I wouldn't say most opamps are FET, there is a good choice of either. Z denotes impedance, the "resistance" of something at AC. Gate (not base) currents in FET's are negligible at DC. At AC there is some current flow as the gate capacitance has to be charged and discharged to follow the AC signal. That varies from small in an opamp to quite significant for a power FET.
Try and understand how opamps works
(Simplified again )
The golden rule is that an opamp with feedback (so that's all the configurations you use for audio) will make the output of the opamp do whatever is necessary to bring the voltage difference between the two inputs to zero. So from that one statement (which isn't even a formula) you can calculate the gain and what is actually going on for any given circuit.
Now imagine a bjt opamp like a 5532. The data sheet says it has an input bias current of 200 nano amps typically and worst case 800 nano amps. Lets work with the worst case figure. It means that 800na flows out of the opamp input pins, and this is the base bias current. If you had a resistor of say 100k connecting that opamp pin to ground then that 800na would cause the opamp pin to become 0.08 volts (80 millivolts) away from ground. Just ohms law. For our statement about the opamp keeping the difference between the inputs at zero to work then the opamp output has to take on a relatively high DC voltage to force the other input (via the feedback network) to equal that 80mv... yes and that voltage at the output of the opamp is what we see and call the DC offset. For example, if the feedback network were a 10k and 1k then you would need 0.88 volts from the opamp to push enough current through that network to give our 80 mv at the opamp to equal and cancel the "error" on the other input pin. Ohms law again. 10k +1k (which is 11k) needs 0.88 volts across it to develop 0.80 mv across the 1k.
The FET opamp has zero input bias current and so even a 100 Meg ohm resistor would not see any voltage developed across it and so there would be no DC offset to correct.
Try and understand how opamps works
(Simplified again )The golden rule is that an opamp with feedback (so that's all the configurations you use for audio) will make the output of the opamp do whatever is necessary to bring the voltage difference between the two inputs to zero. So from that one statement (which isn't even a formula) you can calculate the gain and what is actually going on for any given circuit.
Now imagine a bjt opamp like a 5532. The data sheet says it has an input bias current of 200 nano amps typically and worst case 800 nano amps. Lets work with the worst case figure. It means that 800na flows out of the opamp input pins, and this is the base bias current. If you had a resistor of say 100k connecting that opamp pin to ground then that 800na would cause the opamp pin to become 0.08 volts (80 millivolts) away from ground. Just ohms law. For our statement about the opamp keeping the difference between the inputs at zero to work then the opamp output has to take on a relatively high DC voltage to force the other input (via the feedback network) to equal that 80mv... yes
and that voltage at the output of the opamp is what we see and call the DC offset. For example, if the feedback network were a 10k and 1k then you would need 0.88 volts from the opamp to push enough current through that network to give our 80 mv at the opamp to equal and cancel the "error" on the other input pin. Ohms law again. 10k +1k (which is 11k) needs 0.88 volts across it to develop 0.80 mv across the 1k. The FET opamp has zero input bias current and so even a 100 Meg ohm resistor would not see any voltage developed across it and so there would be no DC offset to correct.
i see, so working with bringing the voltage difference between the two inputs to zero
assume a bjt opamp like the infamous lm741 at temperatures of 25 celsius, it has a worst case input bias current of 500nA, and say i have a 5.7k input resistor and 10k feedback resistor, then the parallel resistance of the resistors (which is 3.63k) would cause the op amp pin to become (ohms law as below):
v = 0.0000005*3630
v = 0.001815
would cause the op amp pin to become 1.815 mV away from ground from the negative side (aka -1.815mV away from ground) since the feedback loop is at the negative input, correct?
hence i would need to put a resistor rated at 3.63k at the positive input so that the op amp would again go 1.815mV "away" from ground from the positive side, hence 1.815-1.815 cancels each other out and i end up with a ground that is at 0v (or whatever voltage the virtual ground is at if i use one)
to conclude. since fet op amps have 0 output current bias, using ohms law
v = 0*anything = 0, hence there will never have any dc offset correct?
