I'm tired of hearing noise!

[Plecto]

Hi. I'm getting various degrees of different types of noise in every type of audio amplifier I make, but I don't know where the noise is coming from or what to do to prevent it I've been getting better at removing power supply noise. I understand the need for star groundings and I'm better at seeing where power supply noise can enter the amplifier. What I don't understand are other types of noises, the most dominant for me is the constant white noise hissing sound that often varies with the position of the input pot. I'm also often troubled by the amplifier acting as a microphone (I can hear it when flick the pcb). A third issue is that the static hissing noise intensifies when I touch the metal of the pot (even though it's not connected to any of the pins on the pot).

I would like to understand where this noise is coming from and know what precautions I have to take to prevent it. Could anyone point me in the right direction of where I could obtain this knowledge?

 

[vicky001.iith]

Not able to understand the problem.Please show your circuit diagram and detail about the audio amplifier you are using ?
I also suffered from the same typo problem default white noise but i i solved it and noise level is very low now.
Happy to help you

 

[Audioguru]

Since your amplifier is microphonic then its gain is much too high. It also amplifies its own noise and the thermal noise of the input resistors.
A properly designed hifi amplifier has noise that is so low that you do not notice it.

An input volume control is quieter if its metal case is connected to 0V in the circuit so it is a shield.

 

[Plecto]

I've made different designs with the same problems, I'm not talking about any particular design. An example would be a simple inverting op-amp design with a mains transformer and rectifier as a psu I made the other day. I had heavy humming at first (as expected since the non-inverting input was simply biased to 1/2 supply), but after replacing the bias resistors with zener diodes the humming disappeared completely. What was left was this hissing noise we've been talking about. I know zener diodes can be noisy, but can the noise really be heard with a gain of 10? The load was a 32Ohm headset, not sure about the sensitivity. I'm also aware of thermal noise from resistors, but I was under the impression that this noise was vanishingly low so that it wasn't an issue in audio amplifier designs? Another example would be an op-amp input stage with a bjt class AB output stage. The power supply on this amplifier was a shitty 24V wall wart, but even after adding a regulator to make the supply smooth, the hissing noise was still quite annoying.

Since your amplifier is microphonic then its gain is much too high.

I guess I could lower the gain quite a bit on many of my designs, but that would make the amp less acceptable to different load impedance's.

 

[crutschow]

What op amp did you use? Some of the old ones, like a 741 are very noisy.

 

[Plecto]

I'm using NE5532, but when making them battery powered, I have no noise what so ever. I was more interested in the theory behind the possible noise sources in audio amplifier designs, not a fix to a particular problem I have. Lets take another scenario; I'm about to make a new amplifier from scratch, what kind of precautions should I take?

 

[betwixt]

NE5532s are exceptionally good low noise amplifiers so either you are not using them optimally or you have the gain far too high. They are not designed to drive 32 Ohms loads though so that may be part of your problem. Consider that if you overload the output you also restrict the amount of it that can be fed back to control the gain so it may be higher than you imagine. Biasing the input with Zener diodes will slightly increase noise, they are actually used as noise sources in some applications but they should not work any better than a potential divider. It sounds like you are not filtering the power supply or biasing networks properly, try using two resistors again but add a capacitor to ground as well to stop signals entering the non-inverting input pin.

Brian.

 

[Plecto]

Thanks for your reply. I guess I could add a regulator, but I liked the idea of using a zener diode. As I said, adding the zener completely removed the humming, but does zener noise sound like the hissing noise we've been talking about? I've tried having a voltage divider with caps in parallel, but I think I need rather big caps to make the humming go away. I did some calculations regarding this. Having 100kOhm bias resistors and a 1uF cap in parallel will lead to a voltage drop of 6V*e^-(0.01/100000*10^-6)=5.42V so if the ripple is less than that, the cap won't do any good. It sounds plausible that it's the NE5532 driving too heavy loads that is the problem, but as I mentioned, I'm not getting any noise at all when the amp is battery powered. I would really like to know the possible sources of this static noise (regardless of amplifier design), like a list if you like I would also like to know how this noise is formed and why it sounds like it does :S

 

[crutschow]

Generally noise from power a power supply is some fixed frequency like 60/120Hz hum or higher frequency if it is a switching supply. A hiss indicates white (wide-band) type noise which sounds like what you hear on an FM radio between stations. White noise is inherently generated in semiconductors such as transistors and zener diodes (shot-noise) and resistors (Johnson or thermal noise).

