Remember where you are measuring all voltages and signals from... its the zero point on the split supply. This is taken as an absolute reference from which everything is measured.
If you used two resistors then the voltage as measured from ground to the junction of these resistors would be a nominal zero volts providing certain conditions were met.
The plus and minus rails would have to be absolutely identical and the resistors would have to be identical in value too. If either rail altered then so to would the voltage at the resistor junction. Also any noise on the rail would be injected into the amplifier input... which you don't want.
Using a resistor to ground to bias and define the opamp input overcomes these problems. Noise on the rails now does not find its way into the amp.
Any imbalance in the rails has no effect on the amplifier output. The amp would be just as happy on a plus 12 minus 6 volt rail or a plus 8 minus 20 volts rail. The output would always be at zero volts. The only effect would be unsymetrical clipping at max outputs.
An input capacitor of appropriate value is always good practice because you never know what someone will connect to the input. Most signal sources are at zero volts DC but it can not be guaranteed.
Have you considered noise (hiss... white noise) and how you can minimise that... at least in theory.
Wonderful explanation 🙂 .. I read a little bit on white noise and I found that it is spread all along the hearing bandwidth, so my question is: how can I minimize it without effecting the gain of the audible bandwidth? If I filter with a sallen-key bandpass does it make sense? But whats the point of filtering unaudible frequencies.. except from 20Hz to 200Hz in this example? Does the white noise or hiss from 20Hz to 200Hz makes that much difference?
Will filtering unaudible frequencies save the op amp a lot of 'resources' and focus more precision on the interested frequencies?
Well I'm pleased it made sense...
Noise... which is ever present in electronic circuits can be minimised.
The larger (in ohms) a resistor is the more noise it generates. If you had a sensitive enough meter you could connect a resistor across it and actually measure the voltage produced, which would be wideband "white noise".
So keeping resistors low in value, for example in the feedback network helps keep the noise down.
How low... depends on the opamp and what it can drive. Most "audio" opamps can drive 600 ohms, so keeping the network around say 1 to 2 k would be a good target value. So the feedback resistors you calculated could usefully be reduced by a factor of 10.
Also the choice of opamp is important and that depends really on the impedances its being used at. FET opamps are a better choice for high impedances, bipolar for low impedances as a generalisation.
You wouldn't normally add a filter to a mic preamp such as this, but you should make sure the design is correctly compensated... and a little roll off at very high frequencies is no bad thing imo.
You might be interested in this on using opamps in practice,
http://www.diyaudio.com/forums/chip...g-audio-integrated-opamps-51.html#post2012422
There's another problem that won't affect this design if you stick to a FET device, and that is "output offset voltage" caused by non equal impedances at the two amplifier inputs. That affects bipolar opamps such as the NE5532 etc.
Noise... which is ever present in electronic circuits can be minimised.
The larger (in ohms) a resistor is the more noise it generates. If you had a sensitive enough meter you could connect a resistor across it and actually measure the voltage produced, which would be wideband "white noise".
So keeping resistors low in value, for example in the feedback network helps keep the noise down.
How low... depends on the opamp and what it can drive. Most "audio" opamps can drive 600 ohms, so keeping the network around say 1 to 2 k would be a good target value. So the feedback resistors you calculated could usefully be reduced by a factor of 10.
Also the choice of opamp is important and that depends really on the impedances its being used at. FET opamps are a better choice for high impedances, bipolar for low impedances as a generalisation.
You wouldn't normally add a filter to a mic preamp such as this, but you should make sure the design is correctly compensated... and a little roll off at very high frequencies is no bad thing imo.
You might be interested in this on using opamps in practice,
http://www.diyaudio.com/forums/chip...g-audio-integrated-opamps-51.html#post2012422
There's another problem that won't affect this design if you stick to a FET device, and that is "output offset voltage" caused by non equal impedances at the two amplifier inputs. That affects bipolar opamps such as the NE5532 etc.
I learned more than my first year of my EE course from your posts 😉 and i will look into the link. thanks
When you mentioned the compensation, are talking about the stablity of the op amp and if so how can i make sure there is no oscillations?
Are their any good reasons or benefits from a low pass filter that would give us a little roll of at the high frequencies?
When you mentioned the compensation, are talking about the stablity of the op amp and if so how can i make sure there is no oscillations?
Are their any good reasons or benefits from a low pass filter that would give us a little roll of at the high frequencies?
Well, it depends- are you trying to design a really excellent mike preamp or one that just meets the outlined specs? If the former, you'd want to block noise (from CFLs, radio stations, TV sets) that's above the audible range but could produce audible distortion.