thanks for all the help along the way, they are all very informative
EDIT: forgot to mention, so deducing from the above conclusion, would it be fair to say that fet op amps are superiour at least on paper versus bjt op amps? since using the same input and feedback resistors, the bjt op amps will always have higher stage impedance due to the need for a "balancing" resistor at ground
assume a bjt opamp like the infamous lm741 at temperatures of 25 celsius, it has a worst case input bias current of 500nA, and say i have a 5.7k input resistor and 10k feedback resistor, then the parallel resistance of the resistors (which is 3.63k) would cause the op amp pin to become (ohms law as below):
v = 0.0000005*3630
v = 0.001815
would cause the op amp pin to become 1.815 mV away from ground from the negative side (aka -1.815mV away from ground) since the feedback loop is at the negative input, correct?
hence i would need to put a resistor rated at 3.63k at the positive input so that the op amp would again go 1.815mV "away" from ground from the positive side, hence 1.815-1.815 cancels each other out and i end up with a ground that is at 0v (or whatever voltage the virtual ground is at if i use one)
to conclude. since fet op amps have 0 output current bias, using ohms law
v = 0*anything = 0, hence there will never have any dc offset correct?
thanks for all the help along the way, they are all very informative
EDIT: forgot to mention, so deducing from the above conclusion, would it be fair to say that fet op amps are superiour at least on paper versus bjt op amps? since using the same input and feedback resistors, the bjt op amps will always have higher stage impedance due to the need for a "balancing" resistor at ground
Last edited:
again fets do have bias current requirement - may be picoAmps if cool but input current is still needed to bias a fet
input V offset also comes from mismatching of input devices - fets are generally worse than bjt, newer semi fab processing, internal trimming can give lower Vos if you look for devices with this spec
input V offset also comes from mismatching of input devices - fets are generally worse than bjt, newer semi fab processing, internal trimming can give lower Vos if you look for devices with this spec
the difference between “is 0” and “is very small” is a factor of infinity
which is important in explaining circuit principles to obvious beginners which may have a very spotty understanding of circuits - may take your "is 0" as an absolute truth rather than as a approximation suitable for just one particular calculation
it can also be a disservice to just answer beginners exact questions when bigger issues with the circuit, variables important to its operation are not brought up by the beginner
which is important in explaining circuit principles to obvious beginners which may have a very spotty understanding of circuits - may take your "is 0" as an absolute truth rather than as a approximation suitable for just one particular calculation
it can also be a disservice to just answer beginners exact questions when bigger issues with the circuit, variables important to its operation are not brought up by the beginner
OK, fair enough to the first point but I find the last remark slightly strange tbh, given the sometimes quite comprehensive posts and follow ups I give to some topics.
Perhaps you could take the time to offer a comprehensive and easily read breakdown of your earlier post,
perhaps even posting a circuit that attends to all the pitfalls awaiting the beginner attempting to build such a circuit.
I am on a phone right now so I may not be able to type in detail and the numbers bight be off
I read the datasheet on opa2227, opa2134, lm741 and 5532
worst case input bias are 10nA, 5nA, 500nA and 800nA
so for worst case input bias, the bias difference of fets vs bjt is about 1.5 order of magnitude
now I just wonder how often are the input biases at their worst case for the respective op amps
I looked at the input bias(in picoamp a) vs temperature(C) graph of the opa2134 and it shows a dpA/dC of about 10 pA per celsius after temperatures are above 26 degrees C. if temperature is below 26 C then the input bias is about 8-10 pA
the graph of the opa2227 shows dpA/dC dropping as temperature increases, so I don't really understand it
I can't find the graph of lm741 or ns5532 so I can't really compare them (well at least I can't find them easily on a phone)
but having made some numbers of cmoys for fun, I seriously doubt that an opa2134 is going to reach 125C often if at all (because if it does then something is probably horribly wrong) to get the worst case input bias current, and I should also add that temperature matters a lot at least for opa2134's case, I will assume the same for opa2227/8 and probably other fet type op amps as well, but I am really not certain about te bjt op amps
