PI Power Supplies

[CAS]

Is there any reason why in the MPP and a few other PI designs I have looked at, the voltages seem to reference the Battery + ? Is it possible to use a more traditional approach and use a inverting switching regulator or charge pump to generate the negative voltages?

Also when using charge pump/switching reg, if a low noise version is used, does it have to be synced to the TX oscillator? The reason I ask is I am looking at varying transmit pulse widths/frequencies and would prefer not to have to sync if possible.

[Davor]

Traditionally the flyback switching transistors were either NPN-s or n-channel MOSFETs. Also traditionally the signal switching JFETs were more common n-channel. Analogue ground is always in a middle between the positive and negative rail. With choice of key components as mentioned above, the most power consumption happens between a negative rail and a ground. Therefore it is only practical to put batteries there, and create a positive rail by a charge pump that may be somewhat frail compared to a negative rail, and therefore much easier to filter properly.
Of course, you may put any other arrangement, but they'll all be a bit more needy.

In case of a radical change in design (H-bridge Tx drive, bilateral switches, rail-to-rail opamps...) you'd need a completely different approach to a power supply

[Qiaozhi]

It's simply a matter of convenience. In fact, if you look at PI designs 1 to 4 in the book "Inside the METAL DETECTOR", you'll see that they all reference the -ve supply line.

Syncing the charge pump to the TX oscillator is good practice. By synchronizing the two oscillators, any noise is effectively removed by the sampling integrator.

[CAS]

I had figured that maybe it was to do with the current required to drive a coil when using a N Channel Mosfet and trying to keep the noise as low as possible, but was just exploring other possibilities. I had a look at those in ITMD and saw that they referenced -ve but figured that was more due to simple design than based on performance.

If the charge pump is synced, would there be an issue if the TX pulse width is being varied?

[Davor]

Preferably any charge pump transitions will not happen during sampling. It is not the case with minipulse right now, as the charge pump switches too fast, but I'll see what's needed to fix it.