I apologize to break one of you critical rules or beliefs but I have to simulate this preamp 😛 .. how will higher frequency distort the audio frequency in the op-amp. I want to explain a little briefly a least about this issue.
Will a simple low pass filter do the trick.. without going into multiple poles for high q factor etc.. I only need a pre amp with the specified requirements..
the last bit asks me to modify the circuit to work with a single +5V supply.. will this limit me a lot in the range of op amps that i can use.
thanks
bdw nice shots of op amp oscillations etc. it's all Gibb's Phenomena is it ?
Will a simple low pass filter do the trick.. without going into multiple poles for high q factor etc.. I only need a pre amp with the specified requirements..
the last bit asks me to modify the circuit to work with a single +5V supply.. will this limit me a lot in the range of op amps that i can use.
thanks
bdw nice shots of op amp oscillations etc. it's all Gibb's Phenomena is it ?
No, not really. You see Gibb's phenomena when the filter applied is near-brick-wall, i.e., very steep slopes. The ringing you see when the compensation is missing or inadequate is a manifestation of positive feedback, i.e., you're running the amplifier near the point of oscillation.
Simulation is fine... but when you actually build a design you also have to prove by measurement and test that is performs as expected.
Compensation as you can see is quite critical to get right, as we are talking only a few pf of capacitance sometimes. On a sine wave input an incorrectly compensated amp still appears to behave correctly... there is no distortion to see... its when a step function such as a squarewave is applied that the problems show. Now it can be argued that a squarewave has little basis in music and audio and that nothing in music ever approaches anything that severe, but what the squarewave does tell us is how well behaved the design is. If there is severe ringing for example you are one short step away from total instability where the thing just oscillates on its own.
I would try and set the compensation so that it gives say a -3db point at 30 khz (or whatever you wish) so that the response is nicely tailored and falls with increasing frequency... and explain your reasons for doing so.
Sy mentions interference caused by radio transmitters/tv timebases etc... all very real problems. These problems cause various non linear forms of distortion, in the case of the radio transmitter the signal can actually be demodulated by the input transistors of the opamp and heard as audio... an old problem with much home built gear. A FET opamp wouldn't do that in the same way though... but if you are designing something as a serious full blown project then it pays to attend to all these things at the design stage. In the case of radio transmitters a ferrite bead slipped over a resistor leg or wire feeding the input to the opamp is sufficient to stop the interference.
5 volt operation... thats more of a challenge... and would need your two resistor bias network at the very least.
So I would think outside the box and argue that performance would suffer (lack of headroom on a restricted supply) and so use a DC/DC convertor to give -/+12 volts from a 5 volt supply.... offer that as one example 🙂
To make this circuit work on 5 volts would require careful choice of opamp... it can be done.
Compensation as you can see is quite critical to get right, as we are talking only a few pf of capacitance sometimes. On a sine wave input an incorrectly compensated amp still appears to behave correctly... there is no distortion to see... its when a step function such as a squarewave is applied that the problems show. Now it can be argued that a squarewave has little basis in music and audio and that nothing in music ever approaches anything that severe, but what the squarewave does tell us is how well behaved the design is. If there is severe ringing for example you are one short step away from total instability where the thing just oscillates on its own.
I would try and set the compensation so that it gives say a -3db point at 30 khz (or whatever you wish) so that the response is nicely tailored and falls with increasing frequency... and explain your reasons for doing so.
Sy mentions interference caused by radio transmitters/tv timebases etc... all very real problems. These problems cause various non linear forms of distortion, in the case of the radio transmitter the signal can actually be demodulated by the input transistors of the opamp and heard as audio... an old problem with much home built gear. A FET opamp wouldn't do that in the same way though... but if you are designing something as a serious full blown project then it pays to attend to all these things at the design stage. In the case of radio transmitters a ferrite bead slipped over a resistor leg or wire feeding the input to the opamp is sufficient to stop the interference.
5 volt operation... thats more of a challenge... and would need your two resistor bias network at the very least.
So I would think outside the box and argue that performance would suffer (lack of headroom on a restricted supply) and so use a DC/DC convertor to give -/+12 volts from a 5 volt supply.... offer that as one example 🙂
To make this circuit work on 5 volts would require careful choice of opamp... it can be done.
The +/- 12V dual supply amplifier will be built on a breadboard and it will be measured with an oscilloscope. Then input will be from a signal generator with sine wave output and from the measurements we will plot a gain frequency response from 200Hz to 20kHz and the DC voltages, while in the simulation I have to find:
dc voltages
mid-band gain @ 1kHz
Lower cut off frequency
gain frequency response
and Phase frequency response.