Generally when building a low noise amp you use low noise ICs, a good ground (preferably a ground plane) with decoupling of all ICs directly from the supply pins to ground, and a careful layout, keeping the sensitive input away from the output and any other possible noise sources. And, of course, (here's the part many beginners don't do) thoroughly read the data sheets of all the semiconductors so that you use them well within all there operational limits (such as not using a standard op amp to drive a 32 ohm load).

 

[betwixt]

Any regulator or Zener arrangement to set the half supply voltage is NOT a good idea as it will try to maintain a constant voltage even if the supply varies. The two equal resistors in a potential divider are the better way to do it and their mid point will track the supply no matter what it is. The NE5532 is a special case in this respect because it has 'head to tail' parallel diodes across it's input pins and thus will not work properly if you try to put more than a few hundred mV between them.

You must understand that ALL components produce some noise, it's a fact of life and caused by random molecular activity in the materials used in construction. The only sure way to kill the noise is to work at absolute zero temperature but that's impractical and also the headphones tend to freeze to your ears. :lol:

From your descriptions, the sources of noise that seem to be troubling you though are a combination of component noise and interference from sources outside the amplifier. The prime suspect is the power supply of course and you want that to be as stable as possible but there are other entry points to consider. What you need to discover is whether the interference is common-mode or differential because each has to be treated differently. All amplifiers increase the signal at their input but they do it relative to another point in the circuit. Often this will be 'ground' but it could also be another input, as is the case with op-amps. A common mode interference is one which appears at both inputs simultaneously and so in theory cancels out but in reality, differences in component values and signal paths may allow a difference to occur. Differential interference is the kind that appears between the ground and input, basically the same as the audio you want to amplify but coming from an undesired source.

Please post your amplifier schematic and if posssible a photograph of the amplifier so we can see it's construction and hopefully spot the possible points of interference entry, it will be much easier to diagnose if we can see it rather than speculate on the many possible causes.

Brian.

 

[Plecto]

So having 50hz humming noise because of the lack of a star grounding would be an example of differential noise? Let's say the input ground is connected to the same ground trace as the rectifieing caps, the trace acts as a resistor creating a voltage drop when the cap is charged/discharged thus creating the humming? I'm not sure what common-mode noise is though

The whole point of the zener was to have a steady bias voltage as I don't want the bias point to track the supply. I know that the zener Vf will have to be 1/2 supply, but I can't see this being a problem if I'm only going to use one type of supply.

Here's a schematic and picture of my amp:

 

[betwixt]

If you don't let the bias track the supply, it will no longer set the amp output to give maximum output swing, the quiescent state could be shifted toward one of the supply rails. You are correct about voltage drops in the PSU wiring causing hum but a properly designed power supply would keep all the charging currents within it's own boundaries and only the fully filtered output would be passed to the amplifier. In essence, the wires between the PSU and amplifier should only carry the amplifier current and nothing else.

I can't see the photograph of your construction, you posted the schematic twice !

Brian.

 

[jiripolivka]

Noise in any amplifier can be difficult to remove. Essentially it cannot be removed but by a good design you can reduce it.
If your audio amplifier is good with battery but not with AC power, you should learn to make a better power supply.
Forget about Zeners, they do generate a lot of (hissing) noise. Use good regulators, again not the 780x series. Use lower gain and good screening.
Best of all, go back to battery power.
To connect audio , professionals use light guides and batteries in the best equipment.

- - - Updated - - -

In your schematic there is GRAVE error opening the way to noise: both positive opamp inputs must be grounded by 10 uF or larger capacitors, otherwise power supply noise i directly amplified.

 

[betwixt]

I'm also curious as to the two different ground symbols and what they mean in real construction. R14 also looks very suspicious. That's why I wanted to see a photograph.

Brian.

 

[Plecto]

Here's the picture:


About the zener. In this case, I'm not in a desperate need of the full output voltage swing and even so, the output isn't that limited as the zener voltage is 6V and the supply is like 12.5V. I don't really like the resistor divider and 10uF+ capacitor as it will only reduce the humming and that solution will require huge rectifying caps to keep the supply ripple at a minimum. If the zener diode is really producing the hissing noise that I am hearing, what about using a 6V reference chip on the non-inverting input (if they produce less noise)? That will allow for far greater ripple on the supply at least.

Use good regulators, again not the 780x series. Use lower gain and good screening.

What regulator would you recommend? And what do you mean by screening?

I'm also curious as to the two different ground symbols and what they mean in real construction. R14 also looks very suspicious. That's why I wanted to see a photograph.

