Hi,
as usual the answer is...it depends.
In this case it depends on the type of filter and the connected to it parts.
Generally the OP-Amp should not react much to different and varying impedances between their inputs. Gross impedance imbalances may lead to excessive output offsets. An example for such a situation are DC-servos/Integrator where one input sees 0Ohm while the other may see 1MOhm. FET-Input devices are less affected in this regard, because of their much lower input currents. If the source impedances the OPsees exceed ~2kOhms, FET-Input devices may be less noisy due to their much lower current noise. The typical FET-Input candidates are also all low enough in voltage noise for usage in low source impedance circuits.
In most filter applications the bandwidth limit of all OPamps exceeds the requirements by far. Very fast OPamps need more care regarding the power supply lines (hence PCB-layout and associated parts) than ´slower´ devices. So there´s in majority no necessity for fast wide bandwidth parts.
So if You are willing to mix technologies bipolar input OPamps may be suitable certain cases. If You want one device for all a FET-input OPamp should be Your choice.
If cost is a factor FET-devices like OPA604, OPA132, OPA134, or AD711, AD825(featuring a OPamp unusual high OL-bandwidth of >10kHz) and bipolars like the NE5534 or NJM4580 and their Dual and Quad brethren are suitable.
If cost is no object FETs like BB/TIs OPA627, OPA827, AnalogDevices´s AD743, AD8610 come to mind as well as the bipolar AD797 and some of the LME-Series of NationalSemi.
jauu
Calvin
as usual the answer is...it depends.
In this case it depends on the type of filter and the connected to it parts.
Generally the OP-Amp should not react much to different and varying impedances between their inputs. Gross impedance imbalances may lead to excessive output offsets. An example for such a situation are DC-servos/Integrator where one input sees 0Ohm while the other may see 1MOhm. FET-Input devices are less affected in this regard, because of their much lower input currents. If the source impedances the OPsees exceed ~2kOhms, FET-Input devices may be less noisy due to their much lower current noise. The typical FET-Input candidates are also all low enough in voltage noise for usage in low source impedance circuits.
In most filter applications the bandwidth limit of all OPamps exceeds the requirements by far. Very fast OPamps need more care regarding the power supply lines (hence PCB-layout and associated parts) than ´slower´ devices. So there´s in majority no necessity for fast wide bandwidth parts.
So if You are willing to mix technologies bipolar input OPamps may be suitable certain cases. If You want one device for all a FET-input OPamp should be Your choice.
If cost is a factor FET-devices like OPA604, OPA132, OPA134, or AD711, AD825(featuring a OPamp unusual high OL-bandwidth of >10kHz) and bipolars like the NE5534 or NJM4580 and their Dual and Quad brethren are suitable.
If cost is no object FETs like BB/TIs OPA627, OPA827, AnalogDevices´s AD743, AD8610 come to mind as well as the bipolar AD797 and some of the LME-Series of NationalSemi.
jauu
Calvin
> If You want one device for all a FET-input OPamp should be Your choice.
My choice in any case.
With modern FET input opamps, the noise advantage of BJT opamps becomes insignificant at line level.
> If cost is no object FETs like BB/TIs OPA627, OPA827, AnalogDevices´s AD743, AD8610 come to mind ...
You should try something modern like the OPA1641/2 for FET and OPA1611/2 for BJT.
And they are not even expensive.
Patrick
My choice in any case.
With modern FET input opamps, the noise advantage of BJT opamps becomes insignificant at line level.
> If cost is no object FETs like BB/TIs OPA627, OPA827, AnalogDevices´s AD743, AD8610 come to mind ...
You should try something modern like the OPA1641/2 for FET and OPA1611/2 for BJT.
And they are not even expensive.
Patrick
If I were to use a Sallen Key with no gain, I basically only need a follower.
In such cases, I shall definitely use a pair of JFETs such as 2SK372 or 2SK369, as in :
http://www.diyaudio.com/forums/soli...-high-performance-source-follower-module.html
Patrick
In such cases, I shall definitely use a pair of JFETs such as 2SK372 or 2SK369, as in :
http://www.diyaudio.com/forums/soli...-high-performance-source-follower-module.html
Patrick
Patrick and all,
Thanks for bringing the OPA164x series to my attention. Looking at the specs, these are very nice opamps for a good price. Of course, specs are good, but they don't always tell the whole story.
Has anyone compared the OPA1641/1642 to OPA627 for how they sound? The OPA627 is pricey, but my choice for high quality signal path opamps. It would be nice to get an opinion or two about how the OPA1641 compares.
Thanks in advance.
Jac
What I've learned from building my active xovers, Sallen-Key configuration using ESP boards, is that the sound of any opamp is magnified by how many are in the circuit. Seems to make sense, but perhaps not readily apparent at first.
I initially used OPA2134 for all stages (input buffer, filter buffer, output buffer). Sounded very musical with good spatiality but a little dull (smooth, but just TOO smooth, somewhat lifeless) on top and flabby on the bottom. There was a bit of noise present.
Switched to all LM4562. Sound became much more detailed but thin with a bit of etched highs and seemed a bit less spacious; gave kind of a hi-fi impression instead of real music. Midrange did not have realistic tonality or body. Introduced listener fatigue. Bass was very well controlled. Very quiet.
Now I have LM4562 input and output buffers, with OPA2134 filter buffers. This seems to be the best combination, providing very full, spacious, smooth yet detailed sound, just a little looseness in the bass but thoroughly musical with superbly realistic midrange tonality and body. OPA2134 introduced a slight amount of noise, but only if listening very carefully right at the speaker with no music playing.
These all used the same power supply, with decent bypassing including a cap directly across the power pins of each opamp. Also put a cap across the feedback resistor of the output buffers to control oscillation.
So the answer, as I have found with so much in audio, seems to be that a combination of two (maybe more) different opamps is better than ANY single device. There is no reason to use only a single type or brand of resistor, capacitor, opamp, wire, etc when building our own equipment, and there may be good reasons for doing otherwise. Manufacturers do it for economy of scale. We are under no such constraints.
Peace,
Tom E
I initially used OPA2134 for all stages (input buffer, filter buffer, output buffer). Sounded very musical with good spatiality but a little dull (smooth, but just TOO smooth, somewhat lifeless) on top and flabby on the bottom. There was a bit of noise present.
Switched to all LM4562. Sound became much more detailed but thin with a bit of etched highs and seemed a bit less spacious; gave kind of a hi-fi impression instead of real music. Midrange did not have realistic tonality or body. Introduced listener fatigue. Bass was very well controlled. Very quiet.
Now I have LM4562 input and output buffers, with OPA2134 filter buffers. This seems to be the best combination, providing very full, spacious, smooth yet detailed sound, just a little looseness in the bass but thoroughly musical with superbly realistic midrange tonality and body. OPA2134 introduced a slight amount of noise, but only if listening very carefully right at the speaker with no music playing.
These all used the same power supply, with decent bypassing including a cap directly across the power pins of each opamp. Also put a cap across the feedback resistor of the output buffers to control oscillation.
So the answer, as I have found with so much in audio, seems to be that a combination of two (maybe more) different opamps is better than ANY single device. There is no reason to use only a single type or brand of resistor, capacitor, opamp, wire, etc when building our own equipment, and there may be good reasons for doing otherwise. Manufacturers do it for economy of scale. We are under no such constraints.
Peace,
Tom E
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.