When it comes to dc voltages I am not sure why it is required as the configuration we talked about doesn't include much dc biasing what do you think. I will simulate it and upload the schematic with graphs if it's not a problem. I am finding your experiences with audio very interesting, tnx
dc voltages
mid-band gain @ 1kHz
Lower cut off frequency
gain frequency response
and Phase frequency response.
When it comes to dc voltages I am not sure why it is required as the configuration we talked about doesn't include much dc biasing what do you think. I will simulate it and upload the schematic with graphs if it's not a problem. I am finding your experiences with audio very interesting, tnx
Thats great that your building it as well as simulating... I'll come back to you in a little while, tea time 🙂
I found a good template at this place:
Analog Devices : Analog Dialogue : Amplifier Circuits
Its in figure 2.. what do you think?
If you can put its description in easier words for me I would appreciate.
Its not only the dual supply... The problem attached.
Analog Devices : Analog Dialogue : Amplifier Circuits
Its in figure 2.. what do you think?
If you can put its description in easier words for me I would appreciate.
Its not only the dual supply... The problem attached.
Figure 2 is just a classic text book non inverting opamp configuration...
Can you visualise how feedback works and how it sets the gain of the opamp so precisely... has that been explained to you ?
Can you visualise how feedback works and how it sets the gain of the opamp so precisely... has that been explained to you ?
Yes I think so.. although it hadn't been explained so well unfortunately.. its all with the help of some good books in 3 three day that I managed to figure out something. The process is all about sampling the output to correct the input as necessary, how is that for a short and sweet description 😉
Does this make sense ?
Look at your simple opamp diagram in post #1
Forget AC signals and think about its operation at DC...
Remember the golden rule that "the output will do whatever is required to ensure the difference in voltage at the inputs is zero".
So lets make R2 9000 ohms and R1 1000 ohms.
If we apply 1 volt to the non inverting input, what happens to the output ? In order to maintain the "difference between the inputs" to equal zero (thats not the same as saying the inputs are at zero... its the difference) we need the output of the opamp to become 10 volts. Using ohms law you can see that 9 volts is across the 9K resistor and 1 volt across the 1 k. So that brings the inverting input up to 1 volt, the same as the non inverting (our signal input). So the input difference becomes zero, both inputs are now at 1 volt and the output is at 10 volts. We have an amplifier with a voltage gain of 10.
That applies for AC signals too, as an AC signal at any instant in time can be thought of as a just a DC voltage of some particular value.
Look at your simple opamp diagram in post #1
Forget AC signals and think about its operation at DC...
Remember the golden rule that "the output will do whatever is required to ensure the difference in voltage at the inputs is zero".
So lets make R2 9000 ohms and R1 1000 ohms.
If we apply 1 volt to the non inverting input, what happens to the output ? In order to maintain the "difference between the inputs" to equal zero (thats not the same as saying the inputs are at zero... its the difference) we need the output of the opamp to become 10 volts. Using ohms law you can see that 9 volts is across the 9K resistor and 1 volt across the 1 k. So that brings the inverting input up to 1 volt, the same as the non inverting (our signal input). So the input difference becomes zero, both inputs are now at 1 volt and the output is at 10 volts. We have an amplifier with a voltage gain of 10.
That applies for AC signals too, as an AC signal at any instant in time can be thought of as a just a DC voltage of some particular value.
Things are much more clear now 🙂 ... I simulated the circuit and I got some interesting results.. (I've been longing to see) They are attached. I have a slight phase shift as you can see at the 1Khz mid band gain because of the filters i added. I will surely comment on them.
Got your images now... can't get them to open off the site, had to download them... strange.
Are you using the OPA2134 and the resistor values you calculated ? And is it just the basic circuit of opamp and resistors ? with you mentioning filters.
I'm calling it a night here 🙂 I'll look in tomorrow.
Pleased that has helped you understand it better anyway.
Are you using the OPA2134 and the resistor values you calculated ? And is it just the basic circuit of opamp and resistors ? with you mentioning filters.
I'm calling it a night here 🙂 I'll look in tomorrow.
Pleased that has helped you understand it better anyway.
And no current flows into the opamp inputs (usually so small it can be ignored). Which means all the current that flows "towards" the input keeps going if it has a pathway.
Applying these two simple rules you can figure out 90% of analogue opamp circuits.
Last edited:
- Status
- Not open for further replies.