| Electronics > Projects, Designs, and Technical Stuff |
| Help needed for simple voltage follower requirements |
| (1/1) |
| Frederico Acardi:
Hello my name is Frederico. I am designing my perfect voltage follower opamp circuit, for audio frequency bandwith, BUT... I am new to opamps :-DD It is for driving a 10K load (and at most 2K). The main objective is to add the lowest amount of noise, unitygain stable without adding parts (everything else I do not care, even how it sounds), parallel circuit is ok as long as it does not use more than dual package (cost not an issue, as long as it's under 3€). I need some informations (for lowest noise) : - Parallel or single ? --> keep in mind that parallel opamps use serie resistor (=noise) + higher current noise. - BJT (lower noise) or FET (lower input current) ? --> I am lost - anything else ? Do you know some parts I can use to get the result I am after ? What do I have to look for a perfect opamp buffer ? What are the specs I have to look for in the datasheets ? I am considering BJT input at the moment (NJM2068/NJM2043/NJM-RC4580/NJM-NE4562) the low noise versions D and A. How do they compare the LT1128 ? Which seems to be perfect but cost almost 10€. Thank you. |
| David Hess:
Noise depends on the input impedance and the ratio between input voltage noise and input current noise. So for high source impedance, the total noise is lower with JFET or MOSFET inputs. For low source impedance, total noise will be lower with bipolar input parts which have lower voltage noise but higher current noise. A part like the LT1128 is only suitable when the input impedance is about 400 ohms or less. The OP-27 or LT1007 which are similar to the parts you mention are more suitable for most audio applications with input impedances of a couple kilohms and this applies to the early eponymous low distortion and low noise bipolar audio parts like the LM833 and NE5532. You might use the LT1128 as a magnetic cartridge amplifier, with a transformer coupled input, or in a 50 to 110 ohm system. Otherwise it is better to choose one of the parts you mentioned. Low noise JFET and CMOS devices tend to have large non-linear gate capacitance at their inputs which contributes to distortion in follower and non-inverting gain applications unless properly handled but with some attention to this detail, they can be competitive with the above parts at higher input impedances. |
| Frederico Acardi:
Thanks to your comment on impedance, I need to define the output impedance of the device I plug into the opamp. Ohm's law again ; current in a R =V. It is still new to me, how opamps are choosen for lowest noise. Should we discard "Input resistance" in the datasheet and rely on ohm's law and the current data ? Let's say I have 2 microphones. One with 500KΩ impdeance and another with 500Ω. Both have a output of 0.4V : - Input Impedance : the LT1007 has 10nA input current =0.4V/10nA=40MΩ input impedance ? - Noise with the LT1007, the 500KΩ mic will have = 500000Ω*10nA =5mV of noise ? and the 500Ω*10nA =0.005mV ? I always read opamps have very high input impedance, so I tough it was very very high. But on the NJM2043/68 (and most BJT from TI) it's only around 30/300KΩ. Now I think I will build 2 inputs : - 1 with a Hi Z opamp (for instruments) = Maybe NJM4562 or LT1007 or OPA1641 - and 1 with Low Z opamp (for microphones) = LT1128 cheaper alternative Does a parallel opamps circuit, has halved input impedance ? That is why it's less noisy, not because of the uncorrelated noise vs the correlated signal ? Is it relevant in a voltage follower in parallel, since there is no gain and the signal is split between opamps (=lower signal =closer to noise floor) Parallel might be even detrimental to noise-to-signal ratio in a V follower configuration ? Thank you. |
| David Hess:
