Noninverting to inverting op amp

I am trying to change a noninverting op amp of an existing circuit so that it is inverting. Since I am just starting out with op amps I am confused as to why simply switching the inputs does not work- I am not getting any output in that case.

The existing circuit uses a dual NE5532 with noninverting input at pin

  1. The signal is DC biased to 50% Vcc, i.e. 2.5V and coupled to an electret microphone. The inverting pin 2 is connected to GND via a 10k resistor and a 10uF capacitor. The inverting input of course also has a feedback connection consisting of a 47k resistor and a 100pF capacitor in parallel connected to the output. This arrangement works fine- switching pins 2 & 3 results in silence!?
Reply to
grubertm
Loading thread data ...

Because when you switch the pins you area also switching the feedback? The feedback should always be negative for amplification.

What you want to do is not switch the pins but put the signal on the inverting side. That is, ground your non-inverting pin and attach your input to the inverting pin instead.

Its probably better to see what than for me to explain:

formatting link

go there and look at the difference between the two and you'll see that the only difference is where the input is attached. If you are physically switching the pins then your putting the feedback onto the non-inverting and getting positive feed back... essentially making a comparator.

Jon

Reply to
Jon Slaughter

They don't work like that. You need to learn some op-amp basics.

Why do you care if a microphone amplifier is inverting or non-inverting?

You could swap the electret microphone and the 10uF capacitor. This would give you an inverting amplifier with a gain of 4.7 instead of the 5.7 you have now. The 10uF capacitor would then be a decoupling capacitor on pin 3, and the 10k resistor would bias the microphone and be its load impedance. The only problem is: 10k might be a bit high for the load impedance. In your original circuit, the load seen by the microphone is the parallel combination of the DC bias resistors.

Reply to
Andrew Holme

Circa Mon, 7 May 2007 11:01:20 +0100 recorded as looks like "Andrew Holme" sounds like:

Maybe he wants to do upside-down voice recording, kinda like the Beatles and ELO did backwards and slowed-down effects. Maybe that's it, huh? Maybe, if he reverses the feedback resistor, he can get backwards effects?!

On the serious side, OP should also be aware that overall gain will change, if he manages to get the thing working at all. The gain now is 1 + Rf/Ri, or 5.7, it will change to Rf/Ri, or 4.7. To keep the same gain, Rf will need to be changed to, say, 56K.

It's not a trivial design modification, and like you, I do wonder why OP wants to make the change. What's the circuit board going to look like when all the experimentation is over? Visions of melting and burning and smoke and solder splash come to mind.

Reply to
Charlie Siegrist

It shouldn't matter why he wants to do it. Its not for you or anyone else to judge. Let him make his mistakes so he can learn. You also don't know his full situation so maybe he has some reason.

Reply to
Jon Slaughter

People ask relatively clueless questions all the time. They are clueless because they lack context, and without that context a good answer often can't be provided.

So somebody asks something, and then if they ever post a second time, it's deep down in the thread, and it turns out what they thought they needed wasn't, so the thread up till that point is a waste. They got answers to what they asked, but they made an assumption about what it wsa they needed, and that was wrong, so they actually need some other solution. Or, they don't have enough history in electronics, so their guess at the solution, and the literal answers they get, turns out to be a less than optimal (at the very least) solution.

Witness someone asking about why they should unplug things from the AC outlet when the devices aren't being used. They come up with some extensive theory about it, basicaly nonsense, yet the replies are pretty much addressing that "theory". I read the post, and ignore his "theory" and explain that things are drawing current when "off".

Often we don't even have that level of context. "I need to use a function generator IC at 2MHz, but it doesn't work properly", says the poster. But then later, it turns out that they don't need the function generator IC, they need a source of 2MHz signal. If the original question was about that, they'd get lots of answers instead of answers about trying to get an IC to work at the top of its frequency range.

