LM5156H: Isolated flyback problem

Hi Matan,

Thanks for using the E2E forum.

Can you please conduct a test to measure the following waveforms

1. Input voltage 

2. Output Voltage

3. Comp pin Signal

4. Switch Node Voltage 

Kindly, also attach the QuickStart calculator values you used for the design.

Best Regards,

Hassan 

Hi Hassan,

Here is the calculator Excel spreadsheet:

LM5155_56_Excel_Quickstart_Calculator_050325.xlsx

I will perform measurements ASAP and update.

Thanks,

Matan

Hi Matan,

Thanks for sharing the excel calculator.

Best Regards,

Hassan 

HI Hassan,

Scopeshot:

CH1 - Yellow - VOUT (after diode)

CH2 - Green - VIN

CH3 - Magenta - COMP pin

CH4 - Blue - switching node

Thanks,

Matan

Hi Matan,

Thanks for the scope plots. I have gone through it. 

The results show that the comp pin is at its max clamp voltage >2.5V and it demands high duty cycle, but the switch node voltage shows that the turn on time for the FET is very small. Its mean that at each turn on time the converter reach the peak current limit and turn off the FET. Due to the current limit the duty cycle is very low. 

In order to investigate the current limit, I request you to send me the scope plot of CS signal.

In order to measure the CS signal, I recommend you to set the scope at 20mV/div and measure the CS signal with small ground loop. If possible, try to measure it with firefly probes https://www.pmk.de/en/firefly. 

And kindly share the saturation current values of the transformer used in your design. 

Best Regards,

Hassan 

Hi Hassan,

I asked for the saturation current values and will share when I'll have them.

Scopeplot:

The Yellow channel is the CS pin.

Unfortunately, I don't have the firefly probe. I've measured with GND spring.

Thanks,

Matan

Hi Hassan,

So it seems there are problems after all. Even though I was able to measure 56V at the output, upon testing it some more - there is a problem, also without load. The problem occurs more often when loading but also without load.

Let me share some scope plots:

CH1 (Yellow) = Switching node

CH2 (Green) = Output

CH3 (Blue) = COMP pin

CH4 (Magenta) = Gate of the FET

(a) In the following plot, it can be seen that the frequency as as expected, around 250KHz. Gate is ~8V. There are some overshoot and undershoots on the COMP pin, but at the overall seems reasonable, I think.

(b) Zoom-in on gate turn-on.

(c) In this plot, missing turn-on cycles are missing. To be more accurate, as will be demonstrated at the following plots, the gate voltage is turned-on but immediately goes to zero volts for some reason. The COMP spikes at those occurrences seem to be bigger.

(c) Zoom in of the gate turn-off very quickly event.

(d) And more zoom in. The gate is turned on but almost immediately turned-off.

The schematics and quick start calculator are already provided in this thread.

Please advice.

Thanks,

Matan

Hi Matan,

Can you please attach the CS signal with these scope plots?

I think the device is turning off because of the peak current limiter.

The scope plot you attached in previous message for CS signal shows that the CS pin is reaching the peak current limit threshold.

Kindly, take a scope plot together with CS, SW node, Comp, and gate signal. 

Best Regards,

Hassan 

Hi Hassan,

Please see following scope plots:

And regular cycle:

Thanks,

Matan

Hi Matan,

Can you please specify the load current for each scope plot case?

Best Regards,

Hassan

Hi Hassan,

All scope plots are without load.

Thanks,

Matan

Hi Matan,

Thanks for the feedback.
Hassan is out of office, so I will take over in the meantime.

To me it looks like some of the sensing signals are noisy, which might cause an unintended trigger of the overcurrent protection.
For an additional check, I would recommend to increase the filter at the CS pin. This can reduce noise at the pin and might solve the problems. (E.g. increase R2 to 1kOhm)

If there is no change, I would have a look at the layout if possible. (If I understand correctly, Moritz only reviewed the schematic)

Thanks and best regards,
Niklas

Hi Niklas,

I changed R2 to 1KOhm and C63 to 1nF. It seems it solved the problem.

I'm still testing.

Thanks,

Matan

Hi Matan,

Thanks for the update and the good news.
Please let me know if testing shows any other issues coming up.

Best regards,
Niklas