It wasn't intentional to use two different ground symbols, I'm just shitty at making schematics R14 and R15 are 0Ohm bridges between the amplifier ground and +V and power supply. I which there was a decent way of making these resistor bridges without adding resistors that seem very out of place when only looking at the schematic. R14 and R15 jumps over the trace going from one output decoupling cap to the mini jack.

 

[betwixt]

Ideally, you want the resistance at the + and - inputs of the amp to be equal so the symetrical input stage of the amplifer is fed from equivalent sources. Using a Zener or a regulator will not do that, I still say use a pair of equal value resistors. There should be no humming at all, either you hear it is because the supply line has ripple on it, in which case stabilizing one input voltage will make matters worse, or you have a coupling source elsewhere to the amplifier input. You could also consider using a split supply in future as this eliminates the need for the biasing network and the output coupling capacitor/resistor and gives you better LF response as well.

Looking at your construction I would say the most likely point of hum introduction is by inductive coupling from the transformer to the PCB tracks, they are close enough for significant voltage to be induced in to them. The other likely entry point is by capacitive coupling between the primary and secondary sides of the transformer. Basically, you couple some AC line noise (some hum and some 'crackle' interference) into the whole circuit from the PSU end and sink the current to whatever you have connected to the input. I appreciate it's never good to have faults pointed out after you have clearly done a lot of work!

This is what I would suggest with the present board:

1.cut it in two so the transformer, rectifiers and reservoir capacitor can be physically separated from the amplifying part, this will reduce or eliminate magnetic coupling. Reconnect the tracks using fairly thick wires, even for a headphone amp the tracks you have used look rather thin.
2. connect the metal body of the volume control to the ground, preferably at the 'bottom' end pin.
3. If those diodes near the IC are Zeners, remove them and go back to resistors but place a capacitor (say 10uF) across the ground side resistor.

Incidentally, you only need one 'half-supply' network, you can link it to both amplifiers.

If you rebuild it:
1. follow the suggestions above and also flood fill the PCB. In other words place copper everywhere between other tracks and components. Make sure it's connected and used as the ground point.
2. place 1nF ceramic capacitors across each of the PSU rectifier diodes.
3. place 100nF ceramic capacitors across the IC supply pins as close to the IC body as possible.
4. make sure you use a double isolated AC mains transformer or alternatively one with a shield between windings which you should connect to ground.

Brian.

 

[FvM]

I agree with most of the points in betwixt's post #16, except for this one

Ideally, you want the resistance at the + and - inputs of the amp to be equal so the symetrical input stage of the amplifer is fed from equivalent sources.

Why should you do that in an audio amplifier? The solely purpose of OP source resistance matching is to cancel the offset voltage caused by it's input current. It matters for DC amplifiers and is effectively meaningless for AC, e.g. audio amplifiers. Furthermore it doesn't apply for amplifiers where the input bias current is in the same order of magnitude as the offset current, e.g. FET input amplifiers and modern bipolar amplifiers with input current compensation like OP07 or OP027.

For audio amplifiers, noise is the most important criterion for resistor selection and it usually ends up in different dimensioning than resistance matching.

Referring to the circuit in post #11, I would add two comments:
Generally, a non-inverting amplifier scheme, connecting the input source to the non-inverting input offers more optimization features towards low-noise, e.g. selecting the feedback network independent of the source impedance and being better prepared for sources of different impedance.

Secondly, the bias voltage for a low-noise audio amplifier should be always RC filtered after deriving it from the power supply. The noise level of voltage regulators or zener diodes is considerable higher than the OP noise. Feeding the bias voltage unfiltered to the OP ruins the circuit's noise performance.

 

[Plecto]

Thanks for a great response Betwixt. I will try and make a new amplifier with the changes you mentioned. About setting the bias point with a resistor divider, as I mentioned before, this will cause ripple if there is ripple on the supply (which it is). Adding a 10uF+ capacitor will of course help, but how much exactly? Calculating the discharge gives this: 6V*e^-(0.01s/100kOhm*10^-5F)=5.94V so if there is ripple on the supply, the ripple on the non-inverting input will still be 60mV, no? I've chosen 100kOhm resistors and I've not taken the input impedance of the op-amp into account.

I would love to have a dual sided supply, but it's hard to find cheap center tapped transformers of this size (the one I'm using is 0.7VA). I could make a virtual ground, but I've been unsuccessful with this in the past. The single sided design I'm using now always seems to work as intended, the only issues are noise.

