INN2604 issue at power up

12 posts / 0 new

In an embedded power supply using INN2604, sometimes we see a problem that seem related to the auto restart state machine that doesn't work well. The problem causes the damage of the electronic board because the INN2604 doesn't regulate the output voltage during the power up (the outputs voltage run up 10volt over the nominal value) and the auto restart state machine runs for an undefined time loop.
The power supply has three outputs voltage. The 5V and 12V on one winding of flyback transformer are cross regulated, the 17V on auxiliary winding is not regulated. At power up there isn't load on 12V, on 5V around 50mA, and on 17V around 20mA.
Attached you can find:
- An oscilloscope screenshot of 12V and 17V outputs during the problem.
- The circuit schematic of the power supply block.

This issue seems occur more often at low input line voltage (around 180Vac), could it be related to UV sensing?
A solution that seems mitigate the problem is to increase C85 (the capacitance of the feedback network) from 470pF to 10nF. Could it be related to the feedback regulation network?

Thank you in advance for support.

첨부 파일Size
Schematic118.39 KB
screenshot1.1 MB

Hi daniele,
Could you please share the primary drain switching waveform during the auto-restart ON time event?
Please also share the transformer design documentation.

In the attached screenshots you can see 12v output at CH2 and Drain voltage at CH1. There is a screenshot that represent the whole problem and two that zoom the drain waveform at start and end of the abnormal regulation.
The transformer is a standard component myrra 74088. Attached you can find the specification sheet.
Best regards

74088-Datasheet.pdf143.35 KB
whole_issue.jpg133.63 KB
issue_start.jpg127.75 KB
issue_end.jpg127.96 KB

Hi Daniele,
Can you please capture waveforms similar to the one I attached and then zoom in the waveform where there is first skipped cycles. After 1 ms of continuous switching, there should be some 2-3 switching cycles skipped (30-60 us of no switching) indicating secondary takes over control from primary. We need to know if secondary takes over control or not since over voltage condition like yours occurs when secondary fails to take control and why is it failing to take control.

Waveforms.png216.27 KB

I suppose that message above is a spam...

However, attached you can find more screenshots related to the problem.

Ch1 - BPP
Ch2 - Drain voltage
Ch3 - VOUT
Ch4 - BPS

During the abnormal regulation seem not to be skipped cycles, I zoomed and scanned all the window, so the secondary seem not to take controls.
I attach also screenshots during normal operation.
Please consider that in bench test I can reproduce the problem more often when I cycle power near the undervoltage threshold.

Attached you can find also an extract of board layout.

Thank you for support.

Hi Daniele,
Was the waveform I sent not visible and went as a spam?

Anyways your waveform 4 is what I was looking for. Once the BPS voltage reaches ~ 4.4 V, the secondary takes over control. This is happening almost after 1.25 ms of start up. I cannot see the initial start up pattern at the moment when the BPS voltage reaches 4.4 V in the waveforms you sent for the abnormal case. Can you please zoom in at this point where the BPS voltage reaches 4.4 V for the abnormal condition so I could know how is the drain switching pattern at this moment?

I'm sorry, yesterday there was a spam message between ours...
Attached you can find the requested screenshots. Sorry for bad resolution, but you can see that the startup event occurs a long time before the issue. At that time the BPS is already at 4.4V.

scope_0.png52.38 KB
scope_1.png27.85 KB
scope_2.png35.14 KB
scope_5.png39.15 KB

Hi daniele,
If you observe in scope 2 and 4 captures, there is almost 50-60 us of no switching activity at the moment when BPS reaches 4.4 V which means that the secondary did take over control.
Does this still happen if you disable Line UV/OV protection?
Is your FB trace close to any noise trace? Can you share your layout ?
How much is the FB pin voltage prior to the triggering of the Auto-restart both for good and bad cases?

I already shared the layout, you can find it in message #5. There aren't noisy traces near the feedback line, but from capacitor to feedback pin, the trace is quite long.
Measured with oscilloscope the voltage at feedback pin seems good, and reflect correctly the output voltage by the resistive divider.
I will try to disable UV/OV protection and so I'll come back to you.
Many thanks.

With input undervoltage protection disabled I can't reproduce the problem.
I have a question. What happens if the feedback pin voltage run up over the Vfb threshold before the BPS reach the 4.4V? It could happen in cases of very light load. During an event such that, if the primary trigger an autorestart (for UV event i.e.), could the regulator lose the secondary control?
Any suggestion is appreciated.

If the feedback pin is running above the Vfb threshold before the BPS reaches 4.4 V, that means it can trigger OV as secondary has not taken control yet. Secondary takes control once the BPS reaches 4.4 V. An auto-restart can trigger if the switching activity stops for ~1 sec if the output voltage is way beyond the required regulated voltage. Is that what you are seeing?

I'm not sure, but I'm doing an hypothesis regard the fact that the only solution that work seems to be to increase feedback capacitor to 10nF. In that way perhaps the Vfb run up slower than BPS... Is there any contraindication doing that? I don't see any overshoot during load changes, so I think that 10nF could be suitable.
Another question. In the new hardware release we can connect the FWD and VOUT pins to 5V secondary winding instead of 12V one. Could it be recommended, in order to comply with the maximum VOUT pin voltage of 15V?