Output Regulation Issue with LED Driver AL1665

[Johnson R]

Hello Everyone,

I recently designed a non-PFC based flyback SMPS with AL1665 for a power of 150W. The schematic from the application note : https://www.diodes.com/assets/Evaluation-Boards/AL1665-AL5822EV1-Flyback-User-Guide.pdf
is followed.

I have designed it for the universal AC voltage range from 90V rms to 300V rms. Output voltage is desired at 24V.
From the design sheet of the AL1665, the minimum operating frequency was to be 66KHz.

The issues that I am facing are
1) The switching frequency is observed to be at around 16KHz instead of 66KHz
2) Output Voltage is not regulated at 24V at no load. Instead the output voltage just keeps increasing till I turn OFF the input(see attachment - Pink is the rectified HV input and blue is the unregulated output)
3) When the output is loaded with just 50mA, the observation that is made in #2 is more faster

I have checked the FB pin, CS pin, NTC pin and the ADIM pin and all these have spike noises during switching which even inspite of the bypass capacitors are not suppressed(see attachment - Pink is the rectified HV input and blue is the VCC and yellow is the FB voltage)

I had made a design for the same power with other controller UC2843 and there I had did not face this kind of issue of unregulated output.

Kindly let me know what am I missing with this design.

[mtwieg]

AL1665 datasheet it has a minimum on-time of 1us typical. Therefore if there is no load, Vout will rise slowly without any ability to regulate itself.

The datasheet also states there is an output OVP function, with the OVP threshold depending on how the FB pin is connected. Can you show your full schematic?

[Johnson R]

Yes, but I see that the OVP is not triggering -> Rather the Output is rising beyond the rated value of the output capacitors. As I mentioned, I tried with some burden resistance of ~ 50mA but that actually accelerated the voltage rise.

Turns ratio between Pri to Sec is 6.5 and Pri to Aux is 8.5

I have attached the schematic for your reference.

[Johnson R]

I now tried checking every signal of the ASIC and here are my observations.

1) Under no load, the switching frequency ~ 26KHz and not 66KHz
        --> My output voltage had to be 24V but it is settling at 34V(average) (OVP limit), the VCC is at ~20V



2) With a load of even 250E at the output, the output voltage is dropping to 13V average
               --> The switching frequency is dropping to 4KHz and the VCC is ~ 24V

 I am suspecting that the inductance value is lower here and somehow the IC is current foldback mode. i have used the calculation sheet from the Diodes Inc website to calculate the inductance it came aroung 100uH.

[xavier60]

Check the phasing of the windings by comparing the waveforms on them.

[Johnson R]

Thank you Xavier.

But do you mean the polarity of the windings? I'll do this check once.

[xavier60]

Thank you Xavier.

But do you mean the polarity of the windings? I'll do this check once.

Yes.
If I understand the data sheet correctly, the FB divider should apply 3V to the FB pin when to output is 24V.
Im struggling to properly understand how primary CC regulation works. Only saw the first example of it a few
days ago. It uses a PFC controller, CM6503A somehow.

[Johnson R]

I made the polarity tests for the three windings that I have.

1) Excited the primary (Pin 1 positive and pin 4 negative) with 10V p-p sine wave (Yellow wave).
2) At the Winding #2 (Pin 8 positive and Pin 9 negative), ~ 2V p-p sine was observed in phase (Pink wave)
3) At the Winding #3 (Pin 11 positive and Pin 12 negative), ~ 2V p-p sine was observed in phase (Green wave)

The winding direction while testing the polarity is marked in the screenshot(winding numbers are marked in RED & winding polarity is marked in corresponding colours matching the waveform)

[xavier60]

That all looks correct. Im not really sure where to go next.
The very short MOSFET on time followed by lots of ringing makes it difficult to make sense of the waveforms shown so far.
It can also cause rectified voltages to not be as expected.
Maybe just a drawing mistake, the way D5 is shown connected makes the spike snubber ineffective.
Have you tried a larger load to see if there is current regulation?
I would also reduce the input voltage to reduce the risk of destroying the MOSFET.

[Johnson R]

Yes, the polarity of the transformers seems correct.

There is an official simulation file in Pspice from Diodes Inc and I tried the simulation with my circuit design. And in simulation it seemed to work as per design. So my initial thought of the inductance being insufficient was not an issue with the simulation.


So I thought the problem might be because of cross regulation and made the following changes
1) I changed R8(pull down resistor for feedback voltage divider) from 13.7KE to 10KE to increase the output voltage during loaded condition but this only increased the voltage by 1V
2) I then modified my pcb such that there was no isolation between the secondary winding and the primary winding ( Shorted GNDD to HV_RECT_MINUS)
3) Regulated to be on secondary winding instead of Aux winding (disconnected VAA connection to R7 & connected VDD to R7)

But even these changes, I was not able be observe the desired output voltage(only 14V was observed)

Feedback Pin:
I probed the feedback signal along with the actual transformer secondary voltage and I observed that during primary conduction period the feedback signal and these spikes were almost reaching 10V & -6V.
This might also been because of the probe. The steady state values were reflecting the transformer voltage.

Current Sense Pin:
I observed similar behaviour on the current sense pin as the feedback pin. The spikes during the primary side conduction were as high as 10V. I had probed the MOSFET source pin and the current sense pin of AL1665.

Comp Pin:
Here I observed the strange behaviour.
I was using PVP3150 probe from RIGOL(termination is 10Mohm and 10pF +- 5pF) and when I probed the Comp pin with this probe, the output was regulated to the desired 27V with no overshoots  . The Comp pin still had spike voltage reaching upto +10V & -5V
When I removed the probe, the output again settled at 14V.
I suspect the issue is either something to do with the layout or the compensation network.
I also changed the C of the RC of Comp pin to 2uF to check if the regulation was fine but the output was still at 14V.


The PCB is a single sided PCB with copper traces only on TOP layer. 

My next step is to add 15pF or 22pF capacitor across the COMP pin and check if the output is able to regulate.

[xavier60]

When looking at spikes, some of what the DSO might be showing is the HF voltage drop across the probe's ground lead. Test by probing the point where the ground clip is connected.
To minimize capacitive pickup of spikes at the FB pin, the resistors should be placed near the IC.