Microcontroller PI

[hobbes_lives]

I'm curious if anyone has successfully built a PI using a microcontroller to sample the signal?

I have done some prototyping using a dsPIC30F4011. I'm using the PWM for the pulse timing, which works well, and the ADC for sampling. I have an LCD connected as well, but that is inconsequential.

The initial approach I am taking is to sample the signal directly, not using an integrator, and then calculate an exponential moving average of the samples within the microcontroller. A problem I've run into with this design is that the gain in the preamp stage(s) has to be higher than in an analog PI for the ADC to have a good signal to sample. The higher gain introduces many noise issues, much of which comes from the noisy microcontroller itself.

So, I'm curious, has anyone built a microcontroller PI that uses the ADC to sample the signal directly, not using an integrator stage? If so, how was this accomplished effectively?

[TheWizard]

I am... my avatar is the board...

[Qiaozhi]

I'm curious if anyone has successfully built a PI using a microcontroller to sample the signal?

I have done some prototyping using a dsPIC30F4011. I'm using the PWM for the pulse timing, which works well, and the ADC for sampling. I have an LCD connected as well, but that is inconsequential.

The initial approach I am taking is to sample the signal directly, not using an integrator, and then calculate an exponential moving average of the samples within the microcontroller. A problem I've run into with this design is that the gain in the preamp stage(s) has to be higher than in an analog PI for the ADC to have a good signal to sample. The higher gain introduces many noise issues, much of which comes from the noisy microcontroller itself.

So, I'm curious, has anyone built a microcontroller PI that uses the ADC to sample the signal directly, not using an integrator stage? If so, how was this accomplished effectively?

Have a look here ->
http://home.global.co.za/~trh/

[hobbes_lives]

I am... my avatar is the board...

I had noticed your avatar before. Are you sampling the signal from the preamp circuit with the ADC? If so, can you provide details?

Have a look here ->
http://home.global.co.za/~trh/

The GoldPic is not sampling the signal from the preamp. Imagine eliminating IC4, Q2, and supporting components, and then running pin 2 of IC6 to pin 6 of IC3. That's the type of sampling that I am trying to achieve.

[1843]

hobbes,
The project Miner is designed like the your idea; you can find it at pulsdetektor.de...
Also see Miner Integrator

[Aziz]

Hi hobbes,

your realistic doubts are well-founded regarding the A/D sampling directly. Even I would do this, if I could do it with a reasonable effort and money.

Bugwhisker did a simple poorman's sampling A/D converter with sample and hold capacitors and converting the samples then with built-in A/D converter slowly. Look into Universal PI Micro project.

You also need to multiplex different gain stages as the signal decays quickly to zero level. A digitally controlled amplifier gain must be done in this case.

Overall, I can not recommend this now: the A/D converters are very expensive, very slow, have low dynamic range and you would need more processing power. Maybe the Blackfin-DSP's could do this job in real-time!

Anyway, you can do it of course, but with loss of much sensitivity and dynamic range. Mixing the analog and digital part would bring more benefits today.

Well, I have decided to use the µC just for some very special tasks only:
- timing logic, window control
- multiplexing analog signals
- controlling integrators
- auto-tune and audio signal generation
- user-interface (mode key switches)

One µC will definitely not be enough for all this stuff. So I have to use many of them to do this job in real multi-tasking manner. Some parts are really very time critical and should not be interrupted by unimportant events.

Just divide and conquer!

Regards,
Aziz