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.

Tina TI parallell op amps

Other Parts Discussed in Thread: OPA1612, TINA-TI

Hi!

I'm trying to simulate noise performance when paralleling two OPA1612 op amps. This should lower the noise by 3 dB, but It doesn't work in Tina TI. It increases the noise. If I do the same setup in LTSpice, it works as it should. Bug or am I missing something?

Jack

  • Jack,

    Thanks for looking at our amplifiers and their noise factors. I have simulated two OPA1612 in parallel with this circuit diagram

    the noise results in TINA-TI are:

    a single OPA1612           102 uV at 132.22 MHz

    two OPA1612 in parallel 144.24 uV at 132.22 Mz

     Please note that 102 u V * sqrt 2= 144.25 uV. This is very close to the simulation.

    Please send me your TINA simulation so I can compare.

  • If you parallel two opamps, the noise is uncorrelated, so the noise partially cancels out. So the noise should decrease with around 3-dB for two opamps. 

    Here's a simulation from LTSpice.

    Single opamp and noise output gives 36 nV/Hz:

    Parallel couple the opamps and you get around 3 dB lower noise, in this case 32.8 nV/Hz.

    This doesn't work in Tina90.

    To quote Douglas Self:

    "If there are two amplifiers connected, the signal gain increases by 6 dB due to the summation. The noise from the two amplifiers is also summed, but since the two noise sources are completely uncorrelated (coming from physically different components) they partially cancel and the noise level only increases by 3 dB. Thus there is an improvement in signal-to-noise ratio of 3 dB. This strategy can be repeated by using four amplifiers, in which case the signal- to-noise improvement is 6 dB. Table 1.9 shows how this works for increasing numbers of amplifiers. In practice the increased signal gain is not useful, and an active summing amplifier would compromise the noise improvement, so the output signals are averaged rather than summed, as shown in Figure 1.10. The amplifier outputs are simply connected together with low-value resistors; the gain is unchanged but the noise output falls. The amplifier outputs are nominally identical, so very little current should flow from one op-amp to another. The combining resistor values are so low that their Johnson noise can be ignored."

  • Jeff,

    This is a very interesting claim that Douglas has made. Can you send me the link to his article?

  • Jeff,

    I have given this issue more consideration. Here are my thoughts.

    I am seeing from the TINA-TI simulation ~ 3dB improvement in SNR. Here is my TINA-TI data.

    As you can see, at 17.04 KHz, the Single amplifier has a SNR of 109.65 dB and the parallel combination has a SNR of 112.66 dB.

    From your article, the text reads

    "If there are two amplifiers connected, the signal gain increases by 6 dB due to the summation. The noise from the two amplifiers is also summed, but since the two noise sources are completely uncorrelated (coming from physically different components) they partially cancel and the noise level only increases by 3 dB. Thus there is an improvement in signal-to-noise ratio of 3 dB.

    Please note that the SNR is not the actual noise pose. For you LTSpice simulation I am not seeing a noise reduction of 3 dB. Rather 20 * LOG(32.8 nV/ 36 nV) = -0.8 dB.

  • "In practice the increased signal gain is not useful, and an active summing amplifier would compromise the noise improvement, so the output signals are averaged rather than summed, as shown in Figure 1.10. The amplifier outputs are simply connected together with low-value resistors; the gain is unchanged but the noise output falls. The amplifier outputs are nominally identical, so very little current should flow from one op-amp to another. The combining resistor values are so low that their Johnson noise can be ignored."

    Regards,

    Jack Zimmermann

  • Jack,

    This information is nearly expected. I assume that this quote is again coming from Douglas Self.

  • Yes, it's from the book "Small Signal Audio Design" by Douglas Self.

    I'm not sure if I'm doing something wrong, or if the Tina simulator doesn't work with parallel opamps, but my measured values shows lower noise levels as do the LTSpice simulations.

    Regards,

    Jack

  • Jack,

    I guess that I am confused.

    From your data: Single opamp you get noise output gives 36 nV/Hz and Parallel couple the opamps you get 32.8 nV/Hz. Although the second number is lower, it is not 3 dB lower rather it is ~ 3.2 nV/Hz lower, rather 0.609 dB lower..

    With the TINA TI data I sent to you has a SNR of 109.65 dB for the single amplifier and a SNR of 112.66 dB for the parallel amps. This is a 3dB improvement. Perhaps you are looking for a negative value for the SNR and you can get that if you change the ratio in the SNR formula. The formula I am using is

    SNR = 20 * log10(Signal rms range / rms noise).

    Since rms noise is less than Signal rms range the value will be positive. So lets change the calculation --

    For single amplifier - SNR = 20 * log10(rms noise / Signal rms range ) = -109.65 dB

    For parallel amplifiers - SNR = 20 * log10(rms noise / Signal rms range ) = -112.66 dB

    You can now see that the noise is reduced by using parallel amplifiers. TIINA TI demonstrates this.

  • Great! Thanks for the help. Much clearer now.

    Regards,

    Jack