Hi,

I have a circuit outputting PCM modulated audio between 0->5V. I need to filter off the modulation frequency (62kHz), reduce the gain and apply a negative bias, so the output signal is -0.5V->+0.5V. Apart from the filter above the human hearing range, I want the circuit to be as linear as possible.

I have used an RLC filter to filter off the modulation circuit.
The input PCM and filtered output can be seen on my oscilloscope here:

I then used a single rail Op Amp design to apply the gain and bias. I only have access to the 5V power rail and Ground (no -5V). I used this guide to help work out values:
http://www.ti.com/lit/an/slyt189/slyt189.pdf
I'm using case 3 on the 3rd page of the document. I worked out that I'd need an inverting amp, as non-inverting created negative resistor values which are obviously not possible! Using these equations I created a circuit simulation. The full circuit simulation can be seen here:

As you can see the simulation suggests it should work. However the output I actually got was:

The scope is set to 0.2V per gridline. So the power isn't going below 0.2V and the bias is wrong (the gain is correct).
I tried playing with R1 as a start. It changes the DC bias, but the voltage still won't drop below 0.2V (it just cuts the waveform).
The op amp I am using is a Texas Instruments TLC082IP. It is designed for single power supplies and I have terminated the unused 2nd op amp properly. The data sheet for it is here: http://www.ti.com/product/tlc082

Any ideas where I could start to debug this issue. Am I using the wrong op amp. The datasheet suggests it would be ideal for the job. Mathematically my resistor values work, but maybe they could be a poor choice for the op amp.

Any suggestions welcome. Many thanks.

Paul

 

The simulation shows negative output voltage, now that will not be physically possible if the negative supply to the op-amp is only 0 volts.

The fact that your output voltage won't go below 0.2 V also suggests that the op-amp is hitting its extreme low output voltage level.

To make a useful output, I would bias the + input at 2.5 V, by making R1 the same 5.6 k Ohms as R2. Then I would put a series capacitor after the output, so it removes any DC offset. This will make a lower cut-off frequency, depending on the load resistance.

The other way would be to use something like an ICL7660 or MAX1044 to generate a negative voltage for the op-amp's negative supply pin, so it will be able to output voltages below 0 V and use your existing resistor values. The ICL7660/MAX1044 can convert 5V into -5V with a few extra components. The TLC082 will be fine with this, but your circuit won't be just a single-supply anymore.

 

Thanks v much. I will give this a try. I think this must be the issue.
I did try making R1 and R2 the same. This pushed the DC bias up, so I could see the whole wave form, but it was now between 0.2V and 1.2V. I thought R1 and R2 were supposed to create a 'virtual earth'. so the op amp can output negative voltages? I may have misunderstood this.
Anyway I'll give a ICL7660 or MAX1044 a try and see how I get on.
I do have an RC high pass filter on as a temporary measure at the moment. I want to remove this if possible as it's removing a lot of the bottom end of the audio (because it's only 6db/oct).
Paul

 

No op amp's output can swing below it's negative supply rail (0V, in your case) or above it's positive supply rail (5V, in your case). If you want your output to be symmetrical around ground, you have two choices. One is to use a negative supply, and the other is AC coupling (high pass filter). Since you don't have a negative supply, the AC coupling is your only choice. You can choose a cutoff frequency that will allow all your audio frequencies to pass unattenuated.
As a side issue, i don't see how your LC filter works. It is a notch filter centered at 29kHz. The 62kHz PCM will not be eliminated by it.

What is your audio bandwidth at the high end of the spectrum, i.e., is it 3kHz, or 10kHz, or 20kHz, or ???

What is the low frequency requirement??

 

I thought R1 and R2 were supposed to create a 'virtual earth'. so the op amp can output negative voltages? I may have misunderstood this.

Nope... the amp's output is limited to the supply voltages it is given.

Another idea is if you have a 10V (actually a little more) supply handy make a 0-10V sine, then run it thru a capacitor to get it to go +5/-5.

 

Thanks for the tip about the notch filter. Changed that out for a 3rd order filter with op amps.