UCC256403: Burst Mode Control for UCC25640x LLC Resonant Controllers

Hello,

Your inquiry is under review and I will get back to you shortly.

Regards,

Hello,

Please see my responses below.

<Question 1>

UCC25640x LLC Resonant Controller

SLUSD90E – JUNE 2019 – REVISED FEBRUARY 2021

In the explanation of "7.4.3 Burst Mode Control" on page 36, the bottom of the page says,

"The Pick Higher block continues to function by selecting higher values ​​of BMTL and FBreplica relative to the control effort Vcomp. In this case, BMTL limits the maximum switching frequency of LLC."

①What is the specific frequency that is limited?

> The minimum switching is just above the capacitor region.

 > The maximum is roughly 400 kHz.

②Also, it says that the condition for continuing control is to "select higher values ​​of BMTL and FBreplica." Is this selection a setting that can be controlled externally?

If so, please tell me the specific setting method.

> A resistor divider off the reference to the LL pin sets the threshold.

> The voltage at the BW pin not only is used for OVP.  However, it set the BTL/BTH ratio

The table below shows the BW voltage that sets the option of

BMTL/BMTH ratio.  The column on the right should have volts in it.

<Question 2>

Due to the circuit configuration, I need control that does not use burst mode.

>This can cause over shoot under light loads.

You would set the design to option 1 with the BW pin.

The current problem is that when the output voltage (-5% to -10%) is changed with almost no output load, the feedback control cannot respond at the control limit, and the FB terminal of this IC is sunk to GND by photocoupler control from the output, and the output voltage rises sharply from 4.9V to about 5.3V, becoming an unstable voltage.

A simple solution would be to connect the necessary dummy load, but this defeats the purpose of increasing conversion efficiency.

>If you do not use burst mode this is the typical solution.

So, is there a way to stabilize and maintain the output voltage within a certain accuracy, like forward PWM continuous control, without using the IC's burst mode control, even at the limit of continuous control with a light load?

>You could try speeding up your voltage loop to reduce the over shoot.

>You also could increase your output capacitor bank to lessen the over shoot and under shoot of larger load transients on the output.

If possible, please let me know the specific solution.

One last solution could be to add a clamp/shunt to prevent excessive over shoot.

Regards,

UCC256403 Question (2nd)

Hello, this is Kurata.
Thank you for your continued support.
In response to the question I received the other day, we adjusted the constants to make the control response of the prototype as fast as possible, based on your company's opinion that the countermeasure should be to "speed up the output control response."
As a result, the malfunction condition hardly changed.
My opinion on the test results is that the UCC256403 device's capabilities have reached a level where it cannot control light loads any further.
In the sudden load change test from 100% to 0%, it was confirmed that the control frequency rose to a maximum of approximately 312kHz (T≒3.2μsec), which is close to the maximum control frequency of 400kHz that you mentioned in your answer.
The results of this test showed that two states occurred: one where the transient response converged to the no-load steady-state value, and one where it did not (unstable voltage), and I feel that the minimum limit power where normal control is not possible has been reached at frequencies close to this maximum frequency.
When this state occurs, the input power consumption of the equipment increases by about four times compared to normal no-load conditions, even when there is no load.
The test results and observation data for the two occurrence modes are attached as a PDF file.
Please confirm the contents.
We would appreciate your company's opinion on this matter and, if possible, we would appreciate it if you could inform us of any improvement measures that would achieve a constant voltage accuracy of 5V ±5% for the UCC256403 control.UCC256403_5V10A output - Transient response waveform measurement during sudden load change.pdf

Hello,

Your inquiry is under review and I will get back to you shortly.

Regards,

Hello,

Your design is dropping out of regulation when the is changed from 100% load to 0%.    Your output voltage drops drastically.  

I don't think this is due to the burst mode not working correctly.  You even mention in your report that switching stops.

You might be triggering over voltage protection causing the design to stop switching.

The BW pin is what detects OVP.  

You should be able to determine if this OVP by monitoring the BW pin, output, LO, SS and output.  Trigger on the output and when it drops out of regulation look at these points right before shutdown.

Regards, 

UCC256403 Question (3rd time)

Hello, this is Kurata.

Thank you for your continued support.

Regarding your answer yesterday, you pointed out that an overvoltage may have been detected at the BW terminal. I observed it again under the same conditions, but no such "overvoltage" occurred. Please check the attached data.

Also, in your explanation of the answer, you wrote that the report we submitted contained a statement that the switching operation had stopped, but I don't think we explained that.

With the sudden change in load from 100% to 0%, I explained that "the FB terminal is rapidly sinking to GND due to the sinking operation of the photocoupler output due to the feedback of the control signal from the output."

I think this is the result of the output voltage rising to a level that exceeds the control reference voltage of the shunt regulator that monitors the output, resulting in negative feedback and sending a signal that causes the photocoupler output to sink. 　To repeat, I think that the UCC256403 device's capabilities have reached a level where it is no longer able to control light loads any further.

In a previous explanation, we said that in a sudden load change test from 100% to 0%, the control frequency rose to a maximum of approximately 312kHz (T ≒ 3.2μsec). However, observations have shown that in the process of changing to this believed to be the upper limit, the control frequency rises in stages from the frequency at rated load, from approximately 75kHz to approximately 140kHz to approximately 200kHz to approximately 310kHz, before reaching the maximum upper limit.

However, this behavior differs in that when it converges: output voltage rises to approximately 5.08V, and when it is unstable: it rises to approximately 5.30V.

When the control state does not settle to the no-load voltage of this static fluctuation and rises to 5.3V, isn't it nothing more than the control reaching the limit of the maximum frequency?

As a result, we imagine that depending on the slight differences in the conditions for detecting the sudden change, two possibilities occur: the control settles to the static fluctuation value, and there is also the possibility of it going over control. To repeat, when over-control occurs at no load, input power consumption is approximately four times that at no normal load, and when comparing the LLC primary current waveform in the attached data, items 3.1 and 3.2, the waveform is far from the clean SW current waveform when it does not stabilize and converge. This state is due to the fact that the peak level of the BW terminal voltage waveform noise in item 3.2 is significantly higher at no load than at rated load, resulting in hard switching, which we imagine results in a significant increase in switching loss and input power.

Please check the attached data.

If you have any questions about the above explanation or if there is room for improvement in operation, we would appreciate your reply and guidance.UCC256403 5V10A output_Transient response waveform measurement during sudden load change 2.pdf

Hello,

Your inquiry is under review and I will get back to you shortly.

Regards,

Hello,

Why is the BW pin sinusoidal on your report?  It should be a square wave.

Regards,

UCC256403 Question (4th)

Hello, this is Kurata.

Thank you for your continued support.

Regarding the point you made in your answer yesterday, I have found out why the waveform of the BW terminal is a sine wave, not a square wave. Previously, in our evaluation board test, we attached a large capacitor (0.01μF) to the BW terminal to eliminate any suspicion of overvoltage malfunction when adjusting the circuit constants. Due to the capacitance of this capacitor, the integral effect with the voltage divider resistor caused the waveform to become sine wave or triangular wave-like.

It was my mistake in the settings. I apologize.

However, this is not related to the malfunction of the overvoltage protection detection of the BW terminal this time, but rather the configuration does not react to momentary excessive noise. 　Therefore, I believe that there is no malfunction caused by noise intrusion into the BW terminal, which led to the concern from the other day.Transient response waveform of 5V10A output by UCC256403 control (3).pdf

UCC256403 Question (4th)

Hello, this is Kurata.

Thank you for your continued support.

Regarding the point you made in your answer yesterday, I have found out why the waveform of the BW terminal is a sine wave, not a square wave. Previously, in our evaluation board test, we attached a large capacitor (0.01μF) to the BW terminal to eliminate any suspicion of overvoltage malfunction when adjusting the circuit constants. Due to the capacitance of this capacitor, the integral effect with the voltage divider resistor caused the waveform to become sine wave or triangular wave-like.

It was my mistake in the settings. I apologize.

However, this is not related to the malfunction of the overvoltage protection detection of the BW terminal this time, but rather the configuration does not react to momentary excessive noise. 　Therefore, I believe that there is no malfunction caused by noise intrusion into the BW terminal, which led to the concern from the other day.Transient response waveform of 5V10A output by UCC256403 control (3).pdf

UCC256403 Question (4th)  Re-edit　

Hello, this is Kurata.

Thank you for your continued support.

Regarding the point you made in your answer yesterday, I have found out why the waveform of the BW terminal is a sine wave, not a square wave. Previously, in our evaluation board test, we attached a large capacitor (0.01μF) to the BW terminal to eliminate any suspicion of overvoltage malfunction when adjusting the circuit constants. Due to the capacitance of this capacitor, the integral effect with the voltage divider resistor caused the waveform to become sine wave or triangular wave-like.

