Beachmaster Digital

[jeep6789]

The magnification of two 6200 is 1000 times. In this case, the output waveform is very shake, far exceeds the accuracy of the 16-bit ADC, I don't know how you deal with it, I think the pre-level enlargement is 100 times behind the ADC detection,

[Olly]

Good, you will be satisfactory

Hoping to be a bit better than that

[jeep6789]

Good, you will be satisfactory

[Olly]

Your ADC detects an AC signal, why not choose to detect DC signals?

Ah, I think I understand what you're asking! You want to know why I'm sampling the flyback pulse directly rather than sampling the integrated flyback window.

What I'm trying to do is to get an idea of the flyback slope and to then hopefully extract some indication of the target tau. With this information combined with the RX coil's amplitude and phase, I'm hoping to perform rudimentary discrimination.

If you check out George Overton's book on the "Voodoo" project it gives an enormous amount of information on this technique which he pioneered. I'm trying to do something similar but with direct sampling.

[jeep6789]

I'm sorry, I don't understand the question ???

Your ADC detects an AC signal, why not choose to detect DC signals?

[Olly]

You can let everyone learn a lot, I want to know why choose to detect communication, not by electronic switch, integrator detection DC, is it to simplify the circuit? Can you tell you what kind of aspects should be written from?

I'm sorry, I don't understand the question ???

[jeep6789]

You can let everyone learn a lot, I want to know why choose to detect communication, not by electronic switch, integrator detection DC, is it to simplify the circuit? Can you tell you what kind of aspects should be written from?

[jeep6789]

The -12V is only used directly by the PI coil, all critical supplies are regulated.
It is a fact that as the battery supply drops then the TX current achieved for a specific pulse width will drop slightly, but this is easily compensated for by automatically increasing pulse width according to monitored voltage.

You are really very powerful

[Olly]

There is also a very important issue, 12V in the circuit. No voltage regulation, will not affect the stability of the machine when using

The -12V is only used directly by the PI coil, all critical supplies are regulated.
It is a fact that as the battery supply drops then the TX current achieved for a specific pulse width will drop slightly, but this is easily compensated for by automatically increasing pulse width according to monitored voltage.

[Olly]

Ok, thank you for explanation, in all PI detectors, why everyone likes to use D ? 5V to supply the op amp supply, why not directly use 3.3V to provide power supply? 6200 lowest voltage is 2.5V, If we use 3.3V power, can this?

In general better SNR is achieved with higher supply voltages and most opamps don't swing rail-to-rail so +/- 5V gives a bit of headroom.
In this specific design, I need the +5V to match the amplifier output to the 5V range of the ADC.

[jeep6789]

There is also a very important issue, 12V in the circuit. No voltage regulation, will not affect the stability of the machine when using

[jeep6789]

Ok, thank you for explanation, in all PI detectors, why everyone likes to use D ? 5V to supply the op amp supply, why not directly use 3.3V to provide power supply? 6200 lowest voltage is 2.5V, If we use 3.3V power, can this?

[Olly]

I think 7905 should be replaced with low noise LDO because it has a large noise, less than 16 precision.

I think we'll be OK - the L79L05 has 40uV output noise, but the LT6200 has 68dB PSRR which is around 2500x so even at the 1000x gain of the 2 x LT6200's the output noise as a result of the L79L05 will be 40uV/2500 X 1000 which is only 16uV.

All things being perfect the bit resoltion of a 16-bit ADC with 5V reference is 5/2^16 = 76uV, so more than 4x the noise from the L79L05.
If this was our only noise source I might consider attending to it, however the Johnson noise from the 1K resistor R14 amplified by 1000x is orders of magnitude greater so in essence the supply noise is not an issue.

Kind regards

[jeep6789]

I think 7905 should be replaced with low noise LDO because it has a large noise, less than 16 precision.

[Olly]

Thanks for the tips Waltr, I will certainly take great care with the grounding and component placement to minimise large circulating currents.
The RP Pico has programmable PIO's and DMA which I plan to use for the ADC SPI. This way I can easily get the data in and stored away in RAM without any intervention from the main program. It also has two cores each running at 125MHz (or 133MHz with a bit of overclocking - but I don't normally use this) for when I need to process the results.

I agree that a bit of oversampling is required to bump the noise down. 16x should give 12dB noise reduction.

My plan is as follows: -

At the falling edge of the TX pulse take 64 samples at 1uSec intervals and DMA the results to RAM.
Do this every falling edge and add the results to an array such that we have a 16x rolling average of each of the 64 points on the decay curve.
We should have more than enough horsepower in the Pico to do this as well as enough RAM (16 x 64 x 2) bytes = 2K. (The Pico has 264K internal RAM and 2M for program storage)

This should leave us with a nice clean decay curve representation for analysis.

Cheers