LM5007: LM5007 turn off issue

Jayce Luo,

VM must be rising after being unloaded which would cause the IC to turn back on. The IC would try to regulate which would pull VM back down which would shut off the IC again, unloading VM which might make VM rise again.

The datasheet shows VCC has a shutdown voltage of 6.3V with a hysteresis of 206mV and a 3us delay. Check to ensure the VIN waveform does not operate in a way to turn the IC back on.

-Sam

Hi, Samuel Jaffe, 

The VM is turned off using an air switch.

There is a big capacitor in front of the VIN pin, and thus when the UVLO is triggered, the MOSFET turns off and then there is no quick route to release the charge on the capacitor.

Therefore, the VM remain around 6.3V and easily trigger the wrong start process.

Two questions to consult：

First, How can solve the problem: release the input capacitor charge after the MOSFET is closed? Divided into 2 situation:

1. The pre-stage is a DCDC.

2. The pre-stage is an air-switch.

Second, What's the potential danger will cause if this problem can not be solved?

Thanks and Best Regards,

Jayce

Besides, can you give an example that how can the shutdown function of LM5007 be easy realized? Is an analog switch necessary? Then how to supply power for the Analog switch? And is a control signal from MCU necessary to control the Analog switch? I feel it is very complex to realize the shutdown function.

By the way, I can not find your name in our email system, could you send me en email so that I can have your email address?

Jayce,

First I'd like to confirm that my theory is correct. Can you please share a waveform showing VM zoomed in to confirm it's rising again.

Please also share the full schematic (showing the big capacitor and any other relevant circuitry. Do you only have 100nF on VIN?) and the layout. We may be able to fix this by adjusting the schematic or layout.

-Sam

Jayce,

I have not heard back. Please let me know if the issue is still happening.

-Sam

Hi Sam,

Sorry for late reply. I am sure there is a big capacitor in front of the input, but I cannot give the more detail schematic or layout information right now (still need to confirm with customer).

However, I made some tests based on the LM5007 EVM Boards and find similar problem.

I add a 100uF capacitor in the input side, follow by 1^st LM5007 EVM Board (set resistor 8.5k/1k for 21V output), then follow by 2^nd LM5007 EVM Board(set resistor 2k/2k for 5V output).

The power source is DC regulated power source (GPS-3303C from GWINSTEK, 90W max output power).

I use two ways to cut off the power supply:

the First one is move the positive power head quickly away from the DC power source to simulate the behavior of an air switch,

the Second is turn off the DC power source (the DC source can be considered a 100 ohm resistor load when turn off).

Result for the First way as below, can see no rising of input waveform but LM5007 restart for 4 times. (ch1 and ch2 ref levels are overlapping)

Result for the second way as below, no restart process.

Current I doubt whether the restart process is caused by the Minimum Load Current or other Restriction?

Please help analysis. Thanks.

Best Regards,

Jayce Luo

Attention that the Second way have a sharper down speed of input voltage (release charge by the DC source resistor)

Jayce,

This is interesting that it's happening on the EVM as well. I've contacted the product expert to help discuss the root cause of this behavior. He will help here and we can continue to discuss on email as well.

-Sam