sensitivity of microphone....

[Audioguru]

On all the other websites he is using a low voltage rail-to-raill Cmos opamp that has no input bias current.

I think the preamp should have most of the gain and that the Sallen and Key active Butterworth lowpass filter should have equal-component-values and a gain of only 1.6.

 

[Kinshoro]

Actually I am using a MEMS microphone and I cant increase the gain above 10 in the first stage.... are there any other mics which are surface mountable but without an integrated preamplifier in them.......

Added after 3 minutes:

The sallen and low pass buttersworth filter should have equal component values?? does that mean that it should be similar to my first stage of amplification????

 

[keith1200rs]

I don't see any reason to change from the microphone you are using. Having a preamp actually in the microphone is good place for it to be. I also don't see why you need to stick to a "Sallen and Key active Butterworth lowpass filter should have equal-component-values and a gain of only 1.6" - maybe Audioguru could explain why? I would probably limit the gain of that filter to maybe 10 and add an extra stage if you want more gain - opamps are cheap and it is easy enough to use multiple stages to get high gain instead of trying to get lots of gain with one stage which can put a demand for high GBW on the opamp.

Keith.

 

[Kinshoro]

ok..... then I will just change my opamp and see how it works....... so far it seems lmv721 isnt working the way I want it to....... I guess I will reduce my ac coupling capacitors as well to around 0.1 uF........ so that the it wont take too long for the values to stabilize in my circuit.......

Added after 48 minutes:

I think low pass buttersworth filters with sallen key design have maximum gain limit... I have confirmed it in this document from TI.... link: **broken link removed**

check page 7 and it will explain the requirements for the sallen key topology...It says that the filter used should be unity gain and the pole order should be below 3.. I think thats why my output at the opamp was unstable..... I was thinking maybe I should use a single stage for filtering and then another for amplifying the signal......

Added after 18 minutes:

I think something like this will work.......... I have made a low pass buttersworth with unity gain and another stage for the amplification....

 

[keith1200rs]

You need to be careful of the tolerance of the values in the filter, particularly if you are using high gain. I would be tempted to run the gain stage will little or no filtering first and see what results you get, unless you are already sure of what filtering you need.

The only problem I see with the LMV721 (apart from the fact that National Semiconductor cannot get the model right) is the high bias current which will give you a voltage shift if you are using 1M resistors in the feedback.

I don't know if you have a PCB for this, but you really need one for this sort of circuit. The signals are low, impedances can be high and a good ground plane will help avoid pickup problems.

Keith.

 

[Kinshoro]

I tried implementing this design(not just simulate)... but it seems the output of my opamp at my third stage is always at 3V....... should something like this be happening in my circuit..??? I did not find anything wrong in the simulation....

 

[keith1200rs]

If the simulations don't match what you have built then one of them is wrong - it is just a case of finding out which one. Sometimes it is the models, sometimes the stimulaus around the models, sometimes you haven't built EXACTLY what you simulated. What are the voltages on the opamp inputs?

Keith.

 

[Kinshoro]

got this from TI:
For a Sallen-Key section: For Q > 1, op amp gainbandwidth
should be at least 100 × Gain × Q^3fn. For
Q ≤ 1, op amp gain-bandwidth should be at least
100 × Gain × fn
Q=order of filter

The filter I am using is 2nd order and my bandwidth is 20 KHZ.... So does that mean it wont work...... I think according to this equation I need a very high bandwidth..... for a sallen key 2nd order filter...

Added after 24 minutes:

**broken link removed**

This website gave me a good idea of what the disadvantages of a sallen key configuration are and also some designing tips... they even have a component MAX 274 to be used as an active filter... Would this be good for my application...???

 

[keith1200rs]

Your filter will be Q<1. While their figures will give a response close to a perfect filter it all depends on how close you need to be. You are more likely to notice it in overshoot. It is something you can simulate. It is one reason for not trying to screw too much gain out of each stage. 10x to 30x is probably a practical maximum.

The TI calculation is fine, but in a practical application I would accept lower than 100xGainxfn, but I would check the effect on both the frequency response & pulse response.

For a 2 pole Butterworth high pass I would say the performance is likely to be acceptable at 20xGainxfn except in critical applications. Simulate it and see.

Keith.

Added after 11 minutes:

If you have Texas Instruments Filter Pro software (a free download) as well as helping design the basic filter designs it gives the required GBW for the filter as well.

I wouldn't recommend the MAX274. It is an old, power hungry, dual supply device. I have designed things like that out of products for companies. Most people now want low power, battery operated 3V only circuits.

Before you get too carried away with filtering - how much filtering do you need?

Keith.

