Rework power supply scheme for the meter probe

[dkonigsberg]

Extensive testing has shown that the root cause of all the issues which have led to #56, #60, and #64 may actually be related to the meter probe itself. This realization was arrived at because these particular USB glitches only seem to occur when testing with the meter probe. They do not seem to occur with any other USB peripherals, including the evaluation board for the controller IC (FT260) used in the meter probe.

First, some basic issues were noticed, which should be addressed:

• The bulk capacitor on VBUS should be increased in size. (FT260 eval board uses a 4.7uF tantalum, whereas the meter probe currently uses a 4.7uF MLCC. It would be a better idea to simply use a 10uF MLCC here.)
• The RESETN pin has a smaller resistor and a capacitor on it in the recommended schematic, whereas it only has a 10k resistor on the current meter probe. For stability, the components on the meter probe here should be changed to match the recommended schematic.

Bodging around these two issues in bench testing did improve resistance to one of the easier power transient tests (lamp transformer on unfiltered AC supply), but did not help against the more difficult tests (AC motor or EFT generator). Regardless, implementing these changes is a good idea.

Then, upon further inspection, the likely root issue was noted. The meter probe runs the FT260 I/O voltage at 1.8V, based on the example on page 34 of the FT260 datasheet. This was done because the TSL2585 sensor needs 1.8V, and results in everything on the surrounding board running at 1.8V. This includes the sensor, EEPROM, push button pull-up, and various FT260 pin pull-ups. This also likely means that the margins on the power rails and many of the pins are now rather small, and could be a lot more sensitive to glitches.

So the plan is to switch to a new power supply configuration, which is a much closer match to how the breadboarded version of the meter probe works:

• Incorporate all the changes noted above
• Switch the FT260 I/O voltage to 3.3V, using its internal regulator, as shown on page 31 of the FT260 datasheet
• Use the FT260 VOUT3V3 pin to provide a 3.3V rail for the EEPROM and the push button
• Keep the 1.8V regulator mostly separate, directly tapping the filtered VBUS 5V rail and providing 1.8V to just the TSL2585 section
• Leverage the fact that the TSL2585 has 3.3V tolerant I/O pins to avoid the need for any additional level shifting

This is essentially how the breadboarded version of the meter probe is configured, using the FT260 evaluation board and a dedicated breakout board for the TSL2585. Testing has shown that the breadboarded version is mostly immune to the power glitch. (it only fails on a repeated burst from the EFT generator, not normal pulses, which is more of a lab test than a real world transient, and one that can likely be handled in software.)

[dkonigsberg]

A new version of the meter probe with this change has been constructed, and is performing far better in initial tests. This issue may be closed shortly, pending a little bit more testing.

[dkonigsberg]

Updated (Rev F) meter probe with these changes is performing much more reliably. There really isn't that much more that can be tweaked here, so closing this issue for now.