lm358 microphone preamp noise

[paulhm81]

Hi

I built a preamp using the schematic in the attachement but instead of 100k R5 I used a 1M pot so I can adjust the gain and it's powered from a 9v battery.  I built it on a breadboard and I know that it picks up a lot of noise but when I touch the ground of the circuit the noise dissapears so I thought why not connect the ground to earth so i connected a wire and it's super silent now. How bad is that idea? What other choises are there?

[Zero999]

Yes, that will help to reduce the noise. It it's not convenient to earth it, then putting it in a screened enclosure will have a similar effect.

The LM358 isn't very good for audio because the output stage is class B and it has crossover distortion. Replace it with the TL072 or NE5532 and it will be better.

[nuno]

What if you put another resistor divider, say with 100K resistors, to bias the node C1-R4 to Vcc/2?

[dom0]

Huh? The bias of that node is already established by R5.

"A lot of noise that goes away when I connect the ground wire" is likely caused by either
- a very large ground loop with the wire not in place
- no common ground between the connected parts, so that the ground is only established by parasitic capacitive coupling.

~> Did you connect the ground of the cinch or TRS plug (or what you used to connect the output) to the circuits ground?

Also, as Hero999 already said, there are many op amps far better than the 358/324 for audio. The 5532 would be a classic and is both cheap and well suited for this kind of amplifier.

A FET-input op amp doesn't improve things here over a bipolar one, since the input impedance is low (1 k?) - whether this is bad or causes significant attenuation of the signal depends on the microphone, but with common types you are looking at an attenuation of maybe 2 to 4 fold ; thus the noise gain - signal gain ratio of this circuit is poor (101 * attenuation / 100). I suggest making the circuit non-inverting, biasing the (+) to C1 node explicitly.

There should be more than plenty of schematics to draw inspiration from, but make sure to filter audiophool stuff out... ESP probably has something well-thought out.

[nuno]

it's not exactly to bias, bad word choosing, the intention was to set a "default" input signal.

[dom0]

?

[paulhm81]

Huh? The bias of that node is already established by R5.

"A lot of noise that goes away when I connect the ground wire" is likely caused by either
- a very large ground loop with the wire not in place
- no common ground between the connected parts, so that the ground is only established by parasitic capacitive coupling.

~> Did you connect the ground of the cinch or TRS plug (or what you used to connect the output) to the circuits ground?

Also, as Hero999 already said, there are many op amps far better than the 358/324 for audio. The 5532 would be a classic and is both cheap and well suited for this kind of amplifier.

A FET-input op amp doesn't improve things here over a bipolar one, since the input impedance is low (1 k?) - whether this is bad or causes significant attenuation of the signal depends on the microphone, but with common types you are looking at an attenuation of maybe 2 to 4 fold ; thus the noise gain - signal gain ratio of this circuit is poor (101 * attenuation / 100). I suggest making the circuit non-inverting, biasing the (+) to C1 node explicitly.

There should be more than plenty of schematics to draw inspiration from, but make sure to filter audiophool stuff out... ESP probably has something well-thought out.

I connected the output through a stereo jack to some active speakers because the goal is to make a two way intercom through a glass window like the ones used by cashiers. The ground for the dc jack is from the common ground. I repeat that I used a breadboard with a lot of long jumpers. I will try again with the op-amps suggested and I will also try the non-inverting option. I'm planning to power the board through a linear 5v regulator from the speakers beacause I think I will use some crappy pc speakers.   If it still has noise whatever I try, it's ok to earth the ground?
Thank you!

[Zero999]

Also, as Hero999 already said, there are many op amps far better than the 358/324 for audio. The 5532 would be a classic and is both cheap and well suited for this kind of amplifier.

A FET-input op amp doesn't improve things here over a bipolar one, since the input impedance is low (1 k?) - whether this is bad or causes significant attenuation of the signal depends on the microphone, but with common types you are looking at an attenuation of maybe 2 to 4 fold ; thus the noise gain - signal gain ratio of this circuit is poor (101 * attenuation / 100). I suggest making the circuit non-inverting, biasing the (+) to C1 node explicitly.