what I really don't know (and would like to know) is how significant is the offset going to be and how much noise would be generated by a set amount of offset, and following that, when will the extra impedance from the "balancing" resistor at ground be worth it/not worth it (like a break even point where the input bias with a set amount of gain is going to be significant enough to justify adding or omitting the resistor at ground)
sorry for messy typing, I will probably edit this when I get home, regardless, all of you have been a great help for me and it has be very educational and informative, I think I will really need to build something to learn about practical application of op amps since not everything is like theory and there are no ideal op amps
I read the datasheet on opa2227, opa2134, lm741 and 5532
worst case input bias are 10nA, 5nA, 500nA and 800nA
so for worst case input bias, the bias difference of fets vs bjt is about 1.5 order of magnitude
now I just wonder how often are the input biases at their worst case for the respective op amps
I looked at the input bias(in picoamp a) vs temperature(C) graph of the opa2134 and it shows a dpA/dC of about 10 pA per celsius after temperatures are above 26 degrees C. if temperature is below 26 C then the input bias is about 8-10 pA
the graph of the opa2227 shows dpA/dC dropping as temperature increases, so I don't really understand it
I can't find the graph of lm741 or ns5532 so I can't really compare them (well at least I can't find them easily on a phone)
but having made some numbers of cmoys for fun, I seriously doubt that an opa2134 is going to reach 125C often if at all (because if it does then something is probably horribly wrong) to get the worst case input bias current, and I should also add that temperature matters a lot at least for opa2134's case, I will assume the same for opa2227/8 and probably other fet type op amps as well, but I am really not certain about te bjt op amps
what I really don't know (and would like to know) is how significant is the offset going to be and how much noise would be generated by a set amount of offset, and following that, when will the extra impedance from the "balancing" resistor at ground be worth it/not worth it (like a break even point where the input bias with a set amount of gain is going to be significant enough to justify adding or omitting the resistor at ground)
sorry for messy typing, I will probably edit this when I get home, regardless, all of you have been a great help for me and it has be very educational and informative, I think I will really need to build something to learn about practical application of op amps since not everything is like theory and there are no ideal op amps
Lets just pick two of them. The OPA2134 has an input bias current of typically 5pa (Picoamps), the NE5532 is around 200na or around 40,000 times higher. The FET devices do show a very rapid rise at high temperatures but that's probably not a realistic scenario for the use we put them too.
If you set up a breadboard so you can switch between an OPA and the NE5532 for example then you will see for real how offset works and how to minimise it by equalising the input bias currents. Many single opamps have an "offset null" facility which allows a trimmer to be connected and for the user to trim the output to zero. That is intended to trim each individual error though, and not to make good an offset that appears due to other connected devices causing an offset etc.
A small offset isn't necessarily a problem but it can cause thumps and clicks when devices with an offset are connected and disconnected from say an amplifier input. Or in the case of a CMOY, when you plug the headphones in. An offset in itself doesn't increase the distortion.
So I would say to rig up an inverting and a non inverting amp on a breadboard and then you can see for real what happens and how different values of resistor affect things. It will all make sense if you do that
If you set up a breadboard so you can switch between an OPA and the NE5532 for example then you will see for real how offset works and how to minimise it by equalising the input bias currents. Many single opamps have an "offset null" facility which allows a trimmer to be connected and for the user to trim the output to zero. That is intended to trim each individual error though, and not to make good an offset that appears due to other connected devices causing an offset etc.
A small offset isn't necessarily a problem but it can cause thumps and clicks when devices with an offset are connected and disconnected from say an amplifier input. Or in the case of a CMOY, when you plug the headphones in. An offset in itself doesn't increase the distortion.
