LM2665: Shutdown mode

Hello TTCose,

Thank you for using our device and sorry that it does not behave as you expect it.

I will need to check the internal structure of the device, so please give me some time to come back with an answer.

I would be interested in your schematic, but you could as well point me to the figure in the datasheet that you are using without the diode.

Thank you for your quick reply, I attached the schematic.

Not putting the diode D1 was actually an oversight : I thought, quickly reading through the datasheet, that it was an application-specific diode, and that removing it would instead allow me to drive down the output of the voltage doubler, which is not possible with D1 (without crashing the input rail). I realized my mistake when carefully reading through the datasheet I saw that it was necessary for startup.

So I decided to try and turn on the device without it first, and add the diode afterwards along with the necessary patch to avoid crashing the input rail when driving down the 10V output.

Currently, the device works as intended by adding a 4K7 resistor on the 10V rail, allowing to keep the 5V rail up while driving down the 10V rail.

The purpose of the 10V rail is to keep a MOSFET grid charged, so while unexpected, the reduced current capacity is not too damaging to the function.

Edit : image wasn't showing up

Hello TTCose,

Thank you very much for the update and great that you solved the issue yourself.

I am a bit confused where you added a 4.7 kohm resistor, as I do not expect it is R86 on your schematic. As you speak about reduced current, you added the 4.7 kohm from the 10V rail to GND, correct? So it seems that there is a low current leakage path from VIN to VOUT that is now no longer charging the output cap, but gets dissipated through the resistor.

There is one paragraph (last paragraph of chapter 9.2.1.2.1) in the datasheet that mentions that the diode is necessary to prevent a potential latch up. I am sorry, but at the moment I do not know where this is coming from and what exactly this means.

Hello Brigitte,

I added the 4.7 kohm resistor at the output of the LM2665 (in series), after the output cap, to greatly add to the global output impedance. As there is (not shown) a MOSFET between the 10V rail and GND, the resistor allows me to crash the 10V rail and not the 5V rail. The image below will clarify.

The MOSFET gate that is driven by the 10V rail (the only purpose of this 10V rail really) has a low capacitance and there is no real-time constraint on its activation. 

For your information, the 5V rail is rated for ~500 mA. In SD mode, the 10V rail is at roughly 5V which means that the CMOS switches are probably closed and this isn't the fault of an internal diode/MOS intrinsic diode, because we would see some voltage drop.

Can you investigate on the internal structure of the device ? This would help identify the real behavior of the SD mode.

Thank you for your time so far.

Hello TTCose,

Thank you very much for sharing the schematic. The investigation will take some time and I am not sure if there will be a result on it.

Hello Brigitte, I added it right after our exchange, a destructive latchup would be unfortunate !

Thank you. Should I mark the thread as resolved ?

Hello TTCose,

If you think the thread is resolved, it would be very kind if you mark it resolved.