Hi!

I have a mic + an amplifier which I'm connecting to a computer and measure sound through Audacity.

When I started, without any filter, I had a lot of noise, which can be seen in one of the images attached. But then I applied a lot of low pass filters.
Most of the noise is gone but there is still a very high peak around 9000 Hz, see image attached, and I don't know why ...
Does anyone have an idea? It's not removed even if I filter after the amplifier.

I power everything with a switching supply, a power bank.
Could it be due to a bad connection somewhere, bad grounding, wires, computer? Help me brainstorming, or better, you might know exactly what causes the peak.

Thank you!

 

Are you sure you've filtered the noise and not just attenuated it along with the wanted signal? What was your filter corner frequency and roll-off? I don't see much evidence from the above of any filtering just some 30+dB of overall attenuation with some increased in-band ripple. If your 12KHz spike is externally injected post amplifier, filtering won't impact it. It could even be an artefact of the computer sampling.

 

hm, I'm a beginner so not sure, but what I did was applying a resistor in series with the voltage input (5V) and then a capacitor in parallell and connected to ground. I thought that would create a passive low pass filter?

To begin with I filtered the voltage supply to the amplifier with a resistor of 47 ohm and a capacitor at 1000uF. Then I connected another low pass filter before one of the resistors that comes from voltage supply at the PCB and directly to the OP-amp. I connected another resistor directly after voltage supply that is 30kHz and a capacitor in parallell which is 10 uF.

I can try to post the whole circuit for you so that you understand better, it was hard describing here.
But I'm very grateful for your help! Thank you!

 

Schematics always help, as does specs/links to the equipment you are using. Its hard to advise in a vacuum of information; the devil is usually in the details...

 

yes, sorry about that!

So the microphone is a Panasonic WM61a, the amplifier is called EK022 if you want to search (but i remodelled it a bit), the OP-amp TLC271. It's powered by a power bank with switching supply (guess that's the problem) +5V.

Right know I only placed a low pass filter containing a resistor and a capacitance. I've tried placeing one both just before the amplifier, and also one at the circuit before the resistor that goes directly into the Op-amp. With similar result.

I tried to shorten the circuit so that the mic wasn't connected, but it also gave me the same noise.

 

It would appear your opamp is configured for high-bias operation (pin 8 grounded).

Your output noise from the opamp alone across 1Hz - 10kHz will be around 0.25mV rms (see video below for explanation). The 100k input resistor will produce double that, around 0.4mV rms @ 25degC, so you have 0.47mV rms total , giving you a noise floor around -66dBV.

If you physically ground the microphone input (on the mic side of the input capacitor, what do you actually get as the noise level?

The 10k/10u on the feed to the mic has a cutoff at 1.6Hz so the mic supply should be reasonably clean. The opamp itself will reject PSU noise by 66dBV or more, though it should have a 10u and 100n in parallel from from pin 7 +V to pin 4 GND as good practice. You might try a 100n in parallel with C3 just to ensure no HF noise is getting in there...

Where is the spectrum analysis being generated? Your rats nest could be picking up that 12kHz signal as it doesn't appear to be screened cable from the output to where-ever its going bottom left of picture. How is that terminated?

 

As an aside, C3 needs to be bigger. With 10u its impacting the LF response of the circuit:

 

Your schematic was huge and sidewise so I made it smaller, turned it and adjusted its brightness.

1) The datasheet for your mic says it draws 0.5mA max. Then your two 10k resistors feeding 5V to it must have 10V across them to produce 0.5mA, plus the 2V that the mic uses.
The resistors will have 5V minus the 2V for the mic= 3V and with a current of 0.5mA the total resistance must be 3V/0.5A= a total of 6k ohms. You cannot use two 3k resistors because the low value resistor at the mic reduces its output level. Instead use a 1k resistor at 5V, increase the filter capacitor to ground to 47uF and use a 5.1k resistor feeding 2V to the mic.

2) Your opamp is biased wrong and has positive feedback.

3) Your TLC271 opamp has 8 times as much noise as a low noise audio opamp.

I fixed the schematic but the gain of only 11 times is much too low for a mic preamp.

 

1) The datasheet for your mic says it draws 0.5mA max. Then your two 10k resistors feeding 5V to it must have 10V across them to produce 0.5mA, plus the 2V that the mic uses.
The resistors will have 5V minus the 2V for the mic= 3V and with a current of 0.5mA the total resistance must be 3V/0.5A= a total of 6k ohms. You cannot use two 3k resistors because the low value resistor at the mic reduces its output level. Instead use a 1k resistor at 5V, increase the filter capacitor to ground to 47uF and use a 5.1k resistor feeding 2V to the mic.