I am sorry for taking so long to respond. It is difficult to talk about operational amplifier noise without getting too complicated. The following Linear Technology design notes cover some of the basics for operational amplifier selection and math. https://www.analog.com/media/en/reference-design-documentation/design-notes/dn015f.pdf https://www.analog.com/media/en/reference-design-documentation/design-notes/dn140f.pdf https://www.analog.com/media/en/reference-design-documentation/design-notes/dn355f.pdf Analog Devices also has several good papers discussing operational amplifier noise. --- Quote from: Frederico Acardi on February 01, 2019, 09:32:48 am ---Thanks to your comment on impedance, I need to define the output impedance of the device I plug into the opamp. Ohm's law again ; current in a R =V. It is still new to me, how opamps are choosen for lowest noise. Should we discard "Input resistance" in the datasheet and rely on ohm's law and the current data ? --- End quote --- The input resistance specification has to do with the change in bias current versus voltage and has nothing to do with noise. It is mostly useless but has some applicability to DC performance of non-inverting amplifiers. --- Quote ---Let's say I have 2 microphones. One with 500KΩ impdeance and another with 500Ω. Both have a output of 0.4V : - Input Impedance : the LT1007 has 10nA input current =0.4V/10nA=40MΩ input impedance ? - Noise with the LT1007, the 500KΩ mic will have = 500000Ω*10nA =5mV of noise ? and the 500Ω*10nA =0.005mV ? --- End quote --- Check the design notes I linked above. The rule I start with is actually wrong because it does not take into account Johnson noise from the resistance of the source and some other things like input bias current cancellation but it is close enough. The source resistance (or impedance) converts the input current noise into voltage noise which adds to the operational amplifier's voltage noise. So the 1.5pA/SqrtHz current noise from an LT1007 multiplies with 500 kilohms to produce a massive 750nV/SqrtHz compared to its voltage noise of 2.8nV/SqrtHz; obviously a different part should be used. With a 500 ohms source, this becomes a very reasonable 0.75nV/SqrtHz. Optimum would be about 1.8 kilohms. But pick a JFET part which would be suitable for audio like the JFET input LT1169 and things are different. 1fA/SqrtHz through 500 kilohms is 0.5nV/SqrtHz which is much lower than the LT1169's low voltage noise for a JFET input of 6nV/SqrtHz. Optimum source resistance for noise purposes is more like 6 megohms which is way beyond the point where parasitic capacitance is going to dominate performance. --- Quote ---I always read opamps have very high input impedance, so I tough it was very very high. But on the NJM2043/68 (and most BJT from TI) it's only around 30/300KΩ. --- End quote --- Operational amplifiers have high input impedance but that has nothing to do with noise performance. Most bipolar input audio operational amplifiers are about a couple of kilohms. --- Quote ---Now I think I will build 2 inputs : - 1 with a Hi Z opamp (for instruments) = Maybe NJM4562 or LT1007 or OPA1641 - and 1 with Low Z opamp (for microphones) = LT1128 cheaper alternative --- End quote --- A suitable operational amplifier for low noise audio at high impedances will have JFET or MOSFET inputs for low current noise. The LT1128 is suitable for very low impedances. The LT1007 or a more traditional bipolar audio amplifier would be much better unless you are dealing with 50 to 300 ohm sources. Incidentally, bipolar audio operational amplifiers do *not* use input bias current cancellation because it adds noise (the LT1128 and LT1007 suffer from this) and generally have PNP inputs because for reasons I have forgotten, PNP input stages have lower noise. --- Quote ---Does a parallel opamps circuit, has halved input impedance ? That is why it's less noisy, not because of the uncorrelated noise vs the correlated signal ? Is it relevant in a voltage follower in parallel, since there is no gain and the signal is split between opamps (=lower signal =closer to noise floor) Parallel might be even detrimental to noise-to-signal ratio in a V follower configuration ? --- End quote --- Paralleling amplifiers lowers voltage noise but increases current noise so the optimum impedance decreases. This might be done for very low source impedances or to lower the voltage noise of a JFET or MOSFET input amplifier where the current noise is already very low but it can cause problems with increased input capacitance. A major source of distortion in non-inverting JFET and MOSFET amplifiers is the change in input capacitance as the common mode voltage (voltage at the non-inverting input) changes. |
| Navigation |
| Message Index |