It was my mistake in the settings. I apologize.

However, this is not related to the malfunction of the overvoltage protection detection of the BW terminal this time, but rather the configuration does not react to momentary excessive noise. 　Therefore, I believe that there is no malfunction caused by noise intrusion into the BW terminal, which led to the concern from the other day.

I have removed the capacitor in question and taken another photo of the operating waveform, so please check the BW terminal waveform again with the attached correction data.

Also, regarding the constant voltage characteristics when the load changes suddenly from 100% to 0%, which is the original issue, I would appreciate it if you could let me know again what the problem is and whether there are any solutions.

Thank you in advance.Transient response waveform of 5V10A output by UCC256403 control _(3).pdf

＜4th question – 2＞

This is my second question today.

In my initial question about "the problem of overshooting beyond the set value," I received two pieces of advice: "Set it to option 1" and "Speed ​​up the control loop." But what is the significance and mechanism behind setting it to the first option, Option 1?

There is no explanation in the data sheet, so I don't understand how this option works. Please advise.

Hello,

If you have questions regarding the UCC25660x family of devices.  Please repost to the e2e with UCC25660x in the thread title so the appropriate applications engineer can answer your questions.

Regards,

Hello Mike

Thank you for your help.

This is Kurata.

Thank you for your answer.

I understand your advice to consider LLC's new device UCC25660.

However, considering the time and cost we have spent evaluating this device up to now, we cannot easily switch to the device you introduced.

If we can clear the current issues, we can see the product coming to market, so we would like to complete the evaluation somehow.

Please cooperate.

<Question 1> About the control frequency

①In the Q&A the other day, you said that the maximum no-load control frequency is 400kHz, but what is the maximum frequency at which the output voltage stabilization can be controlled?

The current state of our prototype is as observed in the attached file the other day, and the maximum is almost 310KHz.

In this state, the FB terminal is pulled to GND by the photocoupler negative feedback signal, and is in an unstable (open loop control) state.

②However, the stable no-load control frequency in static load fluctuations at steady state is around 120kHz.

Therefore, it is unclear why the frequency cannot be controlled even if it rises to 310kHz due to a sudden load change.

Does the appropriateness of the input level setting of the VCR terminal, which captures the resonant voltage through a voltage divider resistor, have any effect on the upper limit of this maximum controllable frequency?

Please advise how to improve the control so that it does not latch up to high control frequencies and allows stable control.

Are there any measures to keep the convergence value of the control frequency at around 120KHz even during sudden changes?

<Question 2> Relationship between LL/SS terminal voltage and FB terminal voltage

After startup, the LL/SS terminal voltage seems to be fixed at 3.5V in the steady state, but what is the relationship between this voltage and the BMT threshold of the FB terminal feedback signal (source current) input?

I understand that the FB terminal input is controlled by the source current, but in continuous mode without using burst mode, what is the minimum control value of the FB terminal voltage due to the FB terminal source current at the minimum controllable output power?

<Question 3> Relationship between threshold and FB terminal voltage

Can the FBReplica threshold (BMTL, BMTH) at the minimum controllable output load be set from outside the IC?

Also, can the threshold be monitored and measured from outside the IC?

Furthermore, please tell me the specific voltage relationship between the FBReplica threshold voltage and the FB terminal voltage during normal control.

Hello,

I am reviewing your inquiry and will get back to you shortly.

Regards,

Hello,

I did not see the attached file.  Could you send it again please?

Regards,

Hello Mike.
This is Kurata.
Thank you for your help.

In my fourth post, titled <UCC256403 5V10A output_Transient response waveform measurement during sudden load change 3> 2024.06.20, I attached the waveform, but I am attaching it again.
Please check the "310kHz BW terminal waveform".
If you have any questions, please reply again.0652.Transient response waveform of 5V10A output by UCC256403 control _(3).pdf

Kurata san,

Please give us some time to respond.

Thanks,

Ning

Hello,

1. I reviewed your report and you are looking at the waveforms after the converter has shut down.  Your are focused on area 1 in the below plot and that is after the event.

2. You should trigger on when the output drops out at 2 and look at the last three switching waveforms before shutdown.  You are interested in the last switching pulses in 3, before the output drops out.

Regards,

E2E question to TI (7th post).pdf

Hello,

Your inquiry has been received and I will get back to you shortly.

Regards,