Good call, I hadn't looked at that...

2) Your opamp is biased wrong and has positive feedback.

Hmmm, I'm not convinced. Its a little unconventional, and your revised circuit is better, but it does work as intended, no obvious positive feedback problems, though I agree that without sufficient decoupling of the bias point it could have issues. Here's both, yours has a better LF response, but the 220nF capacitors are too small for the LF end in both...

3) Your TLC271 opamp has 8 times as much noise as a low noise audio opamp.

I agree, there are plenty in an 8pin DIP package with <2nV/√Hz but at a price. The TI NE5534 is similarly priced to the TLC271 and has a noise figure of 3.5nV/√Hz, much better than 20nV/√Hz!

I fixed the schematic but the gain of only 11 times is much too low for a mic preamp.

Except its actually 101 ( 10k/100+1)

 

With the opamp biased wrong then it has DC positive feedback that forces its output to the positive supply voltage then it doesn't work. Maybe the simulation software does not know that.

Your schematics do not show the TLC271important pin 8 "bias select" setting.
The datasheet of the TLC271 an H version. It is noisy Cmos, not bipolar.
The old bipolar NE5534 low noise audio opamp has a 6V minimum supply rating.

Now you have an output capacitor of only 220nF feeding a new load of 10k which produces -3dB at 73Hz. With no "bias setting" then the output of the TLC271 might not be able to drive it.

 

Hi AGa,

The simulation of the tlc271 has 3 variants. The H suffix shows this is the High Bias setting, ie pin 8 is grounded. I guess this was easier than trying to simulate the pin 8 functionality; there are 'M' and 'L' suffixed files too. It also means you can use the LTSpice opamp2 generic 5-pin symbol rather have to create a new symbol.

In the High Bias setting the amp is spec'd driving a 10k load so I added it to ensure its working as spec'd. The 220nF input & output caps are too small for the 20Hz - 20kHz bandwidth of the mic. They should be at least 10uF, preferably a lot bigger.

 

Thanks a lot everyone. I'm going to have a look!

Can anyone help me with what to do with the braided shielding on the wire that I have from the microphone?
It's suppose to go to ground so I connected it with the blue wire (-) from the microphone which goes to a hole in my pcb which is connected to ground. Is this correct? Or should I connect it to the - hole on the output that goes both from the computer and the power supply? I'm confused if I connected it correctly to ground.

Thank you!

 

As long as the braid is grounded it makes little difference in this situation; it shuld be connected to ground near the mic input. The braid on the screened output cable should be connected to ground near to the output connection.

The important thing is to avoid ground loops by grounding things more than once, or grounding them through long thin wires or traces.

 

So it doesn't matter that I first connected a wire from the hole and then soldered the ground wires and the braid to that wire?

Thank you for fast answering!

 

No, as long as its as short as you can make it

 

thank you everyone! I've tried a lot of things and what I finally realised was that it was the OP-amp greating the peak ....
I bought new ones and they are much better.

I have one last question though. You talked about filtering a bit different from the +5V to the mic. You said that you suggested 1k ohm and then a capacitor of 43uF and then a 5.1 kohm resistor feeding the mic. Unfortunately I do not have those exact values, would it be okay with still using 10 uF and instead lowering the resistor values?

Thank you!

 

That peak at 9kHz could be from the switching power supply.
You might try a linear supply instead.

 

thank you everyone! I've tried a lot of things and what I finally realised was that it was the OP-amp greating the peak ....
I bought new ones and they are much better.

I have one last question though. You talked about filtering a bit different from the +5V to the mic. You said that you suggested 1k ohm and then a capacitor of 43uF and then a 5.1 kohm resistor feeding the mic. Unfortunately I do not have those exact values, would it be okay with still using 10 uF and instead lowering the resistor values?

Thank you!

I said to use a common 1k resistor, 47uF capacitor and 5.1k resistor. A resistor with a value less than 1k and a 10uF capacitor will be fine if the power supply does not produce 120Hz hum. The 5.1k resistor can be a 4.7k resistor which will reduce the mic output level a little.

 

Hi everyone!

I have a new question regarding the circuit, can someone tell me how I calculate the amplification? I know that I should calculate it from the resistor values somehow, but I would like to know exactly how.

I would like to lower the amplification, can I do that by lowering the R2 resistor and the 100k resistor in parallelll?

Thank you!

 

The basic amplification in your original circuit is given by R2/R3 =10k/100 = 100. The 100k across the inputs has no bearing on the amplification.