UCC256303: UCC25630-3 Gate Driver LATCH OFF

Dear

nothing was trigged

high&low side driver shut down and don't restart again.

until VCC was restarted that it can work again.

the waveform

Ch1-> LOW SIDE DRIVER

Ch2->VCC

Ch3->resonant current

Ch4->High SIDE DRIVER

@10mS base

@1mS base

What are you talking about?

There isn't latch off function in UCC25630.

Why does it happen the situation?

Don't only ask us  searching Datasheet?

You don't know what to happen in my project.

I had said that it didn't happen in the unpretection.

I have double-checked the problem.

Can you carefully watch my word and answer my question?

RVCC & VCC always hold on the same the voltage like above waveform.
I have double-checked the probem.

Driver suddenly stops working and never working again.
Unless VCC is under UVLO, It can work again.

Your resonant current changed to very high before stop switching. What happened at that time of circled - is there anything shorted, over loaded?

You can try to use EVM at light load and shut down the IC .
You will see it.

Even if it rings, it still attain to OCP POINT.

CH1 LOW SIDE

CH3 Resonant current

CH4 HIGH SIDE

The waveform shows possibly RVCC is nearby UVLO OFF so turn off both drivers. RVCC off at typical 7V. Your waveform shows that from either driver output.  The waveform still needs further zoom-in and with better resolution. Have you checked RVCC as I asked before? If RVCC ok, please find out why both drivers show around 7V or even < 7V?

You are a Ti Employee.

What do you ask me about such question?

If I knew the problem, I didn't go to the forums to search for the answer.

Can you show a professional and  don't say  nonsense?

Can change anyone to answer the problem,if you don't know the problem?

Please show RVCC together with both driver pulses at switching stop area, that is add RVCC into the below waveform you showed. Please also show better resolution for the waveform since the waveform shows both driver outputs are near 7V which is  nearby RVCC UVLO OFF. Please use better time scale so we can tell what is RVCC and if RVCC is ok why both drivers outputs are at 7V nearby. Please provide these so we can see if any hints. 

Please also zoom in the below which is copied from your 1ms/div scale. We need to see if the wide pulse is always there or just there before switching stops. It looks you can use 200us/div to re-capture the waveform. The purpose is to check if the wide pulse is always there - so if 100us/div cannot make it clear, please use use zoom-in to show. Please add RVCC to the waveform capture. We need to confirm RVCC ok.   

CH1 LOW SIDE

CH2 RVCC

CH3 Resaont current

CH4 HIGH SIDE

This time the waveforms are different from the last time - can you capture the same waveforms at the same operation condition? Is this time what you show actually a normal power off? But anyway, please show capture the waveforms at the same conditions as above.

What's is different form two waveform?
It has latched off ,too.
Though it's not at 7V , it still stops working.

Can you give me any reason to explain the situation?

What do you mean that I don't understand?

I had provided you with waveform which you wanted timing.

Even if it wasn't consistent, it was latched off.

Can you clearly describe what you want?
I will provide it

Add RVCC into the below with your oscilloscope horizontal scale 20us/div, or 50us/div. The below waveforms do not show driver peak clearly. I need to know how much peak voltage on each drive outputs as well as RVCC at the same time.

@ skip mode

the waveform before stop working

it's normally switching.

But next skip cycle  stoped working.

Just make sure you will provide what I am asking:

Please add RVCC into the below with your oscilloscope horizontal scale 20us/div, or 50us/div. The below waveforms do not show driver peak clearly. I need to know how much peak voltage on each drive outputs as well as RVCC at the same time.

I can't capture the same waveform now.
But its condition is the same.

Don't focus on before waveform.
Can you use the last waveform to solve it?
It has being the same situation and I has checked that no protection was happened.



My problem is simple what it had been latched.
The Vcc isn't UVLO.

If you don't know the problem, can change anyone ?

The below is based on your new waveforms The three red circled areas with pulse peak < 7V. Can you zoom in these area with RVCC? I need to know why these three areas pulse peak that low.

I have a standby power to provide .

Schematics

6237.Schematic Prints.pdf

not at skip mode

It still stops working.

1) You use an external power supply to bias UCC256303, right?
YES
2) Please provide the operation conditions when the switching stop issue occurred, such as input voltage, output load, etc.

It's random, but it often happens at light load .
if it's not at skip mode, it sometimes happens.

3) Your schematics show you are using a load sharing device, so please let me know when the issue occurred, is UCC39002 in operation with the other converter, or the issue happened without parallel with the other converter?

I used to remove it ,and the issue happened .

at single power

4) The schematics show LLC ENB, can you remove Q103 and re-test to see the issue still present, so to exclude pulling down falsely by Q103?

I used to try to do the way ,but still shut-down .

As these are confirmed, one possible reason we can think is ISNS pin signal gets multiple zero crossings which can confuse the internal function. Can you probe ISNS zero crossings along with LO/HO latch off? If you do see multiple zero crossings, you can try to increase C126 in your design (may also need increase a bit of C114) to eliminate the multiple zero crossings to see if the issues can be resolved. 

I guess that you mean to  increase the capacitance on Isne pin.

I had tried to increase the value of capacitance before you made the remark.

It can improve the situation, but it still sometimes happens.

If the value is too large, power won't start up.

I try to capture Isne waveform but it is plenty of noise on it. 

I think that it may be have other problems in it.

The waveform is normally operating.

It still has a plenty of noise on Isns.

How come it doesn't be generated latching function?

Maybe the  noise  is captured by probe.

Do you have other ways to solve it ?

CH1 Isns

CH2 LOW SIDE

CH3 resonant current

CH4 HIGH SIDE

What can I do for eliminating multiple zero crossings?

Increasing capacitance is no use and it will effect my system steady.

You can see the waveform which isn't zero crossing.

But Driver stops working.

CH_2 LOW SIDE

CH_3 RESONANT CURRENT

CH_4 HIGH SIDE

If increase capacitance cannot be sure to eliminate ISNS signal multiple zero crossings, the next that can be is to review the layout to see it can be improved. The ISNS should not have multiple zero crossings which needs to confirm.

Our layout is following the datasheet's suggestion.

Here we can get only two solutions which are checking datasheet and layout.

Can you give me another more better solution?

You said that latched situation was appear to result from  multiple zero crossings.

But it isn't now.

What dose other problem be able to  cause latch off ?

Zero Crossing should be used during died time.

The waveform can watch the section no zero crossing, but it is stopping work.

There is a PFC to supply steady 400V, and it works well.

CH1 LOW

CH2 ISNS

CH3 RESONANT CURRENT

CH4 HIGH

layout.rar