Power Interrupt Protection (holdup) with LT8228

Hi All,

I’m working on a power supply concept that can ride through short power interruptions using the LT8228.

Concept Overview:

• Under normal conditions, 18 V is applied at V.

• The LT8228 runs in boost mode, producing 56 V and charging a capacitor bank totaling 8.8 mF.

• When 18 V is lost, the LT8228 reverses to buck mode, supplying the load from the capacitor bank.

• Once 18 V returns, the LT8228 should switch back to boost mode.

The approach is inspired by the application schematic on page 64 of the latest LT8228 datasheet.

I’m using the EVAL-CN0556-EVZ evaluation board (see my setup below).
The switch is mechanical.
There is no load (i.e., R_load = ∞) during this test, giving ample time for manual switching..

5 Ohm resistor and diode in parallel makes the capacitor charging less problematic. 

Modifications:

I’ve changed R1N, R2N, RUV1A, and CSS values to adapt the eval board to my use case, see bellow:

Observations:

• With 18 V applied, the board properly switches to boost mode and charges the capacitor bank to 56 V.

• If I open the switch (disconnecting V), it transitions to buck mode and supplies 14 V to V2 as expected.

Problem:

If I reapply 18 V while there is still residual charge on the capacitor bank, the LT8228 enters a persistent FAULT state pulsing.

Please see the oscilloscope capture below. Traces are Labelled.

Scope Details:

At the trigger point, I disconnect V — V2 drops from 18 V to 14 V (as FB2 is set for 14 V).
Shortly after reapplying 18 V, the FAULT pin begins pulsing continuously.
It never recovers, after ~46 FAULT pulses I manually disconnected V.

Below is a zoomed-in view of the FAULT onset:

In the plot below, I marked when FB2 crosses 1.3 V and 7ms window later (≈1024 cycles at the converter clock 146 kHz)

We see that after reapplying the 18V V2 starts going up, so does the FB2.

Expected vs. Actual:

According to the LT8228 datasheet, the device should switch to boost mode 7ms after FB2 > 1.3 V. 
Yet in this case, it enters FAULT instead.

Important voltages during the measurement above:

• FB1 (V1 target is 56 V) is below 1.21 V.

• V1 under voltage set to 12.9 V, and the UV1 pin was above 1.21 V.

• DRXN constant 3.3V

I think this behavior doesn’t align with the datasheet’s mode transition diagram.

Question:

Can anyone help explain why the LT8228 is entering FAULT after VIN is reapplied while the output caps are still partially charged?
Any insight would be greatly appreciated.

Thank you.