(LYT5228D Buck) Problems with SVM (Stroboscopic Effect) and inaccurate Magnetics Designer using LYTSwitch-Buck Design1.pixls

Hi,

You need to grind the core first until the recommended 235uH+/- % margin is achieved as recommended in the PIExpert. A 400-450uH is too high that can violate the minimum time on of the Lytswitch which results into single pulse/switch that causes the PSU not to power up. 

If grinding the core did not resolve the issue, you can provide here your PIExpert design and design files (schem, layout) for us to review.

Regards,
PI-Nanoe

Hello,

thanks for the response. However, I have no means to grind cores myself. Instead, I got sample pre-ground cores from my supplier, according to the airgap suggested in the design files.

Okay, I managed to get the Transformer situation sorted (though the suggested air gap was way off; in the PIExpert file, it said 1mm, however the airgap ended up being 1.8mm to get the desired Inductance). The device turns on and runs reliably.

However, the SVM value is still bad. We have an allowable limit of < 0.4 SVM. However, the value sits at 2.72 SVM. If I use a larger electrolytic capacitor to buffer the LEDs, the SVM value goes down, however I would need a truly large one to get it low enough, which isn't possible, and it should not be necessary if everything was working correctly.

Hi,

Can you share the schematic design or the PI XLS file so we can review accordingly? 

-Gavin

Hello,

see attached files. Part 15/R10 is a potentiometer used to adjust the LED current and total power to the values set in the PI XLS file, so it is not actually 10 Ohms, but usually a value close to what the PI XLS file suggests.

Hello,

Aside from increasing the output capacitance (C8) further, adding an active ripple current filter circuit at the output as shown in the attached image might help reduce the SVM value further. The ARCF circuit comes from the DER-543, which is an 18.4 W (385 V, 48 mA) Boost LED Driver Using LYTSwitch-5 device, so the voltage and current rating of Q1 should be changed accordingly. The voltage rating of capacitor C11 should be reduced as well to 50 V since your output is only 36 V, and its capacitance value can be increased depending on the SVM result. Resistor values are also subject to optimization if needed.

I hope this helps. Let us know the result once it is available on your side. Thank you and have a great day ahead.

Regards,

Tommy

Hello Tommy,

thank you for the response and the potential solution. However, I am not familiar with ARCF circuits, and I cannot quite work out how to calculate the appropriate resistor values. Do you have some resource that can be used to calculate those, or else something you can point me to where I can read about this specific type of ARCF circuit that could help me get the answer?

Choosing an appropriate capacitor is not an issue, and for the transistor I assume any that has appropriate voltage and current ratings for my specs should work (I have a BD135-16 handy that should fulfill that purpose, at least for the initial tests), however the resistor values currently have me scratching my head a little, I would really appreciate if you could help me out here. Thanks in advance.

Kind regards

Alex

Hello Alex,

The ARF is almost similar to a linear series pass voltage regulator which comprises of NPN BJT (Q1) and an RC low pass filter (R4, R6 and C11). The low pass filter lowers the Q1 base drive current inducing a voltage drop across Q1 collector emitter. The voltage drop on Q1 CE reduces output ripple voltage, thereby reducing the output ripple current.

Unfortunately, we don't have any documents that shows how to calculate the resistor values. However, you can use the current values as a starting point. Just ensure that the BJT transistor operates in linear region.

Regards,

Tommy

Hello Tommy,

thanks for the response. However, as this circuit is used in DER-543, someone (or something, if an Excel-file was used) must have calculated the values for it, I'm sure. Having that calculation as a sample would help me immensely.

Kind regards

Alex

Okay, I have tried to get this to work, and unfortunately I cannot get the circuit to run. If I turn it on, the lights will briefly flash, then the circuit shuts off and draws no power. I also cannot find any ACRF-circuit that is built like this one online so I can read up more on it. Most either use several transistors or none at all. 

I have double-checked everything, followed the design spreadsheet to the letter, but I cannot get this to work with an acceptable SVM, and trying to use the ACRF-circuit just means that the device will refuse to work at all. I really need someone to tell me exactly which parts to use so the ACRF-circuit works with the values provided for my circuit.