I don't quite get how the noise is induced from the transformers to the traces or how this would be simulated. It's not like this noise is induced to the whole circuit, right? It's only induced to the supply traces? If so, can't I just treat this noise the same way I treat 50hz humming noise, trying to prevent it from entering the amplifier? Using a voltage regulator was mentioned in this thread, what if I go with that? I have to use a low noise regulator though, but shouldn't this make it so that the amplifier won't be able to distinguish between being powered from a battery or from a transformer, a completely smooth supply? Also, if the hissing noise comes from the supply, how does it enter my amplifier when I have a zener diode to set the bias?

About the list of changes I should do to the new design. What purpose will the 1nF caps have? And about the 100nF cap, I can understand that a cap close to the supply pin of the op-amp is important for stability, but my amp doesn't oscillate or anything so what good will it do here?

Again about the zener diode. I don't quite understand what you mean by having equal resistance on the + and - input. I faintly remember that input resistances can be set to remove offset voltage from uni, but as FvM pointed out, how can this be relevant? I still don't understand why using a zener diode is not a good idea (unless the noise I'm hearing is indeed from the zener), to me it seems like it makes the amp almost immune to supply ripple. FvM: A low-pass filter to set the bias point sounds like a good idea, but what is the best way to implement this?

One last point. I also mentioned microphoning and I'm curious to what can cause this. To me this phenomenon seems impossible and I don't understand it at all In the past, this problem has made me rage quit a project. Let's say I just managed to lower the noise of an amplifier to acceptable levels, but then suddenly Murphy makes it so that my amp starts acting as a bloody microphone, that's kind of the last drop

 

[betwixt]

I take the point that equivalent resistance at the two inputs is not particularly important in an audio amp - I must spend too much of my time in control systems! (about 25 hours every day)

The figure of 60mV is unrealistic and most certainly would account for lots of hum. The supply itself should be almost ripple free, maybe 50mV or less and a filtered half-supply point should have only a few uV on it.

Placing 1nF capacitors across the rectifiers may do nothing at all but it isn't unknown for some diodes to produce switching noise as they recover from reverse bias condition. It's usually at higher than audio frequencies but adding capacitors with such low value will do no harm and could prevent problems occuring. The 100nF is essential, if the IC is unstable, the chances are it will be oscillating at a frequency higher than you can hear and that can cause all kinds of unpredictable effects, including additional noise level.

Inductive coupling from the transformer will be into all traces nearby, include those in the amplifier part of the board. It's voltage induced by the magnetic flux escaping from the transformer core. A toroidal transformer is better at constraining it's magnetic field than a conventional EI cored type but I'm guessing from the PCB photograph it's the latter type you are using.

The 'microphoning' is technically called "microphony" and is caused by changes in component properties as they are physically stressed. It's the same effect that is used to advantage in a real microphone but obviously other components are designed to exhibit it as little as possible. In your circuit it shouldn't be noticable as the voltages should be very tiny. You don't show any values in your schematics but I would guess you are either using extremely high gain or it is actually unstable already and the effect you see is from 'tuning' the oscillation rather than amplifying a voltage.

You don't need a center taped transformer to create a split supply, you can still do it with four diodes and one extra capacitor.

Brian.

 

[Audioguru]

I've made different designs with the same problems, I'm not talking about any particular design. An example would be a simple inverting op-amp design with a mains transformer and rectifier as a psu I made the other day. I had heavy humming at first (as expected since the non-inverting input was simply biased to 1/2 supply), but after replacing the bias resistors with zener diodes the humming disappeared completely. What was left was this hissing noise we've been talking about. I know zener diodes can be noisy, but can the noise really be heard with a gain of 10? The load was a 32Ohm headset, not sure about the sensitivity. I'm also aware of thermal noise from resistors, but I was under the impression that this noise was vanishingly low so that it wasn't an issue in audio amplifier designs? Another example would be an op-amp input stage with a bjt class AB output stage. The power supply on this amplifier was a shitty 24V wall wart, but even after adding a regulator to make the supply smooth, the hissing noise was still quite annoying.



I guess I could lower the gain quite a bit on many of my designs, but that would make the amp less acceptable to different load impedance's.

You do not use noisy zener diodes to bias an opamp. Instead you use two resistors as a voltage divider and a filter capacitor to prevent hum.
Many opamps like the antique 741 are very noisy. Use a modern low noise audio opamp instead.

A half-decent power amplifier has an extremely low output impedance of 0.04 ohms or less and has a lot of negative feedback. Then it can drive a 2 ohm speaker (if it does not burn out from the high current) at exactly the same level as a 16 ohm speaker or 32 ohm headphones.