Sound check!

Thread Starter

xox

Joined Sep 8, 2017
678
A friend of mine has asked me for some help with a little audio-visual project he's been working on lately. Basically, he just wants to convert his voice into pulsating lights. He's got one of these, a Shure SM57 dynamic microphone, which requires no power.

Shure-SM57.jpg


Now I am no sound engineer but this one did seem JUST simple enough for me to tackle. All the circuit needs to do is detect fluctuations in voltage. So this is what I came up with. Not particularly sophisticated mind you but possibly (?) good enough to get the job done.

voxlumen.png


The whole thing runs off of 5V and the op amp is in "single supply" configuration. The AC signal represents the mic, which produces voltages in the millivolt range. The output here is a single LED but just think of it as a 5V output driving the actual lighting display.

Now one thing that does worry me is that the big resistors would probably have to be PERFECTLY matched, otherwise the circuit may not work at all. Besides that it almost just looks too simplistic. Sure it runs in the simulator...but will it really work once everything is actually put together?
 

AlbertHall

Joined Jun 4, 2014
12,023
I don't think this circuit will work well visually with real audio. In you diagram, for instance the LED would be flashing at 220Hz which would be seen as continuously lit.
 

BobTPH

Joined Jun 5, 2013
4,923
Even if the resistors are matched, this would not work with a real opamp. The input offset will likely overpower your input, driving the output one rail or the other.

And, as stated, it does not give the effect you want.

What you need is a preamp using a normal AC amplifier opamp circuit feeding an envelope detector.

Bob
 

Audioguru again

Joined Oct 21, 2019
4,690
With no negative feedback the voltage gain at DC and low frequencies is almost one million times.
A microphone produces about 10mV when it is held in front of you so you need negative feedback to limit the voltage me gain to
about 250 times and use a peak detector to see voices.
I have a Sound Level Indicator circuit I made driving LEDs at different levels with an LM3915 IC. Its peak detector is like this:
 

Attachments

AnalogKid

Joined Aug 1, 2013
10,062
What you have is a comparator that turns the complex audio sine waveform into a square wave at the audio frequencies. If the circuit works, you will not see the LED blink because your eye cannot respond to something flashing 1000 times a second.

What you want is the basis of what used to be called a "color organ". Specifically, you want the bass channel. Filter off the mid and high frequencies, and envelope-detect the audio waveform just like in an AM radio. The difference here is that the audio frequencies are the "carrier", and the syllables are the information. The LED will blink with the inflections in a voice, or the beat of music.

Here are some example color organ designs:

https://images.search.yahoo.com/sea...pdnM-?p=led+color+organ+schematic&fr2=piv-web

ak
 

sghioto

Joined Dec 31, 2017
3,294
Now one thing that does worry me is that the big resistors would probably have to be PERFECTLY matched
Not exactly. For the circuit to work as designed the voltage on the negative input needs to be slightly (2mv) more then the voltage on the positive input. If the positive input is higher then the LED will always be on.
 

Thread Starter

xox

Joined Sep 8, 2017
678
Simulators are pretty good at having perfectly matched components and simulated opamps tend to have very low offsets, so you might want to add a “balance pot”.
Good idea. I think I'll put one on each voltage divider actually. That will make balancing the circuit (not to mention making crude sensitivity adjustments) a bit simpler.

I don't think this circuit will work well visually with real audio. In you diagram, for instance the LED would be flashing at 220Hz which would be seen as continuously lit.

I agree and in fact I told him the very same thing. He seemed fine with it though so I didn't press the issue...

Even if the resistors are matched, this would not work with a real opamp. The input offset will likely overpower your input, driving the output one rail or the other.


And, as stated, it does not give the effect you want.


What you need is a preamp using a normal AC amplifier opamp circuit feeding an envelope detector.


Bob
Interesting point. I am actually going to try this out on a cheap LM324 and to be honest I wouldn't be surprised if that is indeed the case for this particular one. Ideally I would like the circuit to work with virtually any opamp, so this may be something I will have to explore further down the road anyway.

With no negative feedback the voltage gain at DC and low frequencies is almost one million times.

A microphone produces about 10mV when it is held in front of you so you need negative feedback to limit the voltage me gain to

about 250 times and use a peak detector to see voices.
Thanks. Would you mind going into a little more detail (or maybe provide some links) about the importance of negative feedback for this type of signal?

I tested your circuit on a breadboard using 1% resistors a MCP601 op amp with a Sony F-K30H mic and it works just fine.

Sweet! That helps a lot actually. Now I know that there is at least ONE opamp that could handle such a relatively unrefined signal.


Not exactly. For the circuit to work as designed the voltage on the negative input needs to be slightly (2mv) more then the voltage on the positive input. If the positive input is higher then the LED will always be on.
Yes, of course. Those adjustment pots should make it easy to fine-tune for that.
 

Audioguru again

Joined Oct 21, 2019
4,690
Your opamp needs some negative feedback to reduce its low frequency gain from almost one million times to a reasonable amount of gain so it does not amplify its DC offset voltage and background noise.
 

Audioguru again

Joined Oct 21, 2019
4,690
Recently i had my hearing tested. It is done in a "sound-proof" room but I could hear the rumbling of the air conditioner next door. With low frequency gain at almost one million then the opamp will be saturated by very low level noise.

Voices have no output at 10Hz. Negative feedback can be added in a highpass filter so that only voice frequencies will activate the LED.
 

sghioto

Joined Dec 31, 2017
3,294
With low frequency gain at almost one million then the opamp will be saturated by very low level noise.
I don't understand where you get this gain of almost 1 million. The circuit is a comparator. As long as the -input is appx 2mV above the + input the output is zero. The negative peaks of the audio signal trigger the comparator. The LED lights up with each word spoken so it appears to blink or flash giving a visual indication of speech.
 

AnalogKid

Joined Aug 1, 2013
10,062
I don't understand where you get this gain of almost 1 million. The circuit is a comparator.
Don't confuse what it does with what it is. A comparator is a linear amplifier with some of the parts left out. It has a finite (non-infinite) open loop gain that has a maximum value at DC amd decreases with frequency. A comparator usually is uncompensated, which increases its speed. It also makes it vulnerable to oscillation at low gains, but since it usually is operated wide open (open loop), low phase margin is not a problem.

A comparator circuit can be many different things, but a comparator component is an opamp with very high gain, no compensation capacitor, and (often) an open-collector output.

As a reference, look at the datasheet internal schematics for the LM358 and LM393.

ak
 

Audioguru again

Joined Oct 21, 2019
4,690
Sghoto recommended and tried the opamp MCP601 in post #6.
The 10nF input capacitor into the two 1M bias resistors passes 32Hz and above frequencies are reduced 6dB per octave.
Ambient sound and electronic low frequency noise will activate it but maybe it can have an adjustable offset voltage to do what you want with the low frequencies. It will have much less sensitivity to voice frequencies, a comparator would work better.
 

Attachments

Top