Hi Alex,

Good day and I apologize for the delayed response. We really appreciate the information that when the ARCF circuit is connected, the design is not functioning correctly.

The R1 and C1 in the attached ARCF image are responsible for filtering the ripple voltage. The corner frequency [f = 1/(2πRC)] of R1 and C1 should be set below the line frequency (50 Hz / 60 Hz) to reduce the 2x line frequency ripple present at the output. Resistor R2 limits the base-current during the output short circuit condition.

If we calculate the values from DER-543, the corner frequency is around 12.06 Hz, which is below the line frequency. Since the current values from DER-543 are not working with your design, let's try tweaking the values further while keeping the corner frequency below the line frequency.

We're able to find one internal evaluation report where they've used the following values for their ARCF circuit (please see image "Evaluation using ARCF") and the board RD-257, but they are using the LinkSwitch-PH instead of LYT.Switch-5. Please see the other attached image as well for the test data results of the evaluation. Even if the devices are different, the concept of ARCF is still the same. The RC values shown in the image give a corner frequency of around 10.26 Hz. Can you try if the values of R1 (330 ohms), R2 (300 ohms), and C1 (47 uF) will also work for you? Please note as well the ratings of transistor Q1.

If this still doesn't work, another option we can suggest is to use LYTSwithc-6 device in your design.

I hope that the information above helps. Thank you and looking forward to the results.

Regards,

Tommy

Hello Tommy,

thank you for the response. I will have to order some parts so I can try this out, I will post the results as soon as I can build and test the circuit.

For my further understanding of the ARCF circuit used, you say that the corner frequency should be below the line frequency. Am I assuming correctly that the circuit is more effective the further it is from the line frequency?

I am asking because we are planning to use a similar circuit based around the LYT5228D for a few different configurations with the output values modified as follows:

Case 2 (driving 18 LEDs): Voltage 54V, Current 320mA

Case 3 (driving 24 LEDs): Voltage 72V, Current 320mA

Case 4 (driving 30 LEDs): Voltage 90V, Current 275mA

Case 5 (driving 36 LEDs): Voltage 108V, Current 228mA

Obviously I need to adjust the ARCF circuit for these (assuming the values you provided work for the first use-case), and the capacitor in particular will need to be rated for a higher voltage, which might also cause space concerns on the circuit board. It would be easier to get an appropriately rated and sized capacitor if the capacity can be lowered, but that causes the corner frequency to rise. I understand that the corner frequency needs to be kept below the line frequency, but if my assumption is correct, a higher corner frequency would mean worse filtering results. Is this correct?

Hi Alex,

Theoretically, yes, because the RC circuit (R1 and C1) is basically a low-pass filter. However, since we don't have any additional test data, I can't confirm this yet. You can try experimenting with different combinations on your side to verify its actual impact on the design. Let us know once the results are available on your side. Thank you and have a great day ahead.

Regards,

Tommy

Hello Tommy,

I received the parts and assembled the circuit. As a result of the suggested changes, the device works, the LEDs power on. However, while the ARCF did manage to reduce the SVM down from 2.72 to 1.62, that is still far from the required <0.4. 

What steps would you suggest next? Unfortunately, switching to a different driver is, for now at least, not an option, as unfortunately our company has already ordered 2500 LYT5228D that are now in storage, and will have to be used in some fashion. 

Kind regards

Alex

Hello Alex,

Good day and thanks for the update.

We're sorry, but we don't have any other solution that we can think of aside from optimizing the ARCF circuit and increasing the output capacitor to reduce the output current ripple to further reduce the SVM.

Thank you for your understanding.

Regards,

Tommy

Hello Tommy,

I might be on the right path to optimizing the capacitor and ARCF circuit. I will let you know once I have finalized it.

Kind regards,

Alexander Lehmann

Hi Alex,

Okay, let us know the results once it is available on your side. Happy Thanksgiving!

Regards,

Tommy

Hello Tommy,

will do. Happy Thanksgiving to you as well!

Kind regards,

Alexander Lehmann