The original poster in this thread may have a good reason for wanting no inversion in this op-amp stage. But, it may be that just as he is having a bit of problem getting it to switch from inverting to non-inverting, he may have some mistaken reason for seeing a need for it. We should help him do what he wants, but not help him in the deeper misunderstanding? Or for that matter, if he really needs to switch between inverting and non-inverting, point him in other directions, such as merely adding an inverting stage after this one, and have a switch to put it in or out of the circuit (which sometimes may make the design easier than switching the first stage from inverting to non-inverting).

Michael

Reply to
Michael Black

The opamp is pretty symmetrical with respect to the two inputs (the output is driven by the difference between the voltages on the two inputs, with a very high gain), until you connect the output back to one of them. Then there is a very big asymmetry. Connecting the output back to the - input generally tends to produce negative feedback, that lowers the overall gain, while connecting the output back to the + input generally produces positive feedback, that drives the output to saturation towards one supply rail of the other.

With no feedback to this node, and the inherent high impedance of the input, the opamp is essentially invisible to the microphone, which sees only the load of the bias divider.

That is because it puts positive feedback around the loop from output to input. Assume the - input is now held at 2.5 volts by the bias divider, but the + input is at a slightly more positive voltage. The opamp amplifies this difference, and since the more positive voltage is on the + input, the output is driven in the positive direction. But that also drives the + input in the positive direction (through the

47k feedback resistor), so the input difference voltage gets even larger, driving the output positive, even harder. This process continues till the output can go no more positive.

Why do you want to reverse the sense of the gain for the microphone signal?

Reply to
John Popelish

Yes, but the what happens in this case is that the person ends up having to think. Not only have the learned from there mistakes and having to fish out the details they will learn something new. If you try to think for the person and give them all the answers to they get the best solution without doing much work then it hasn't done anything for them. If you give them only what they ask and let them do what they want with it then they will be forced to think about what they are doing. If they are wrong a about it then it be much more significant to them in similar situations.

I would say the only counter to this is when there is something dangerous involved.

Well, There is a difference between someone posting a conclusion that is wrong and someone asking for help. No one knows better what they are trying to do than the person doing it. This isn't entirely true but I think its best to assume that as long as there is nothing dangerous involved. Ultimately it leads to that person learning more.

Right, but then that person probably will learn why its difficult or why it doesn't work... Someone else might learn something too.

Well, I just don't think its ok for others to try and assume what he's doing is wrong. To me that is arrogant. Theres many people in these NG's that think that they know everything about everything. If some person asks a question then they think its there job to come in and tell them how wrong they are and that it isn't going to work. What good does that do? People like that have no business helping others because they are not helping in the first place. No one wants to be told what to do. They want information.

If, say, the original poster is barking up the wrong tree then he will find out sooner or later. But guess what? He has learned to solve the issue of the op amp that he had. Now he knows more than someone telling him that he wrong in doing what he's doing. If that wasn't his real issue and he still runs into trouble he will post again with the real problem.

The main issue I have is that in general it does no good to try to analyze every possible context the person is working in. Sure it makes no sense to me why he's trying to invert an audio signal since our ears don't hear the difference... but maybe he has a very good reason. Or maybe, he will try it and it won't do what he expects. Either he will figure out why or he will post again. But reading minds doesn't work.

I do see your point but I think ultimately it hurts more than helps(in most cases). Its ok to be wrong and its ok to learn from your mistakes... and sometimes thats the best way to learn.

Jon

Reply to
Jon Slaughter

(snip)

I agree, completely. This is electronics.basics, and I see its purpose not so much as being about helping people build basic gadgets that do some task for them, ans to help people learn basic electronics.

I often pass on posts asking for design help where it is obvious that the poster has no real interest in electronics, but just wants to own some gadget, even if he has to build it.

I like trying to imagine the thoughts in a poster's brain as they come to grips with the fundamental processes involved in basic electronics. There is no point in answering a question that has no been asked, because their brain is not (yet) in a state where the answer has a place to be connected to everything else the person understands. I'm not here to build circuits so much as to build thoughts.

Reply to
John Popelish

is

trying

its

I'll heartily second that as well. The posts I particularly enjoy responding to, are those where I understand the posters line of thought and where they seem to be coming in from. Sort of 'empathy' as at some point, I've also been there, done it and got the T shirt. Sometimes I know they are heading off in the wrong direction but it doesn't matter one iota. It's the journey itself which forms the path to their true enlightenment.

