10
\$\begingroup\$

Opamps U1/2/3 are current-feedback opamps used in a RF signal conditioning circuit. U1/2 control the gain through R8 and S1. U3 along with U7 allow for polarity toggling. What I don't understand is the 4.7 ohm resistors in series with U1/2/3 supply pins. I suspect the current-feedback topology of the opamps has something to do with it? Is current sourced from these resistors in some way? What would removing these resistors do to the opamp performance (generally)? circuit

\$\endgroup\$
4
  • 3
    \$\begingroup\$ Not specific to CFAs. Noise. Look up regulator PSRR. \$\endgroup\$
    – DKNguyen
    Jul 5 at 13:40
  • 1
    \$\begingroup\$ @DKNguyen great comment. Very detailed and informative! \$\endgroup\$
    – earl
    Jul 5 at 13:41
  • \$\begingroup\$ Popular cheat to low pass filter anything. If your device downstream consumes very little current, the voltage drop across the resistor is low. If it consumes a lot of current, you don't have this option and need either CLC or active filtering. \$\endgroup\$
    – winny
    Jul 5 at 13:54
  • 1
    \$\begingroup\$ The resistor isolates the ICs from the +6V rail, so not only is the IC benefiting from some RC filtering, the +6V is somewhat protected from current spikes from one IC polluting the +6V rail and affecting other ICs. \$\endgroup\$
    – td127
    Jul 6 at 17:28

3 Answers 3

9
\$\begingroup\$

PSRR Power Supply Rejection Ratio is 65 dB min and unity gain BW is 300 MHz means the PSRR drops to zero near the GBW so PS filtering is essential to avoid degrading the SNR of the output signal.

You may need to define your PS ripple spectrum but the harmonics are attenuated by this filter.

RC= T= 47 us step edge noise is reduced 37% and 20 dB /decade above the breakpoint.

It is always advisable to include RC filters for all UHF IC's to decouple noise from the DC power source. (if you care about high SNR and avoiding positive feedback between sources and inputs that lead to spurious oscillations)

added

TY @JohnDoty yes adding the RC LPF not only reduces dV/dt but also dI/dt in the external power loop traces that may have stray H fields or conduct ground shift noise from load steps.

In other words, it improves both ingress & egress spectral attenuation both conducted and radiated with a damped cap across IC.

\$\endgroup\$
1
  • 1
    \$\begingroup\$ It's not just PSRR. The decoupling resistors also reduce the injection of rapidly varying current into the supply rails, thus reducing inductive crosstalk. And it's not just UHF: even at much lower frequencies decoupling is critical for achieving high SNR. \$\endgroup\$
    – John Doty
    Jul 6 at 15:59
9
\$\begingroup\$

What I don't understand is the 4.7 ohm resistors in series with U1/2/3 supply pins.

You get an improvement in PSRR (power supply rejection ratio) when your potentially noisy supply lines are locally noise decoupled to each op-amp using an RC filter. This comes at the expense of slightly lower full-scale output voltages. Net PSRR improvement in blue: -

enter image description here

Image from EL5163 data sheet.

I suspect the current-feedback topology of the opamps has something to do with it?

Not one bit.

Is current sourced from these resistors in some way?

Current flows through resistors and, unless you are talking about thermal noise emissions of resistors (that are very tiny for 4.7 Ω) current is not sourced from a resistor.

\$\endgroup\$
3
  • 1
    \$\begingroup\$ Is RC decoupling common? In reading datasheets for opamps, I've only really seen suggestions of decoupling capacitors on the supply pins. \$\endgroup\$
    – earl
    Jul 5 at 13:46
  • \$\begingroup\$ They are on the supply pins and enhance the presence of the decoupling capacitor. This technique used to be quite common but, as op-amp improvements have been made over the years, it is seen less and less. \$\endgroup\$
    – Andy aka
    Jul 5 at 13:47
  • 3
    \$\begingroup\$ You always have some impedance on the power supply lines, and at high frequencies potientially a lot due to inductive reactance, so there is always some low pass filtering provided by the decoupling capacitors. Putting the resistors adds a well-controlled resistance so that lower frequencies are attenuated too. You see this done a lot with things like photodiode amplifiers on the photodiode bias so that the amp won't amplify the power supply noise at the input. \$\endgroup\$ Jul 5 at 14:00
3
\$\begingroup\$

" Is current sourced from these resistors in some way?"

Yes, it is. But just having a resistor is not the whole story. Equally important is the capacitor attached to it. The two combine to (somewhat) isolate the other op amps from power supply variations caused by the varying current drains as each op amp varies its output voltage.

" What would removing these resistors do to the opamp performance (generally)?"

Generally speaking, removing the resistors will increase the likelihood that the op amps will interact with each other, usually in such a way as to produce high-frequency oscillation.

This can be a problem even for audio-range op amps, particularly in the old days when multilayer pc boards were not widely used, and ground and power distribution could be less than optimal. Going to ground planes and voltage planes generally avoids the issue. Except, of course, for high-frequency op amps like the EL5163. Those puppies work out to 500 MHz, and it's easy for parasitics to cause unwanted coupling between op amps.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.