Hey guys, I hope this topic has not yet been discussed in that way.
I am interested in how to choose the right Op-Amp for specific circuits, like:
I wonder in which circuits it is beneficial to use a Bipolar, JFET or CMOS input Op-Amp
As I understand some things are just down to personal preference, but I would like to understand the technical background behind the decision.
So which specifications are taken into account?
The goal for me is minimal distortion and noise.
Happy to hear about your decision process
I am interested in how to choose the right Op-Amp for specific circuits, like:
- Unity gain buffer (non-inverting)
- Summing amplifier (inverting)
- Filters (i.e. Linkwitz-Riley, Butterworth etc.)
- Phono preamps
- Headphone amps
I wonder in which circuits it is beneficial to use a Bipolar, JFET or CMOS input Op-Amp
As I understand some things are just down to personal preference, but I would like to understand the technical background behind the decision.
So which specifications are taken into account?
- Gain bandwidth product
- Slew rate
- Input noise
- Offset voltage
- Input bias current
- Common-mode rejection ratio
- Input impedance
- Open loop gain
- Output current
- Capacitive load drive
The goal for me is minimal distortion and noise.
Happy to hear about your decision process
Phono preamps: MM need low current noise, so JFET or NE5534A typically, for MC low voltage noise is paramount, so AD797 is a strong contender, as are hybrid designs with low voltage noise transistors like the ZTX951.
buffers/summing amp/filters - all the same requirements really, but if there's a pot a JFET opamp doesn't require caps in the wiper path to prevent scratchiness. Headphone amps require more current drive than many opamps.
buffers/summing amp/filters - all the same requirements really, but if there's a pot a JFET opamp doesn't require caps in the wiper path to prevent scratchiness. Headphone amps require more current drive than many opamps.
Op-amps are designed for a variety of applications and so being good at some parameters is not essential for audio. For example, input bias and offset voltage are important for measurement.
Op-amps have extremely high open-loop gain to minimize error in the feedback loop. However, the gain starts falling at a low frequency (10-100Hz). Make sure that enough open-loop gain remains at 20KHz.
Ed
Op-amps have extremely high open-loop gain to minimize error in the feedback loop. However, the gain starts falling at a low frequency (10-100Hz). Make sure that enough open-loop gain remains at 20KHz.
Ed
Messing around with various opamps lately I’ve had very good luck with the old warhorses of audio - 5532, 2068, 4580, 833, etc… as all of these were designed when audio was a market leader and the big engineering fabs would put top people and top money into development. Are there better ones today? Probably, but as analog audio has not appreciably changed in 40-50yr, so the engineering that was cutting edge then is still doing an admirable job today.
If the circuit doesn’t work well and sound great with a 5532, there’s something wrong with it.
If the circuit doesn’t work well and sound great with a 5532, there’s something wrong with it.
OPA1642/1652/1656/1612 spring to mind, LM4562 as well
Some circuits need the low low bias currents of a JFET opamp such as the OPA1656. JFET inputs mean you can lose large electrolytics from the wipers of potentiometers too, and are usually less sensitive to EMI (more of an issue these days).
Golden rule though is ensure decoupling is adequate and stability isn't borderline.. Either can lead to degraded performance.
Some circuits need the low low bias currents of a JFET opamp such as the OPA1656. JFET inputs mean you can lose large electrolytics from the wipers of potentiometers too, and are usually less sensitive to EMI (more of an issue these days).
Golden rule though is ensure decoupling is adequate and stability isn't borderline.. Either can lead to degraded performance.
Thank you for all the replies so far.
I summarize:
MM Phono Preamp (High Source Impedance):
- low current noise
MC Phono Preamp (Low Source Impedance):
- low voltage noise
Suboptimal Decoupling/Stability
- low slew rate
DC-coupling Potentiometers
- low input bias current
Headphone Amp
- high current drive
In another thread some popular op-amps were simulated in both inverting and non-inverting amplifier topology and the results showed, that all the op-amps showed less distortion in the inverting topology. the best op-amps in the test for the non-inverting amplifier were LM4562 and OPA1612. I wonder which feature of these op-amps made them give the lowest THD in the non-inverting mode?
So still unclear to me:
Non-inverting Buffer:
?
Summing Amplifier:
? (same as inverting amplifier?)
Active Filters:
? (high load drive capabilty?)
I summarize:
MM Phono Preamp (High Source Impedance):
- low current noise
MC Phono Preamp (Low Source Impedance):
- low voltage noise
Suboptimal Decoupling/Stability
- low slew rate
DC-coupling Potentiometers
- low input bias current
Headphone Amp
- high current drive
In another thread some popular op-amps were simulated in both inverting and non-inverting amplifier topology and the results showed, that all the op-amps showed less distortion in the inverting topology. the best op-amps in the test for the non-inverting amplifier were LM4562 and OPA1612. I wonder which feature of these op-amps made them give the lowest THD in the non-inverting mode?
So still unclear to me:
Non-inverting Buffer:
?
Summing Amplifier:
? (same as inverting amplifier?)
Active Filters:
? (high load drive capabilty?)
Last edited:
Erm, the OPA189 cannot drive low impedances well, its distortion rockets as the load impedance falls from 10k, being pretty shoddy by 600 ohms. The distortion also rises rapidly with frequency, so the 1kHz distortion figure is misleading. It's a chopper opamp designed for low frequency DC precision applications. The low 1/f noise is attractive, but it looks like care is needed to get the best from it - its full power bandwidth is less than 20kHz for instance, ruling it out for some applications with high (intermediate) signal levels. 5Vrms is about the limit at 20kHz.
So limit it to 10k loads, low signal levels, and don't expect particularly low 10k distortion figures - and its going to fare well.
So limit it to 10k loads, low signal levels, and don't expect particularly low 10k distortion figures - and its going to fare well.