So I would say to rig up an inverting and a non inverting amp on a breadboard and then you can see for real what happens and how different values of resistor affect things. It will all make sense if you do that
I can't say I like Mancinni Op Amps for Everyone - but it is free
Wallt Jung's Op Amp Applications tome has some good parts, still not the best learning tool
there's the Philbrick manual, particularly http://www.analog.com/library/analogdialogue/archives/philbrick/013-019.pdf
but basically I don't have good recommendation to get someone starting out up to the level of appreciating Samuel Groner's work http://www.sg-acoustics.ch/analogue_audio/ic_opamps/
Wallt Jung's Op Amp Applications tome has some good parts, still not the best learning tool
there's the Philbrick manual, particularly http://www.analog.com/library/analogdialogue/archives/philbrick/013-019.pdf
but basically I don't have good recommendation to get someone starting out up to the level of appreciating Samuel Groner's work http://www.sg-acoustics.ch/analogue_audio/ic_opamps/
Last edited:
sorry for going invisible for a fortnight. college and visa problems got in my way (my F-1 visa for staying in the US had some problems and i almost got deported for that)
i actually did build a simple circuit to test out inverting and non-inverting designs as well as running two inverting op amps in series at the weekend, heres a picture of my breadboard
in the picture, i used a TL072 jellybean because i have tons of them lying around the place, the circuit is only a one channel circuit with two op amps, since i do not need stereo for testing the sound quality
the results are very interesting, i will call the first op amp in series A and the second one B, hence the output from the first op amp is A(out) and second one B(out) and etc
A(out) is connected to B(in), both A and B have an inverting feedback loop that has a gain of 2, hence gain at A(out) is 2 whilst gain at B(out) is 4
since the TL072 is a JFET, it has quite a low input bias current, however, just for plain curiosity, i placed some "neutralising" resistors (which is 3.5K ohms which is approximately the parallel resistance of my input and feedback resistors) connected to ground and non-inverting input of the op amps. the resistors actually gave a not-so-significant but very noticeable difference in terms of noise reduction
the results are very interesting, when listening from A(out), it is very "run of the mill" and sounds just like what i would have expected, however, when listening from B(out), the sound is really, really bad (i thought it would have sounded better but apparently i am wrong)
another interesting find is when i have B(out) connected to something that has a very high resistance/impedance (such as my body) and then connecting the high impedance object to the headphone input, i could still listen to the music very clearly (i had my finger touching the alligator clip and my other finger touching the input of my headphone) whilst when i do the same for A(out), the sound is significantly muffled, what is this effect called?
i actually did build a simple circuit to test out inverting and non-inverting designs as well as running two inverting op amps in series at the weekend, heres a picture of my breadboard
in the picture, i used a TL072 jellybean because i have tons of them lying around the place, the circuit is only a one channel circuit with two op amps, since i do not need stereo for testing the sound quality
the results are very interesting, i will call the first op amp in series A and the second one B, hence the output from the first op amp is A(out) and second one B(out) and etc
A(out) is connected to B(in), both A and B have an inverting feedback loop that has a gain of 2, hence gain at A(out) is 2 whilst gain at B(out) is 4
since the TL072 is a JFET, it has quite a low input bias current, however, just for plain curiosity, i placed some "neutralising" resistors (which is 3.5K ohms which is approximately the parallel resistance of my input and feedback resistors) connected to ground and non-inverting input of the op amps. the resistors actually gave a not-so-significant but very noticeable difference in terms of noise reduction
the results are very interesting, when listening from A(out), it is very "run of the mill" and sounds just like what i would have expected, however, when listening from B(out), the sound is really, really bad (i thought it would have sounded better but apparently i am wrong)
another interesting find is when i have B(out) connected to something that has a very high resistance/impedance (such as my body) and then connecting the high impedance object to the headphone input, i could still listen to the music very clearly (i had my finger touching the alligator clip and my other finger touching the input of my headphone) whilst when i do the same for A(out), the sound is significantly muffled, what is this effect called?
Last edited:
Perhaps head phone load at gain = 4 exceeds the output rating, unless you have a finger (around 1k at that voltage) in series with the load.
And now you know why it is good to minimise as well as balance the input resistances to op amps. Note that while fet inputs have high DC resistance they also have large input capacitance, it is the capacitive current with AC signals that acts like an input bias current and causes amplitude distortions when unbalanced rather than DC offsets.
I hope you are now learning to carefully evaluate what people here say.
And now you know why it is good to minimise as well as balance the input resistances to op amps. Note that while fet inputs have high DC resistance they also have large input capacitance, it is the capacitive current with AC signals that acts like an input bias current and causes amplitude distortions when unbalanced rather than DC offsets.
I hope you are now learning to carefully evaluate what people here say.
While you have had lots of advice on resistor values around op-amps no one has mentioned the power supply.
Make sure you use star grounding or you will get massive hum.
I threw together a pcb for a 6 channel mixer and when I built it the hum was at an unbearable level.
I found the hum was coming from smoothing capacitors charging up and causing a pulse on the ground line.
I re-laid out the pcb with the power supply separated and it worked a treat.
Make sure you use star grounding or you will get massive hum.
I threw together a pcb for a 6 channel mixer and when I built it the hum was at an unbearable level.
I found the hum was coming from smoothing capacitors charging up and causing a pulse on the ground line.
I re-laid out the pcb with the power supply separated and it worked a treat.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.