waveform. The simulation should show more distortion, but the slew rate may not be modeled very accurately.   

  • 0 in reply to Ron Michallick
    Prodigy 20 points

    Thank for your detailed explanations. To answer your "curiosity", it happens i use OPA991 for another high-voltage application, and i was lazy to select another op-amp, because when i looked at the datasheet, 4.5MHz GBP and 21V/us, supply down to 3v is probably enough for my 1MHz  buffer application.

    So as i understand now, this is clearly not enough. Can you give me a sort of 'rule of the thumb' to determine the minimum GBP? Would a 20MHz GBP op-amp such as OPA323 suffice for this application? 

  • +1 in reply to Zijiang Yang
    TI__Guru*** 142926 points

    Good question. The rule of thumb depends on the output quality expected. 

    Here is the OPA323 unity gain buffer sim. At 1/1000 the BW (20kHz), gain peaking (gain error) starts. This happens when open loop gain's phase drops below -90 degrees. The gain peak isn't very large (on the simulation) so 1/10 the BW (2MHz) is not much gain error.

    For output phase lag; it starts around 1/100 BW (200kHz).
    At 1MHz, 0.1dB gain error and -2 degrees lag expected.  Good enough for many applications.  

    Above is the small signal (think tiny) expectations.

    For slew rate, I always get suspicious when data sheet slew rate (in V/us) exceeds BW in MHz. (33>20). That makes me think slew boost gives this high number.

    I like a natural slew rate at least double the peak rate needed, so at least 12.56V/us. The natural slew of OPA323 should be close to that. For best output sine quality and phase preservation, I'd go a little higher on device BW 

    However, there's a good chance OPA323 can provide a decent 1MHz 2Vpp output. 

  • 0 in reply to Ron Michallick
    Prodigy 20 points

    Thank you Ron, your are indeed an analog expert! I have no further questions. Hopefully OPA323 will satisfy my application.