Senior Design Project - Audio Buffer Question

Status
Not open for further replies.

Monkeyman87

New Member
I am currently working on a senior design project that involves sampling an audio signal from an portable mp3 player using a TI TMS320F28023 DSP/MCU.

Using a unipolar 3.3V supply, I need to convert the input signal 0Vdc, ~200mVpp (ac coupled through 10uF cap) to a 1.6Vdc, 3.2Vpp signal with low impedance (several hundred ohms).

I attempted to offset the input signal with a voltage divider and then amplify the signal with an inverting op amp circuit, but this is not working. I'm assuming because of the unipolar supply.

I also tried an non-inverting level shifting amplifier, but again this gives me some strange results.

Does anyone have any ideas on a simple way to do this?

Thanks,

Sam
 

Google 'single supply opamp circuits'. For single supply opamp circuits, buffers, amps, filters etc.. since you're dealing with AC, they require a bipolar power supply, so, a 'virtual ground' is created at half the supply voltage. For a very simple voltage follower this would be exactly as you first tried....a signal, via a capacitor, connected to the positive input of an opamp. The positive input is biased at VCC/2 via a voltage divider, the parallel resistance of which determines the input impedance. So, using two 1Mohm resistors, gives an input impedance of 500k.

At 3.3v (very common now days for analogue) you'll need an opamp which can run at this voltage. Look for 'common input range' and output range, you may not need a rail-to-rail input opamp, as you're only feeding it 200mv p-p ...after the capacitor, thats 100mv above and below the bias voltage, 1.6V meaning its 1.5-1.7V. This means the opamp you use must be able to input voltages within 1.5V of the power rails (most opamps can handle this).

However, as you need 3.2V p-p on the output, the output of the opamp must be able to swing within 500mV of the power rails. So...its should be a rail-to-rail output opamp. Otherwise it clips horribly and distorts your signal.

Ther alternative to all the above? Use a higher supply voltage for the opamp. This will give the output headroom, allowing it to be the full 3.2V p-p as required. I could list some opamp part numbers, but frankly, just use what you have to hand. Check the specs for slew rate and bandwidth to make sure it can deal with the bandwidth of your signal (a simple LM358 isn't that good for audio). Then just power it with 5-12V. Still a unipolar power supply, but you can worry about getting hold of a R-R opamp later when you want to run the whole thing from the same 3.3v line.

*removed stuff about sigma delta ADC's, as your chip doesn't have one!*

I'm sure TI have provided documentation with example circuits for the use of their ADC's for this range of DSP's.

Blueteeth

Edit: just checked the datasheet for that chip, seems its not a delta sigma ADC, so you may need a higher pole filter for antialiasing...at least if you're after accurate sampling.
 
Last edited:
Thanks Blueteeth.

I do have a 3.3V rail to rail opamp available (AD8541) so I will try to use a virtual ground. I have done quite a few searches for buffer circuits and looked at TI's ADC interfacing document. The problem was trying to level shift and amplify the signal at the same time. I think I will just use to op amp stages to accomplish this.
 
Hey,

The AD8541 looks pretty much ideal for the job, its bandwidth isn't all that but a quick look at the datasheet, seems its good up to 100kHz

Whast you must remember is, by using an opamp with avirtual ground (single supply) you are level shifting. The output of a single supply voltage follower is centred at VCC/2. In order to amplify this, a simple non-inverting amplifier sohuld do the trick.

I have attached a pic from LTspice simulation showing a very basic rail-to-rail opamp, with 200mV p-p input signal (no DC component offset) and its output. The voltage gain you need is 16, here I have used 12, because the opamp I simulated with can only go within 200mV of the supply rails, and so it would clip.

As you can see the output is centred around VCC/2 (the virtual ground) which is 1.65V. If youre opamp clips at sucha low voltage, you could power this circuit with 5V, then remove the DC offset form the output (which would be 2.5V), then re-bias the output at 1.65V.

Please note, for sampling with an ADC you'll need an antialiasing filter (lowpass filter with a sharp cut off before half the sample rate).
 

Attachments

  • opamp_buff1.png
    12.7 KB · Views: 601
  • opamp_buff2.png
    21.2 KB · Views: 317
Status
Not open for further replies.
Cookies are required to use this site. You must accept them to continue using the site. Learn more…