LM5017: Soft start unstable

Kuan Wei Chen (Kevin Chen)

Genius 9710 points

Part Number: LM5017
Other Parts Discussed in Thread: LM5107

Dear Team,

We want to implement the soft start circuit here.

But we are not sure why the Vo is not stable.

Base on the waveform, the issue is Vcc didn't regulate at the voltage which defines on the datasheet.

Do you have any ideas what is the root cause?

Kevin

over 5 years ago

0 Timothy Hegarty over 5 years ago

TI__Guru 63497 points

Hi Kuan,

There should be a VCC capacitor to GND (RTN) to stabilize the internal VCC regulator. See Figure 12 in the LM5017 datasheet.

Regards,

Tim

0 Kuan Wei Chen (Kevin Chen) over 5 years ago in reply to Timothy Hegarty

TI__Genius 9710 points

Dear Tim,

After adding the cap the Iout inrush is smaller as below:

So the Vcc pin waveform is ok?

I think it should be around 7.6V

The original waveform without soft-start circuit as below:

As you can see this one have faster Vout establish and higher Io spike.

But without Vout ripple.

This one is the zoom-in version of Vout after adding the soft-start circuit:

Do you think is because of need to fine-tune the compensation circuit?

Or what can we do to improve it?

Kevin

0 Timothy Hegarty over 5 years ago in reply to Kuan Wei Chen (Kevin Chen)

TI__Guru 63497 points

Kevin,

Yes, the VCC voltage should be 7.6V. Can you check that it operates correctly with the soft-start network disconnected.

Note also that you have downstream output capacitors, most likely with low ESR. These appear in parallel with R131/C301, so they will reduce and distort the triangular ripple generated and cause the SW node voltage to appear unstable. I recommend installing a type 3 ripple injection circuit to avoid this. you can use the LM5107 quickstart calculator to find appropriate values.

Regards,

Tim

0 Kuan Wei Chen (Kevin Chen) over 5 years ago in reply to Timothy Hegarty

TI__Genius 9710 points

Dear Tim,

For the Vcc pin it turns out to be the customer probe the wrong position.

For the compensation, why without the soft-start circuit the Vo is quite stable?

If the type 3 compensation is necessary, I think Vout will unstable with/without soft-start circuit.

Can you also implement the soft-start circuit on the EVM and compare the before and after Vout waveform?

Kevin

0 Kuan Wei Chen (Kevin Chen) over 5 years ago in reply to Kuan Wei Chen (Kevin Chen)

TI__Genius 9710 points

Dear Tim,

Attach the waveform.

0 Timothy Hegarty over 5 years ago in reply to Kuan Wei Chen (Kevin Chen)

TI__Guru 63497 points

HI Kevin,

The FB node is noise sensitive, so PCB layout is critical. Keep the FB trace as short as possible and the feedback resistors close to the FB pin.

Note that the current limit will engage if needed during startup of the LM5017 (just in case you need to depopulate the soft-start circuit).

Regards,

Tim

0 Timothy Hegarty over 5 years ago in reply to Timothy Hegarty

TI__Guru 63497 points

Also check that the downstream output caps are disconnected as they are shunting the ripple generated by R131.

0 Kuan Wei Chen (Kevin Chen) over 5 years ago in reply to Timothy Hegarty

TI__Genius 9710 points

Dear Tim,

Please check the input.

I am not sure what you mean for the R131 and C301 here.

Do you mean with the R131 and C301 will cause the output capacitors C502, C112 and C105 will be meaningless here, why?

It didn't have any issue when without the soft-start circuit.

複本 LM5017QuickStartCalculator.xls

0 Timothy Hegarty over 5 years ago in reply to Kuan Wei Chen (Kevin Chen)

TI__Guru 63497 points

Hi Kevin,

The output capacitors C502, C112 and C105 will attenuate the ripple generated by R131 and C301. The output ripple should be triangular, but your plot shows a more sinusoidal behavior (too much capacitive ripple and not enough resistive ripple).

Please install the type 3 ripple injection network as advised in the datasheet when using ceramic output caps.

Regards,

Tim

Power management

Power management forum

LM5017: Soft start unstable