I agree with you about a FET input op-amp not giving any advantages over a BJT op-amp but the TL072 is still a heck of a lot better than the LM358 for audio. A bipolar op-amp such as the NE5532 would be superior to the TL072 in this case but it costs a bit more.

[Audioguru]

Your opamp is inverting with a 1k input resistance that shorts away a lot of the signal level from the electret mic. Then you need a lot more gain to make up for the signal loss then you hear more noise. The opamp should be non-inverting with a much higher input resistance then the input capacitor can be a lower value film type that has no polarity. Which is more positive, the mic or the opamp bias voltage? Maybe your huge electrolytic input capacitor has its polarity backwards.

Your hum problem is caused by your breadboard. Its rows of contacts and many jumper wires all over the place are antennas that pickup all kinds of interference. Use a compact pcb or stripboard layout and it will be fine.

I use TL071 single, TL072 dual and TL074 quad audio opamps for all of my audio circuits. Here is my mic  preamp circuit:

[Audioguru]

A TL07x opamp minimum supply voltage is 7V. Your 5V powered pc speakers much be cheap and tiny. Mine are powered from 12VDC.

[dom0]

Also, as Hero999 already said, there are many op amps far better than the 358/324 for audio. The 5532 would be a classic and is both cheap and well suited for this kind of amplifier.

A FET-input op amp doesn't improve things here over a bipolar one, since the input impedance is low (1 k?) - whether this is bad or causes significant attenuation of the signal depends on the microphone, but with common types you are looking at an attenuation of maybe 2 to 4 fold ; thus the noise gain - signal gain ratio of this circuit is poor (101 * attenuation / 100). I suggest making the circuit non-inverting, biasing the (+) to C1 node explicitly.

I agree with you about a FET input op-amp not giving any advantages over a BJT op-amp but the TL072 is still a heck of a lot better than the LM358 for audio. A bipolar op-amp such as the NE5532 would be superior to the TL072 in this case but it costs a bit more.

I don't think that matters, I can buy a TL072 (national/TI) here for about 35 cents, similarly a NE5532 (nat/ti) costs a little less than 30 cents - not really a relevant difference... although somewhere else the prices might be much more different, and much higher, of course...


For 5 V operation none of these work, you'll need RRIO audio op amps for that, which are typically more expensive.

[paulhm81]

It's not about the few cents for the parts. I could buy a 150 $ system that just works. It's about building it myself and learning something in the process and if I get away with a few bucks for the parts instead of buying it then I'll be happy about it. Hero999 said I could build a cage for the project but what I would like to know is what's wrong with connecting the ground to earth if it works. I'm thinking there's something that I'm not aware of. I could use a dead notebook charger that has those aluminium plates as a case for the circuit.

[Audioguru]

I built ONE audio circuit on a solderless breadboard and it picked up so much hum and interference that I NEVER used a solderless breadboard again. I quickly learned how to plan a compact stripboard layout (today there is software that does it) and everything is soldered so there are no intermittent contacts like on a solderless breadboard. The copper strips that are cut to length with a drill bit form most of a pcb and a few very short jumper wires form the remainder of a pcb.
None of my thousands of audio projects built compactly on stripboard needed a metal box or metal cage. Usually I used a plastic enclosure.

I hope for audio connections you are using shielded audio cables. 

[tooki]

Weird. I've prototyped a number of audio circuits on breadboards and haven't had any problems with hum, at least not caused by the breadboard. (Just the usual ground loops or unshielded cables, which of course affect permanently constructed circuits as you said.)

What is the stripboard design app you recommend?

[Audioguru]

Each contact row and each of many long messy jumper wire on a breadboard is an antenna that picks up interference and mains hum.
I do not use a modern app for stripboard layouts but some are available. I use my brain instead. The challenge to make it compact is fun. I made some very complicated huge digital circuits.