Skip to content

TXCO Overvoltage  #19

@bschulz1701

Description

@bschulz1701

Issue:

In most situations the TCXO on the logger experiences an overvoltage condition.

This oscillator is supplied by an ORed rail from BCKP_GPS and SYSCORE. BCKP_GPS is within spec, as the max battery voltage is 3V, however, SYSCORE is tied to the battery rail which can be up to 4.2V

The max recommended voltage of the oscillator (Si1552) is 3.63V, well exceeded by 4.2V.

While we have not observed an oscillator failure as of yet, this adds significant stress to the device and is an unacceptable operating condition. Also it can cause the oscillator to deviate beyond the accuracy specifications.

Retrofit:

On existing loggers (<= v1.7) this issue should be resolved via retrofit. This can be done by adding a 510kΩ resistor in series with the SYSCORE supply to the ORing system. This sufficiently limits the current to reduce the voltage applied to the oscillator to the specified range without impacting performance.

  1. Cut trace between via and pin 2 of D4
  2. Solder a 510kΩ 0603 resistor between via and pin 2 of D4

Potential Solution:

To fix this problem permanently a regulator could be added between SYSCORE and the ORing system. This linear regulator should have as low of a quiescent current as possible. The regulator should drop the voltage down to 1.8V in order to prevent impact of dropout voltage, even when the battery voltage is low. In this case, due to the extremely small current draw (~2.5uA max), efficiency of the regulator is almost irrelevant compared to the quiescent current, so this should be the focus of optimization.

The TPS7A0218 may be a desirable solution

Design Log Entry:

Upon investigating I noticed that the RTC oscillator (Si1552) on Kestrel (<=MrK 1.7) had a topology issue which resulted in an over voltage under many conditions.

The oscillator recommended voltage limit is 3.63V

The oscillator is powered by an ORing of GPS_BACKUP and SYS_CORE

GPS_BACKUP is limited to ~3.1V, which is within the limit. However SYS_CORE is tied to the battery input voltage which can reach (and often is around) 4.2V. This exceeds all limits for the oscillator and causes significant concern

As of 5/2/2023 no oscillators have failed as a result of this over-voltage, however this is not an acceptable mode of operation.

Potential Workarounds:
Add a series resistor between SYS_CORE and ORing diode to drop voltage
Add a low power voltage regulator in series with SYS_CORE to deliver a fixed voltage
Should be a low voltage to account for drop out even at extreme range of battery, 1.8V would be ideal

Using series resistor as patch for v1.6 design

Required voltage drop = 4.2 - 3.63 = 0.57V

Expected max current draw from SYS_CORE is RTC backup draw (0.925uA) + OSC max (1.52uA) ~= 2.5uA

Min expected current = OSC min = 1uA

In order to sufficiently reduce the voltage in all situations, 1uA*R > 0.57 -> R = 570kΩ

However, practically, the current draw by the clock and oscillator begin to be limited by the resistance before this

In practice (given the actual current draws), 470kΩ is sufficient to prevent over voltage in all tested cases. To ensure a sufficient margin, 510kΩ is chosen

Metadata

Metadata

Assignees

Labels

bugSomething isn't working

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions