Flyback PSU - Instability Problem

• May 9, 2017

• #1

E

ElecDesigner

Member level 5

Joined

Jul 22, 2014

Messages

94

Helped

4

Reputation

8

Reaction score

5

Trophy points

1,288

Activity points

2,190

Hi,

I am testing a Flyback convertor that uses the LM5021 IC.

I have an issue where at some load / Vin combinations every other PWM pulse is a different width. I am using the IC variant that allows 50% duty cycle max and as far as I can see , for example, one pulse might be close to 50%, the next 30%, then 50% and so on.

I wouldn't have thought it would be the Voltage feedback loop but I have slugged that anyhow, with no effect. I am not an expert but it sounds like the sort of thing you might get with Duty Cycles above 50% (requiring slope comp to fix it).
As the controller I am using is limiting the Duty cycle to less than 50% I am confused as to why this is happening.

 

• May 9, 2017

• #2

KlausST

Advanced Member level 7

Joined

Apr 17, 2014

Messages

25,113

Helped

4,866

Reputation

9,753

Reaction score

5,529

Trophy points

1,393

Activity points

168,039

Hi,

we have about no chance to help you with the given informations.

We need YOUR schematic with all the voltages and device values and device names.
Also we need an exact description at which situations you see the unstability and at which situation it is stable.
Input voltage, current. Output voltage, current.

Klaus

 

• May 9, 2017

• #3

E

ElecDesigner

Member level 5

Joined

Jul 22, 2014

Messages

94

Helped

4

Reputation

8

Reaction score

5

Trophy points

1,288

Activity points

2,190

Sorry for the frustration with lack of information.

Regarding schematics - I keep to a minimum for confidentiality reasons.
I was hoping for some generic pointers on the effect rather than something specific.

It seams to be happening at some window where the duty cycle is not near the 50% limit but not narrow either.
The its designed to work at about 170V min, sat at 175Vin the instability happens between 8W and 15W (30W is full load).

I attach a screenshot from the scope. Yellow is Gate Drive, blue is current sense. The current sense has some bias on to keep it running continually at light loads (see datasheet) but removing that has no effect.

View attachment TEK0045.BMP

 

• May 9, 2017

• #4

KlausST

Advanced Member level 7

Joined

Apr 17, 2014

Messages

25,113

Helped

4,866

Reputation

9,753

Reaction score

5,529

Trophy points

1,393

Activity points

168,039

Hi,

from the picture (please in future use the "Insert image" button) it seems..
* the indutor is not saturating (good)
* you have a current mode controller (OK)
* PWM stops at the desired current (good)
* but it starts with different current (why?)
.. I have no clue what may cause this..

Other users may be more familiar with LM5021...

Klaus

 

• May 10, 2017

• #5

E-design

Advanced Member level 5

Joined

Jun 1, 2002

Messages

2,014

Helped

507

Reputation

1,016

Reaction score

408

Trophy points

1,363

Activity points

14,193

Check your transformer calculations and design. It looks like it goes into CCM mode. The application note says it should remain discontinuous.

 

Attachments

• 
  ramp.png

  41.7 KB · Views: 109

• May 10, 2017

• #6

E

ElecDesigner

Member level 5

Joined

Jul 22, 2014

Messages

94

Helped

4

Reputation

8

Reaction score

5

Trophy points

1,288

Activity points

2,190

E-Design
Indeed it is operating in CCM (I have thought about doing a DCM version for comparison).
I read that bit in the application note as a choice for that particular design, I don't see that the controller can only operate in DCM.

KlausST
You point about the start current is interesting. Its difficult to determine "cause and effect" here. Where the current ramp starts at a higher value, it is due to the fact that the DC was higher on the previous cycle giving less diode on time (which makes sense).

 

• May 10, 2017

• #7

E-design

Advanced Member level 5

Joined

Jun 1, 2002

Messages

2,014

Helped

507

Reputation

1,016

Reaction score

408

Trophy points

1,363

Activity points

14,193

Can you show the current waveform at a duty cycle where you don't have the sub-harmonic switching?

It is difficult to help you if you can't share more details of your actual design unfortunately.

 

• May 10, 2017

• #8

E

ElecDesigner

Member level 5

Joined

Jul 22, 2014

Messages

94

Helped

4

Reputation

8

Reaction score

5

Trophy points

1,288

Activity points

2,190

Doesn't look too much different apart from the obvious.......

 

• May 10, 2017

• #9

E-design

Advanced Member level 5

Joined

Jun 1, 2002

Messages

2,014

Helped

507

Reputation

1,016

Reaction score

408

Trophy points

1,363

Activity points

14,193

Without seeing your diagram and layout, I am not sure how we can assist you. It may be a design problem, a layout problem, control loop problem, or a combination with other things.

 

• May 10, 2017

• #10

FvM

Super Moderator

Staff member

Joined

Jan 22, 2008

Messages

52,362

Helped

14,743

Reputation

29,768

Reaction score

14,086

Trophy points

1,393

Location

Bochum, Germany

Activity points

297,824

Subharmonic oscillations are a well-known problem of current mode converters operating in CCM with > 50 % duty cycle.

LM5021 has a slope compensation feature to suppress subharmonic oscillations. Did you utilize it in your circuit?

 

• May 11, 2017

• #11

E

ElecDesigner

Member level 5

Joined

Jul 22, 2014

Messages

94

Helped

4

Reputation

8

Reaction score

5

Trophy points

1,288

Activity points

2,190

FvM,

Yes I read about subharmonic oscillations in such circumstances.

I went for the version of the LM5021 that does not allow a duty cycle of greater than 50%, hence it does not have the slope comp feature.

That said, I done some further work with a lower inductance transformer (covering more of the line/load range in DCM mode) as an experiment. The oscillations do appear to be related to CCM mode or the DCM/CCM boundary.

 

• May 11, 2017

• #12

FvM

Super Moderator

Staff member

Joined

Jan 22, 2008

Messages

52,362

Helped

14,743

Reputation

29,768

Reaction score

14,086

Trophy points

1,393

Location

Bochum, Germany

Activity points

297,824

Classical subharmonic oscillations shouldn't occur with < 50 % primary duty cycle. But the current waveform in post #3/#5 is somehow unusual. What causes the plateau current?

 

• May 12, 2017

• #13

E

ElecDesigner

Member level 5

Joined

Jul 22, 2014

Messages

94

Helped

4

Reputation

8

Reaction score

5

Trophy points

1,288

Activity points

2,190

I have attached part of my schematic and part of the 5021 datasheet.

The plateau current is caused by my R33 biasing the current sense signal to prevent skip cycle mode. I have tried removing R33, the waveform changes to have a more classical current sense ramp look about it but the oscillation still occurs.