LM76003-Q1: Strange behaviour during input transients

Yannis Karatzaferis

Part Number: LM76003-Q1
Other Parts Discussed in Thread: LM76003

Hi,

I am trying to evaluate LM76003-Q1 during load/input transients and I am observing a strange behaviour, during input voltage transients.

My goal was to test 16V->32V transients and 32V->16V transients. The problem occurs when the input voltage is dropping. Here is what the response should look like:

The above response was for an input step response from 27V to 16V. Channel 1 is input voltage and channel 2 is output voltage (AC coupled).

Now here is what happens when I try to increase the input ΔV:

Output voltage is dropping by almost 1V, which is huge.

Examining the switching node at the same time reveals a strange behaviour. At first, LM76003 is responding well, as you can see the duty cycle is increased as the input voltage drops. But at some point it just gives up and leaves the low switch turned on permanently:

I have probed all important pins (Vcc, RT, FB) and I didn't find anything strange. I am using the LM7600xEVM board, where I have replaced the LM76003 with LM76003-Q1. I am using FPWM and setting 500kHz frequency with Rt=80k6. For feedback, I am using Rft=100k, Rfb=24k9 and a 10pF Cff. Everything else is configured the same way as in the LM76003EVM board.

Any ideas why this is happening?

Thanks

over 5 years ago

0 Frank De Stasi over 5 years ago

TI__Guru 73515 points

Hello

Based on your FB divider I assume you have 5Vout set.

What is your inductor value? and output capacitor?

Can you take a scope shot of the VCC along with the ones you presented here.

Thanks

0 Yannis Karatzaferis over 5 years ago in reply to Frank De Stasi

Prodigy 50 points

Hi Frank,

Yes, it is a 5V Vout setup.

L is 10uH and Cout is 89uF (4x22uF+1uF+0.1uF)

Unfortunately, I didn't keep the Vcc scopes, but it was quite stable (I didn't notice any significant change at 50mV/div during the transient).

I will record it next week, we only visit the workplace/lab when its necessary.

0 Frank De Stasi over 5 years ago in reply to Yannis Karatzaferis

TI__Guru 73515 points

Hello

One more question: What load current do you have when conducting the test?

Can you test at different loads.

Thanks

0 Yannis Karatzaferis over 5 years ago in reply to Frank De Stasi

Prodigy 50 points

Hello,

This happens under all load conditions, I usually test under no load, 0.2A and 1.5A load.

I am back at the lab today, I have a scope of Vcc, AC coupled @50mV/div. Nothing strange there.

It is not resetting, as this would mean it would soft-start again, right?

I am running out of ideas...

0 Frank De Stasi over 5 years ago in reply to Yannis Karatzaferis

TI__Guru 73515 points

Hello

Try the same test with the original, un-modified EVM; this will give you a baseline.

Also, try to slow down the edges of your input voltage.

Thanks

0 Yannis Karatzaferis over 5 years ago in reply to Frank De Stasi

Prodigy 50 points

Hello,

With smaller dv/dt it works fine, however the dv/dt corresponds to iso7637-2 transients and internal standards, so no point in slowing it down.

The design is not that far from the original EVM, most changes were done to replace X5R capacitors with X7R.

I can try switching back to the original 6u8 inductor but I doubt it will make a difference.

Thanks

0 Frank De Stasi over 5 years ago in reply to Yannis Karatzaferis

TI__Guru 73515 points

Hello

Using the original EVM is a good way to check to get a baseline of performance.

I believe that the input slew rate is contributing to the performance you are seeing.

If you can not slow down the edge rate, then a filter on the input will be needed to slow down the

transient that the device sees at its input rail.

Thanks

0 Yannis Karatzaferis over 5 years ago in reply to Frank De Stasi

Prodigy 50 points

Hello,

Yes, this a solution I will be trying.

I was just wondering why this is happening or if this has happened before.

It just seems strange. Also, when the input transient is positive there is no problem at all, even when dv/dt is much higher.

Thanks

0 Frank De Stasi over 5 years ago in reply to Yannis Karatzaferis

TI__Guru 73515 points

Hello

Sometimes the fast edge of the input voltage can pass through to circuits that are

not filtered by the internal LDO. This can cause unexpected behavior.

Thanks

Power management

Power management forum

LM76003-Q1: Strange behaviour during input transients