--
Posted via a free Usenet account from http://www.teranews.com
Reply to
john jardine

Hello John & Jon,

Thank you for your answers and thanks not second guessing!

Okay, that makes perfect sense. When I was looking at it I think I mentally inverted the input twice and thought it would be stable that way. Guess it was too late last night..

Now, if keep the negative feedback path intact and add the signal to it, what happens with the DC bias? If I had a split supply and the signal was positive I can see the op amp creating a negative output which gets fed back. But if I hold Vp at 0 and simply apply 2.5V (plus signal) at Vn, then isn't the output going to swing to at most 0V ? So should I instead hold Vp at 2.5V and Vn at 2.5V plus signal ?

For voice recognition in a car I am trying to eliminate the background noise. Since I have two microphones I want to place one near the speaker and the other further away so that it picks up more of the engine/tire noise, then subtract one from the other. I intentionally left this part out to not get suggestions on novel circuits that are unrelated to the parts I have at hand..

- Marco

Reply to
grubertm

snipped-for-privacy@gmail.com wrote: (snip)

The inverting input is essentially a current summing node that holds the voltage at whatever is applied to the non inverting input. So you could put a bias divider on the + input to hold it at 2.5 volts, so that the feedback would hold the inverting input at that same voltage. Then you could change your microphone voltage signal to a current by adding a series resistor between it and the inverting input. When the input voltage swings positive, the output will swing negative to whatever it takes to force the feedback resistor to remove any current delivered through the input resistor. Unfortunately, this setup lowers the input impedance to the value of the input (voltage to current converting) resistor. The high impedance of the non inverting form is usually important to maintain the frequency response of electret mic inputs. It is usually easier to add a second opamp connected as an inverting stage with a gain of 1 to produce an inverted result.

Keep in mind that sound does not travel at an infinite velocity, so that mics at different locations will pick up signals phase shifted by the travel time for sound to pass between them, depending on which way it is traveling. This effect greatly complicates cancellation strategies.

Understood.

Reply to
John Popelish

Circa Mon, 07 May 2007 18:17:52 -0400 recorded as looks like John Popelish sounds like:

This solution has the benefit of keeping the original design intact (given the OP's inexperience with op amps), and pretty much making an identical but inverted signal for the subtraction of the noise signal. However, I see a possible design problem.

OP wishes to subtract the inverted road noise signal from the ambient signal to be used for the voice recognition. The design he envisions places the noise mic closer to the road noise than the ambient mic. This will likely produce a stronger inverted noise signal, and thus the subtraction won't be complete, but will be additive in the negative direction (I hope that makes sense).

So, I would make the gain on the inverter stage variable, say from 0.25 to

1.25, so that the resultant subtraction can be nulled under road-test conditions.

Further, the two signals must be subtracted somehow, and to do that, either a fourth op amp circuit is needed, designed as an inverting summing amp, or a passive summing network must be designed. Maybe this is already in the works?

So I will now posit that for really good brainstorming, the more info presented up front, the better!

Reply to
Charlie Siegrist

Sounds to me like there are problems in the basic design if noise is getting in. Decoupling needs sorting. Sort out any ground loops. Be careful with the layout of wiring, keep outputs away from inputs. Keep inputs away from mains transformers.

formatting link

Reply to
Marra

That's noise as in sound waves, not noise as in unintended AC. But I'll tell Hyundai to decouple the engine and tires ;-)

Passive summing network sounds nice. What happens if I just connect the two op amp outputs directly and use that sum them ?

Reply to
grubertm

It is common to invert one channel so that it can be Nilled out from the rest of it for example, later on using the left and right channels to cancel.. etc..

--
"I\'m never wrong, once i thought i was, but was mistaken"
Real Programmers Do things like this.
 Click to see the full signature
Reply to
Jamie

One will go into output current limit and the other will dominate the sum. The pair will also suck power like the dickens. Add a series resistor to each (say, 1k to 10k, and the summing works just fine.

Reply to
John Popelish

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.