UC1846: Gate skipping operation of UC1846

Nandha Kumar5

Part Number: UC1846

Hi,

I have a poorly designed flyback converter which operates with gate skipping at low loads. As per my understanding of the IC's operation, gate pulses should occur at even multiples of the oscillator/clock pulses. However, I see gate pulses at 3rd/5th distance from the first clock pulse. Is this a problem with the converter. If yes, please explain the IC internal function which lead to this mode of operation.

Thanks

Nandha

over 5 years ago

0 Daniel Hartung over 5 years ago

TI__Mastermind 26305 points

Hey Nandha,

By gate skipping do you mean at light loads the device is in burst mode?

Burst mode is when the device will switch for a small period of time followed by a period of no pulses.

If this is what you are asking about, this is normal behavior.

Thanks,

Daniel

0 Nandha Kumar5 over 5 years ago in reply to Daniel Hartung

Intellectual 470 points

Hi Daniel,
I did not mean the burst mode. As you mentioned during the burst mode, the device will switch for a few times and turns off for a longer time and this pattern is repeated. In this burst mode, the switching frequency will be half as that of the oscillator. In my case, the pulses occur at random intervals (say 2/3rd of the switching frequency and etc.,) I will soon attach the waveform for reference. See if the description suffices till then

0 Nandha Kumar5 over 5 years ago in reply to Daniel Hartung

Intellectual 470 points

The waveforms as captured.

Ch1- Gate waveform

Ch2- CT pin of UC1846

Ch3 - CS+ Pin of UC1846

The converter is a current-mode controlled flyback converter operated in DCM.

0 Daniel Hartung over 5 years ago in reply to Nandha Kumar5

TI__Mastermind 26305 points

Hey Nandha,

This behavior is not considered "normal".

Are you able to try different values for Ct/Rt?

Thanks,

Daniel

0 Nandha Kumar5 over 5 years ago in reply to Daniel Hartung

Intellectual 470 points

I've used 1nF/7.5k as of now. Will get back to you after trying alternate values

0 Nandha Kumar5 over 5 years ago in reply to Daniel Hartung

Intellectual 470 points

Trying this has added more problems. I tried 5.6nF/1.3k for the oscillator. This set the oscillator frequency at 180 kHz instead of around 280kHz as expected.

0 Daniel Hartung over 5 years ago in reply to Nandha Kumar5

TI__Mastermind 26305 points

Hey Nandha,

Did the device still exhibit the skipping behavior?

Can you provide a full schematic at all? I can contact the email of your account if necessary.

The current waveform of the device looks strange, what are your current specifications of your transformer (peak RMS ect.)

Are you not using Bout at all?

Thanks,
Daniel

0 Nandha Kumar5 over 5 years ago in reply to Daniel Hartung

Intellectual 470 points

Yes, it still does. The peak current (designed value) is around 4A. I'm using a 0.1 ohm series resistor to sense the current. Also the Bout pin is open. Sharing the schematic might be a bit hard as of now

Also, Rt/Ct graph is a bit elusive to me. If I use a 5.6nF Ct, what would be the Rt value to get an oscillator frequency of 280kHz? The graph isn't very clear and by using the formula for 1.3k Rt, I get an oscillator frequency of 180kHz

0 Daniel Hartung over 5 years ago in reply to Nandha Kumar5

TI__Mastermind 26305 points

Hey Nandha,

What value are you using for EA+/- and what is the COMP voltage?

I am trying to make sure the device doesn't have a reason to skip a pulse.

Thanks,

Daniel

0 Nandha Kumar5 over 5 years ago in reply to Daniel Hartung

Intellectual 470 points

Hi Daniel

I'm using the Vref and dividing by 2 for the EA+ input. The EA- input is generated from a 15V output. Haven't checked the COMP waveform yet. Please clarify the axie values for the oscillator. The one in the datasheet is confusing and doesn't match with hardware.

0 Daniel Hartung over 5 years ago in reply to Nandha Kumar5

TI__Mastermind 26305 points

Hey Nandha,

What are "axie" values.

