Mic vs Line level

Hello ELEC23, 

Where do I start?... 

First there needs to be a good understanding of voltage levels and measurement standards. The datasheet shows the full scale voltage as an RMS number which is an industry standard way to report it, but it is not really helpful for figuring out the headroom and clipping point. You really need to look at the peak or the pear-to-peak value to understand if the signal will exceed the power supply rails and clip. By the way, the clipping point is not always the power supply rail, often it can be less than the power supply rail. You have to read the datasheets carefully for the converters and the OpAmps. 

This calculator will be handy for all the analysis: Utilities:VRMS/dBm/dBu/dBV calculators | Design Center | Analog Devices

The 1.9Vrms * 1.414 will give you the peak voltage from what you can call a zero audio signal. 

1.9Vrms * 2.828 will give you the peak to peak voltage of the entire sine wave from the most positive peak to the most negative peak. This enters into the discussion that for audio, we need the numbers to be positive and negative. So the converters will sit at a mid-scale voltage level we call the "common mode" voltage. That is the zero crossing point of the sine wave so the signal will go positive and negative relative to this common mode voltage. This would be a good topic for a video lecture... I will note that for the future. 

The common mode voltage on the AD1937 is 1.5V. So a sine wave will go above 1.5V + the Peak level and will go below 1.5V -  the peak level. Clipping on the positive and negative sides will not be the power supply rails, it will be greater than zero volts and less than the 3.3V for the power rail. 

The reason I mention all this is to understand where clipping happens and to be able to calculate the peak positive and negative values. 

You are correct that most pro audio interfaces will clip at around +24dBu. Some will clip a little less like +23 or +22dBu. So what is the voltage levels?

If you use the calculator, a +24dBu signal is 17.36V peak!!! The is almost 35V peak to peak!! This means your input or output amplifiers need to be able to handle or reproduce a signal that is around 35V. So you probably need a power supply rail of +/-18V. That gives you 36V for headroom if the amplifier will go to the rail. Now you know why most pro audio gear uses 18V rails and some will use 20V rails! 

Clearly the ADC with a full scale voltage level of 1.9Vrms cannot handle a signal that high. So you need to pad it down with an input stage. There are some applications circuits in the datasheet on page 32 and also I would look at the user guide for the evaluation board to see how that was designed. That one uses a +/-12V to power the OpAmps so it will not be able to go to +24dBu. Getting into the design of these interface stages is really beyond the scope of this post. There are simulators and design help available and you should be able to see examples on our site and on the internet. 

Your questions about mic level. Yes, a microphone level is roughly -36dBu is what I use on the Audio Precision to test mic preamps. Most preamp gain pots will allow you to get to clipping and the minimum gain setting will be about -10 to -15dB below what the preamp would consider the standard operating level. I usually use +4dBu as the standard. Most of our PGA amplifiers will only go to around 33dB of gain. Therefore, a -36dBu signal plus 33dB of gain will not get you to +4dBu. Some of our codecs have another stage where there is a +20dB amp you can engage. That is helpful. In other devices the PGA goes into a DSP and in the DSP you can apply more gain. There are tradeoffs. The PGA at max gain will have a low input impedance and that can load down the microphone and change the sound. (frequency response) If I were to be designing a high level pro audio preamp I would want a better analog front end that has at least 60dB of gain available. 70dB would be even better if you want to use something with a low output like a ribbon mic. The PGA internal to our parts are designed to work well with an electret microphone for speech applications. So keep that in mind. 

I hope this all helps. 

Dave T

DaveThib said:

This would be a good topic for a video lecture... I will note that for the future. 

Hi Dave,I hope you doing well. Dave is there any news about this video?I look forward to it.

Hello ELEC23,

You can keep poking me for this one. I have a couple of others that I need to do first. I make most of these videos on my own time and things have been insane. Summer was busy doing work around my house getting stuff done before I had shoulder surgery in mid July. I am getting back to normal now, I moved my mouse back to the right side of my keyboard! But, now it is fall and I teach at two different universities in the fall so that is occupying my weekend time and two evenings a week but I will get to it. 

Dave T

Hi Dave.I hope your shoulder is fully recovered and I'm sending you positive thoughts for a full and speedy recovery. Dave I was re-reading your answer and now I have a new question. When we speak about +24dBu in professional audio device, its about differential values or not? I mean when we want to connect a +24dBu to an ADC input of a codec for example AD1937 we have to design a attenuator circuit for reduce 17.36 to  1.9Vrms/2 for each pin or +24dBu already is for a differential signal and we have to reduce +24dBu/2 to 1.9Vrms/2 for each pin of ADC? Thanks for your helps.

Hello ELEC23,

Thanks for your kind words. It is going well, I am being great at doing my exercises and workouts every morning before I make my smoothie! It takes about a year to fully recover but it is doing really well and I am do almost anything with my right shoulder at this point. 

You are correct that is you look at it from an individual pin's perspective the voltage will be half. I would treat it as a differential signal and design an H-pad. If you sort of take apart an H-pad you divide the shunt resistor in half ( the middle of it is virtual ground) and then you will see each individual pad reduces each pin by X voltage so the full H-pad reduces by 2X. Or you can say 1/2X and X. 

If you use two single-ended pads then you involve ground which can cause ground loops. It is best to keep it differential as long as you can. 

Dave T