V3-0 - Input Overvoltage protection improvement by Schottky
- The overvoltage protection of the diodes is not perfect. Recommended
not to change as no better alternative is known and the current
solution is good enough.
Discussion
- JGR: I think the 3.3V overvoltage protection with D2 diodes won't
work properly. According to the CDCLVC1102 datasheet, its absolute
input voltage is Vdd + 0.5V (limited to 4.6V maximum). The BAR66 has
a Vf of about 0.8 - 0.9V, therefore higher than the 0.5V
recommended. Because of this, the chip will be stresses when the
input voltage is 5V TTL, and "may cause permanent damage to the
device". I would recommend changing the BAR66 by Schottky diodes.
- NV: Big discussion in perspective… We already tried using schottky
diodes (HSMS-2802-TR1G) but were worse than the BAR-66. If anyone
has a better idea…?
- EB: This has been reported as an Issue and the resulting
recommendation was to have the schematics with the BAR66. See
https://www.ohwr.org/work_packages/607. Recommend not to change.
- JGR: I'm sorry but here I disagree. If a redesign is going to be
performed, this issue must be fixed with a high priority. The design
must avoid any cause of reliability decrease.
Here I think there would be three options:
1.- Using the BAT64 Schottky
diode:
http://www.digikey.es/product-detail/en/BAT%2064-05%20E6327/BAT64-05E6327INCT-ND/1287651
Maybe you can buy it and make a fast test to see if it works
properly and, if so, we can use it right now.
2.- If the BAT64 doesn't work and BAR66 must be used, you could also
generate a 2.5V (or similar) where connect it for biasing instead of
the 3.3V.
3.- For this pre-series and series, and based on the previous
option, we could change the D1 diode and generating 2.5V instead
P5V0. We should remove D2 and mount the D1 BAR66 diodes and change
the F1 by a resistor. Also we must to change the SENSE1 input to the
power monitor TPS3307. I know it, it's a more difficult choice.
- NV: We will try option 1.
- NV: The option 2 is only feasible on a new version.
- NV: For option 3, we cannot replace F1 by a resistor, because the
termination resistors (R7-R9) are after it. But what can be done is
to replace the zener diode (D1) by a 4.3V (for example BZV55-C4V3).
This will ensure a proper clamping at 5V. But this does not solve
our problem, because the CDCLVC1102 cannot accept more than 3.8V on
its input. The resistor R10 can also be increased (100Ohms) to limit
the current on the input of the CDCLVC1102.
- NV: If we plan doing a new design, we maybe can think about
replacing the CDCLVC1102 by another buffer that accept +5V on its
input.
- EB: This has been reported as an Issue and the resulting
recommendation was to have the schematics with the BAR66. The
conclusion of the report that the solution is good enough and that
Schottky diodes have a worse behaviour than the recommended BAR66.
See https://www.ohwr.org/work_packages/607. Recommend not to
change.