Thanks,
Daniel

0 Nandha Kumar5 over 5 years ago in reply to Daniel Hartung

Intellectual 470 points

Hi Daniel

Apologies for the typo. I was referring to the graph given in the application note (Fig6 of U-93). As I had mentioned before, I had used 5.6nF/1.k for the Ct/Rt and obtained an oscillator frequency of 180kHz as compared to the formula which gives 300kHz.

0 Daniel Hartung over 5 years ago in reply to Nandha Kumar5

TI__Mastermind 26305 points

Hey Nandha,

I agree the oscillator is not giving the correct frequency compared to the equation. While I do see the equation be off at times, this is a very large difference.

I am also unsure of why you are seeing the device not have the outputs skip switches.

I am unsure of what more I can do without seeing a full schematic.

Thanks,
Daniel

0 Nandha Kumar5 over 5 years ago in reply to Daniel Hartung

Intellectual 470 points

Hi Daniel,

Getting you the schematic might take me some time. See if you can come up with some explanation in the meantime

Thanks
Nandha

0 Daniel Hartung over 5 years ago in reply to Nandha Kumar5

TI__Mastermind 26305 points

Hey Nandha,

Were you able to get a schematic?

Thanks,
Daniel

0 Nandha Kumar5 over 5 years ago in reply to Daniel Hartung

Intellectual 470 points

Hi Daniel,
Not yet. The problem I face is similar to the one I see on this thread

e2e.ti.com/.../2539580

Would the same reasoning apply to the 1846?. I see similar internal circuitry after the comparator.

0 Nandha Kumar5 over 5 years ago in reply to Daniel Hartung

Intellectual 470 points

Hi,

I can share the block diagram for now

0 Daniel Hartung over 5 years ago in reply to Nandha Kumar5

TI__Mastermind 26305 points

Hey Nandha,

Its possible that what you linked is what is wrong, but if that was the case I would expect that to change when I asked you to change the Rt/Ct components.
The dead time is controlled by the Ct value, and when you changed the Ct value you didn't see any improvement.
If you want to experiment by making Ct larger again you would be able to tell if you were having the same issue.

Given the shape of the current in the original waveforms I believe something may be wrong at the schematic level, as the current should ramp mostly linearly up from 0, and the large positive and negative spike doesn't make much sense.

Thanks,
Daniel

0 Nandha Kumar5 over 5 years ago in reply to Daniel Hartung

Intellectual 470 points

Hi Daniel,
One major issue I traced was the the current sense resistor not being properly soldered. Rectification of this led to better performance. The IC now gives pulses on even multiples of the oscillator ramp. However, i still can't understand the IC's previous non-standard operation.

0 Daniel Hartung over 5 years ago in reply to Nandha Kumar5

TI__Mastermind 26305 points

Hey Nandha,

Are you saying that one output is doing it on every other pulse?
This would be expected as the outputs should switch off pulses

Thanks,
Daniel

0 Nandha Kumar5 over 5 years ago in reply to Daniel Hartung

Intellectual 470 points

Hi Daniel,
In response to your previous query.
No, what you said is true. It is to be expected that the outputs switch off after one oscillator cycle (which is what limits the ICs max duty cycle to 50%). What happened in my case was (refer wave form attached previously) the output was suppressed for two oscillator cycles before turning ON again.
Thanks,
Nandha

0 Daniel Hartung over 5 years ago in reply to Nandha Kumar5

TI__Mastermind 26305 points

Hey Nandha,

You mentioned a change in behavior and that now the devices were "turning off in even multiples"

What does that mean then?

Thanks,

Daniel

0 Nandha Kumar5 over 5 years ago in reply to Daniel Hartung

Intellectual 470 points

Hi Daniel,
I meant that the output (Aout) is now seen at even multiples of the oscillator pin. You can contact me on my e-mail if you need clearer information

0 Daniel Hartung over 5 years ago in reply to Nandha Kumar5

TI__Mastermind 26305 points

This has been taken offline.

Power management

Power management forum

UC1846: Gate skipping operation of UC1846