 

[Kinshoro]

I tried using a mic with no integrated amplifier in it... and I implemented the design with first stage of amplifier with only the filter and unity gain..... and the second stage with an amplification of 10..... I dc-biased both the stages at half of the input voltage... around 1.5v..... It seems to be working perfectly.......... The problem is if u dont dc bias all the stages the dc offset becomes unstable at the output of the stage which is not dc biased..... Now the problem with the mems mic is that I dont think it has been dc biased or ac coupled inside... so I would have to take it apart so that my opamps can work with it....... Is there a better solution??? Can anyone look at the data sheet of my mic and suggest a way to dc bias my mics integrated preamp and ac couple it as well????? (mic model: sp0103be3)

Added after 9 minutes:

for a start I just need 2nd order filtering... Dont need a fast roll off... maybe later on I will need to increase the order for a faster roll off..............

Added after 3 minutes:

ya according to the current design of my low pass buttersworth filter with unity gain I would need a GBP of 2 MHZ.....

 

[keith1200rs]

Now the problem with the mems mic is that I don't think it has been dc biased or ac coupled inside

It certainly isn't AC coupled but it has a DC bias, just not a fixed bias. I do not understand what your current problem is. If you don't want to accept the DC bias of the microphone (which is a little low, and variable) then you need to AC couple the microphone to the next stage. That stage would then have to have its own DC bias point (presumably Vdd/2, but not necessarily). If you want to add an extra stage after that then you can DC couple PROVIDED you haven't introduced any significant offsets from offset amplification or the bias current across the feedback resistor, otherwise you need to AC couple again. You can share DC bias points with subsequent stages but you need to be careful not to do that across too many stages with significant gain otherwise you could end up with compromising the performance or in the worst case building an oscillator.

As I mentioned earlier, I have used the Knowles microphones (FG series) with no problems. Changing the microphone will not solve the circuit design problems you are having - the microphone isn't at fault.

It would help to see diagrams of the circuits that work and those that don't work. If you want to send me a PM, that may help to sort out your problems quicker.

Keith.

 

[Audioguru]

I have never seen a Sallen and Key Butterworth lowpass filter with a gain as high as 100. Usually their gain is 1 (the feedback C has double the value of the grounded C) or a gain of 1.6 (equal R and C values).

The filter must be fed from the very low output impedance of an opamp, not from a microphone.

 

[keith1200rs]

I have never seen a Sallen and Key Butterworth lowpass filter with a gain as high as 100. Usually their gain is 1 (the feedback C has double the value of the grounded C) or a gain of 1.6 (equal R and C values).

You can have any gain you like with a Sallen-Key filter design, but high gains can cause problems with stability, component tolerances and required GBW.

The filter must be fed from the very low output impedance of an opamp, not from a microphone.

The microphone has a built in amplifier.

Keith.

 

[Kinshoro]

thnx a lot keith... now it seems to be working fine with my mic...... man ur a life saver.......... I really appreciate all the help u have given throughout this week... I hope I can do the same for everyone that has helped me here in this forum...

 

[Kinshoro]

is it ok if I amplify a 100 times in the third stage... or should I amplify the signal first and then filter it???? which way is better?????? currently I am filtering first and then amplifying the signal......

 

[keith1200rs]

It depends on whether the frequencies you are filtering out my overload subsequent stages. I have done circuits where I have to filter very early because the interfering signal is considerably larger than the one I am trying to see. If I don't filter early it will clip the amplifier.

Keith.

 

[Kinshoro]

Is there a way to see the noise level characteristics in decibels vs the frequency in a simulation...... I actually need to find out how my filters are working and at what frequency my signal stops rolling off because of the high pass and low pass filter I have used.... I basically need to do an AC analysis of my circuit.....

 

[keith1200rs]

You can do a noise analysis (.NOISE). It will give either the equivalent input noise or output noise - I am not sure exactly what features your simulator will have. Usually you can get total output noise (as opposed to nV/rt(Hz)) as well.

Be aware that some opamp models don't model noise correctly so if you get silly answers that may be the cause.

Keith.

 

[Kinshoro]

I implemented my microphone(not simulation) and checked the FFT using an oscilloscope..... it seems the mic is not detecting any frequency below 63Hz.... Is there any way I can improve the sensitivity of the mic or its dynamic range....... basically the relation between the amplitude of the mic signal and frequency is in quadrature.... Hence the amplitude of the mic decreases considerably as the frequency of the mic signal decreases.....????

 

[keith1200rs]

Maybe you could post the latest circuit, but my guess is that you need to increase the capacitors to ground and any coupling capacitors to improve the low frequency response.

Keith.