Noise from PWM Audio circuit affecting Hi gain INA126 circuit

Hi Jerry,

I don't believe your schematic was successfully attached so you may need to try again. It may also be helpful to see the layout of your PCB to determine if placement and routing can be improved.

PWM can be nasty! The fast edge rates on the PWM waveform have a habit of capacitively coupling into sensitive circuits all over a PCB. Also, it is unlikely that the INA126 has any remaining CMRR at the high frequencies associated with fast edge rates on a square wave. 

Some thoughts on routing your PCB: Keep the sensor and INA126 as far away from the speaker and amplifier circuit as possible on the PCB. I would place the INA126 as close as possible to the sensor and it may be beneficial to use low-pass filters on the inputs of the INA126 and the following stage (ADC or otherwise). Be sure to run the input traces to the INA126 as close together as possible and ideally surround them with ground plane. 

Is the speaker on the board or does a connection go off-board to connect to it? Shortening the connection between the speaker and the amplifier circuit, reduces the ability of this circuit to radiate EMI.

Also, it would be best to configure the MSP430 so that the PWM output and any ADC inputs you are using are on opposite sides of the chip.  

Thanks John,

I did most of the things you mentioned. I replaced the speaker with an 8 ohm resistor and I get the same results so its not being radiated from the off board speaker. I think its through the power supply because I  noticed I get similar disturbance if I use a power supply instead of the 3v battery supply. When I increment the supply from say 3.0 to 2.9 or to 3.1 quickly, I see a similar pulse on the output of the INA126. I have bypass caps of 0.1uF close to the chip. The speaker circuit and PWM output is on the opposite side of the board. I'll resend the schematic and ground layouts tomorrow. The 5V analog looks clean and stable, but with a gain of 700, I wouldn't expect to see much on the scope.

Thanks

Jerry,

In high gains (e.g. 700) you are MORE likely to see effects on the scope, changes in the power supply cause a change in the input offset voltage, this is multiplied by the gain of the amplifier.

0.1 uF capacitors may be ineffective at such high frequencies (most likely above their self-resonant point and they are inductive), consider placing a smaller value capacitor (0.01uF C0G) in parallel with these capacitors. These capacitors are also only as good as the impedance they see to ground so a ground plane is really crucial and not just a trace. 

I'm having trouble uploading the files. It says they are uploaded but then they dont appear on the post. Obviously doing something wrong. As far as the ..01u bypass, I know that and tried that and it made no difference. I'm not sure it's high frequency noise. Anyhow is there another way I can email the files.

Thaks

Jerry

looks like only one file each post

3d file

Jerry,

There are a couple things I should point out. First, the input signal to the INA2126 is being applied as a common-mode voltage (same polarity at each input), instrumentation amplifiers are designed specifically to reject common-mode voltages and the INA2126 will not produce an amplified output signal in this configuration:

Second, are the long blue traces or the areas outlined in dotted blue lines the grounds? I ask because I do not see a connection between the ground areas and the ground pin on P2.  In either case there will be a very large amount of inductance in the ground return path of the INA2126. This will limit the effectiveness of power supply decoupling capacitors and also has the potential to radiate and receive EMI. For a design such as this one with a crowded board that includes both low noise analog and high current PWM,a multilayer board with a continuous ground plane is really preferred (single ground for both analog and digital). 

john,

i am not trying to amplify the signal, but looking for an output when there is a minute changes in the input. C11 and C13 inegrate the input signal and dont change as quickly as the minus input to the INA126. This circuit works as it should. The input to the voltage follower is basically a bridge and the output sits ast 2.5V

 Second yes all the grounds shown on the top left of the schematic are connected, just not shown on the schematic. The traces for grounds are very large and direct almost 1cm wide. The cost on this product prohibits multilayer boards.

HELP PLEASE! How can I move this entire post t0 the Power Management forum. I think it might get more reponse there.

Thank You.