LM2596: Output voltage overshoot after removing short circuit at the output

Hello

Can you test at different input voltages and see if there is any difference.

Also, can you show us images of your PCB layout.

Thanks

Hi, sure

layout with feedback path

24V

15V

12V

8V

Hello

Can you put a scope on the input voltage and make sure your input is not falling during the short and recovery.

Thanks

Hello

Measure A:

yellow trace is CIN Voltage, 1V/div @ 12VDC nominal, voltage drop is ~0,7VDC during short circuit condition

blue trace is Regulated Output Voltage

Measure B:

at the input PIN 1 node, with short loop probe, I see ~2,5V Vp-p high high frequency during short circuit condition

Measure B setup:

Hello, I performed another test, I shorted again the output (at the output capacitor) and measured this node, PIN2 "Output":

This is the measure, switching frequency ~156KHz:

Is this normal behavior for a short circuit output?

Hello

The switching behavior looks normal for a short circuit.

Can you try to parallel the input capacitor with a small ceramic such as 1uF or so.

I have ordered an EVM for this device and will try to test in our lab.

Thanks

Hello, I am waiting for the LM2596S-ADJEVM, I ordered it a couple of days ago; I am also waiting for some known good parts from your TI store.

Hello

I just received the LM2596S-ADJEVM with 5VDC fixed output; the short circuit condition is removed and there is no load at the output, this is the resulting oscillogram

the overshoot settling time is reduced if 200mA load is applied with an electronic load, but it is still there

I am waiting for parts LM2596S-3.3/NOPB 

Hello

Looks similar to your results.

When we get our EVM we will test in our lab.

Thanks

Hello

finally I received known good parts LM2596S-3.3/NOPB and tested on my board, here the resulting oscillogram

~6VDC is way above any absolute maximum ratings for the 3,3VDC nominal supply voltage of the devices attached to this power supply rail.

In my application, the Ethernet IC fails after a short circuit on the 3,3V power supply rail.

Could you please confirm this behaviour on your EVM ?

Hello

We will check this in our lab as soon as possible.

You may need a Zener diode clamp in your application.

Thanks

Hi Bob,

I'm also seeing similar behavior under a short circuit transient on the EVM - as Frank mentioned, using a fairly high power rated zener at ~3.6V will help.

Alternatively, if you're not locked in to the LM2596, you could consider a different/newer part - for instance, the TPS62933 datasheet has a plot of the short circuit recovery (https://www.ti.com/lit/ds/symlink/tps62933.pdf), which has virtually no overshoot. There are lots of other TI solutions for the operating conditions you mentioned, you can find them all on Webench.

If you need more help, feel free to continue the thread!

Hello , 

considering part number LM2596S-3.3/NOPB is fixed 3,3V DC output,  could you suggest zener topology protection with some reference Zener P/N for testing?

Hi Bob,

You could try placing a SMBJ3V3-E3/52 in parallel to COUT and see if it stops the high transient. 

Thanks,
Andrew

Thank you Andrew for the suggestion, I will test and share the results

Hello, 

I have tested your solution, part number SMBJ3V3-E3/52 , Manufacturer Vishay Intertech

unfortunately, no change

Hi Bob,

This is with the diode reverse-biased across VOUT - GND?

This is the topology, Cathode connected to +VOUT, Anode connected to GND = in parallel to Output capacitor

Hello Andrew, 

no change:

1k load at the output:

Has anybody actually tested the clamping circuit before suggesting?

Any other suggestion, different than changing IC family, is most welcome at this stage.

Best regards

Hi Bob,

Yeah, I think the voltage surge voltage you are seeing is a bit too low for the zener diode which is why it's not catching the surge. You could try a zener diode designed for a lower voltage, but there will be leakage at 3.3V.

You could try implementing an external soft start circuit like below.

slva307b.pdf

The reality is that there's not really an easy work around for this issue that doesn't involve a bunch of external components. The issue is that the device doesn't turn off in the short circuit condition which is why there's a huge overshoot. This is quite the old part, and I'm sorry that this issue is a huge problem for your application specifically, and I'd be more than happy to help transition you to another part.

Thanks,

Andrew

Hello Andrew, 

your last response is very useful; if the suggested SS circuit could solve my problem in a new PCB revision, it's nevertheless a solution.

I will definetly test this

Thank you for your suggestion

Best Regards