LM5164: Stability test

Maor Matityahu

Intellectual 600 points

Part Number: LM5164

Hi,

I am using the LM5164 IC for fly-buck converter.

Power supply parameters:

Input voltage: 30-60V from batteries

Main Output voltage: 3.3V, 0.3W

Secondary voltage 1: 12V, 1mA

Secondary voltage 2: 14V, 0.5mA

Switching frequency set: 75kHZ

The main transformer has 3 windings:

Primary: 100uH

Secondary 1: 1444uH

Secondary 2: 1849uH

The output voltage starts to decrease after 0.3W at the 3.3V output

I am trying to test the stability using the bode 100 device, this is what I get for 30V, 0.3W:

SW voltage:

Output Voltage Rising:

From the SW voltage waveform it seems that the converter goes to sleep mode (this is what we want for light loads, because the supply is from batteries).

Because the converter goes to sleep mode, Is the bode plot result I see (in the first picture) is real? Is there a meaning to test the bode plot when the converter goes to sleep?

Thank you

over 2 years ago

0 Daniel Li14 over 2 years ago

TI__Mastermind 27470 points

can you share the schematic?

thanks

0 Maor Matityahu over 2 years ago in reply to Daniel Li14

Intellectual 600 points

Hi,

Thank you

0 Daniel Li14 over 2 years ago in reply to Maor Matityahu

TI__Mastermind 27470 points

1 as LM5164 is PFM model, pls make sure the primary current no go to 0.

2 if LM5164 work at CCM model, you can reduce C359 to 4.7nF and C360 to 10nF to see any improvement (it will increase the ripple injection to FB)

0 Maor Matityahu over 2 years ago

Intellectual 600 points

Hi Daniel,

1. Why do i need to make sure the current dont goes to 0? I want it to go to 0 so that i will be in sleep mode.

2. How did you get this capacitance numbers? I set it c359 to 10n and c360 to 33n and i get 3 pulses in the SW node until the IC goes to sleep for 1ms and than wakes up again. Is it ok?

3. Is it ok to check the stability with bode plot even when the IC goes to sleep?

Thank you

+1 Daniel Li14 over 2 years ago in reply to Maor Matityahu

TI__Mastermind 27470 points

1. if only buck model, no problem, but if you want to deliver the energy to secondary, you need to make the voltage on transformer. but when current of LM5164 go to 0, the transfomer will not able to deliver the energy or proivde the same votlage ratio of Vo_pri to secondary.

2.the ripple injection into FB node will be helpful to stablize , can you also open secondary winding to let it work at buck model?

3. no need to check the stability as its working frequency is pretty low.

0 Maor Matityahu over 2 years ago in reply to Daniel Li14

Intellectual 600 points

HI Daniel,

About 3, i don't need to check the stability even when i am increasing the load and working with CCM?

Thank you

0 Daniel Li14 over 2 years ago in reply to Maor Matityahu

TI__Mastermind 27470 points

Yes, you can test the loop but it may still looks strange , so I more suggest you can do the load transient to verify the loop stability too。

you can using fastest load transient to see whether output have oscillation at this moment.

Thanks

0 Maor Matityahu over 2 years ago in reply to Daniel Li14

Intellectual 600 points

Hi Daniel,

Is it because i am working with 75kHz? Why the bode will look strange in low frequency operation?

I checked the bode and i got 50Hz cutoff frequency, is it normal? This is what you meant by strange?

Thank you

0 Daniel Li14 over 2 years ago in reply to Maor Matityahu

TI__Mastermind 27470 points

I mean in middle frequency range, it looks not smooth. if you get proper phase margin and gain margin, then I think it is fine.

0 Maor Matityahu over 2 years ago in reply to Daniel Li14

Intellectual 600 points

Hi Daniel,

I changed the FB network because i got double pulses and the transformer entered into saturation and now i get one pulse and everything is ok.

I have 2 questions:

1. When i am working with 0.5W i don't enter sleep mode, and this is what i get in the bode plot:

I get 50Hz cutoff frequency, is this ok or i do need to get higher cutoff frequency? The PM is 102 and the GM is 62.

2. Why i don't have to check the loop at low working frequency (at light loads)?

Thank you

+1 Daniel Li14 over 2 years ago in reply to Maor Matityahu

TI__Mastermind 27470 points

1 if you whole system is stable at 0.5W, I think it is OK. can you share the SW waveform at this condition?

2. in CCM condition, worse case of its loop condition happened at full load. for DCM, the power stage AC model will changed, and cross frequency will become lower.

Thanks

0 Maor Matityahu over 2 years ago in reply to Daniel Li14

Intellectual 600 points

Hi Daniel,

Here is the waveform:

Yellow - SW pin

Green - FB pin

Red - primary current

The PM i get is a real reault?

Thank you

+1 Daniel Li14 over 2 years ago in reply to Maor Matityahu

TI__Mastermind 27470 points

The waveform looks good, for the loop, I have no idea why it is so low , but you can check the transient to see whether the loop response fitting for your system.

Power management

Power management forum

LM5164: Stability test