Sub-audio Modulator

Some respondents suggested a 10KHz carrier which seems fine. I could then nominate 1KHz as the cutoff. See my circut at the bottom of this post.

Here is my interpretation of the circuit you have kindly described above.

Any suggestions for improvement?

I understand that sound cards and MP3 players swing pos/neg.

One concern is that the DC biasing in the first stage would change whenever the volume of the MP3 player (that feeds in the signal) was adjusted by the user. There would be no way to tell without a scope.

One solution might be to design the recorded signal in software so the active part, in its entirety, occupies only the top half of the swing.

Would this work?

Mike Towner

The fundamental (heart beat) is around 1Hz, so all of the important detail is faster than that. I've never done ECG analysis, either, but it seems worth a try. Deconvolution with the measured response would extend the bandwidth a bit, if needed.

My bad. Your MP3 idea looks good

I am resending Mike's response to Dave Platt's detailed post in case he missed it. Mike and I are grad students at SCU working on the same project.

Sorry to everyone else for the duplication.

David King

(snip)

Here is my interpretation of the circuit you have kindly described above.

Any suggestions for improvement?

I understand that sound cards and MP3 players swing pos/neg.

One concern is that the DC biasing in the first stage would change whenever the volume of the MP3 player (that feeds in the signal) was adjusted by the user. There would be no way to tell without a scope.

One solution might be to design the recorded signal in software so the active part, in its entirety, occupies only the top half of the swing.

Would this work?

It looks as if the high-pass (DC-blocking) filter between the two stages has a 1.5-second time constant.

Is this long enough to pass an ECG waveform properly, without generating enough phase shift in the lower-frequency components to distort the waveform?

You might want to go up to a 470 uF cap, or even larger. A 1000 uF

Right... but that's often *because* they have a DC-blocking cap on their outputs. A lot of these players and sound cards have only a unipolar DC power supply (i.e. one battery, or a +5 voltage). The DAC chip automatically biased its output signal up to an average voltage of half of the supply. In a 5-volt DAC, the audio signal might come out of the chip at an average voltage of 2.5, and swing a volt or so in either direction.

Feed this through a DC-blocking capacitor, and you end up with 1 volt swinging on either side of ground - that's what goes to your amp.

That wouldn't matter, if you use a series capacitor to restore the DC average level to zero. The 100 uF cap in your schematic will do that just fine.

That's exactly what I was referring to when I said "biasing the signal".

Just take your desired waveform values (numerically calculated, or whatever), and add a constant to each sample, with the constant being sufficiently large to keep the (adjusted) samples from ever dropping down to zero.

Thanks Dave. Your help has been invaluable.

I will build the circuit up when we return to the lab next week.

David King

The circuit below includes the rectifier/LP filter that Dave Platt kindly assisted me with, plus two symmetrical amplifiers which I have added to drive solenoid coils.

Everything works fine up to the amp. But I am unable to obtain any output from the LM358 amplifers.

I suspect biasing, but since the design is slightly unusual, the solution is escaping me.

Can someone please look at the above diagram and tell me where I am going wrong?

Many thanks.

David King

This is not my forte, but the biasing circuit for the final two opamps looks like an awkward circuit. Are you getting the output you expect from the second opamp, pin 7?

What voltages are you trying to put on the two coils? Do you intend to have a DC bias on them?

The low pass filter has no ground reference, so the DC output voltage is ab le wander about. A 22k resistor from C2 negative to ground would settle it down.

What is the purpose of the two output amps ?

Can't say if a circuit will achieve an unknown one.

As shown, they cannot work linearly because there are TWO competing DC inpu t levels and the op-amps cannot drive enough current into the pin 2 junctio ns ( via 100 kohms) to correct this.

Normally, one applies DC bias to ONE op-amp input and lets the amp deal wit h the other via the feedback loop. The golden rule with of op-amps is the o utput always tries to make both inputs exactly the same voltage, your job i s to make sure that it is possible for it to do so - or else the output wil l swing hard up to one DC rail.

.... Phil

Thanks Phill. I will fix this as you recommend.

Each amp is driving a coil toward ground. One from the positve rail and the other from the negative. It is intended to produce two identica,l but offset, signals.

I will try changing the DC offset adjustment from the non-inverting to inverting input and see what happens.

Is there any way to isolate the DC input level of each amp from the other amp with a diode?