LNK623 DESIGN - 12V 150mA - Kind request for help

The waveform you're seeing when you attach the 23.5 ohm resister is called auto-restart and it is used for protection against output fault conditions.


Something I'm confused about:
- Your design is for a 12V @ 150 mA design, which is a 1.80 W output.
- At 47 ohms, your output power is around 3 Watts
- At 23.5 ohms, your output power would be around 6 Watts.
- If you're performing tests at 3 and 6 watts of output power, why did you design the power supply for a 1.8 watt output?


A couple of other questions:
- Did you have your transformer built as shown in the PI Expert design files? Did you verify that your magnetics vendor built the transformer correctly?
- I can't view the gerber files directly. Can you generate a PDF of your layout so I can take a look at it?
- Are you using the output pre-load resister (R5) that you have shown on your schematic or did you remove it? If you removed the pre-load resister, this could explain why your output voltage is going up so high. Is the 40V you're seeing during no-load conditions?
- Have you checked your drain voltage/current waveforms at different line and load conditions?

-The Traveler

Dear Traveler,

thank you so much for your reply.

In contrary to your maths which are absolutely correct, it seems that it was not just the psu in shutdown/reset mode...my brain must have also been in shutdown mode as well, as I intended to put the resistor in series and not in parallel..

So far to recap; the PSU correctly enters thermal shutdown mode when driven at 47R load (3watts) and it works beautifully on a 94R load (1.5watts) producing a nice steady 12.3V output.

Please note that I still have the issue of 33-40V outputs however when the output is floating (no load). The resistor R5 that you are mentioning is connected to pin 4 of the transformer and back to FB1 pin of the LINK (also to R6).

I tried experimenting with different resistor values as an external (additional to the schematic) load so as to see what is the minimum value that results in a stable target voltage of 12V. Finally i concluded that putting a 7.5K resistor across + (schematic TP4) and return (schematic TP3) results in a stable voltage output of 11.9-12.1V.

So in overall i am happy with that I guess.

My next step is to investigate what sort of protection circuitry i need to add to the output of this psu. My target application outputs a PWM by fast switching a FET which in turn switches on and off the 12V coming from the PSU to an external sink...In your opinion, how do switching power supplies differ in robustness to a typical linear one? Are there any guidedlines as to how to best protest the switching PSU electronics?

Many thanks,

thanos-


No worries...I seem to have my own shutdown/reset days as well :-)


Regarding your output pre-load resister: There isn't much way around this. The LinkSwitch device is going to have a certain number of switching events in a given time period...even at light-load and no-load conditions. In a situation with no-load, there is nothing to dissipate the power getting delivered to the output caps, hence the voltage rise. The simplest solution is exactly what you've done, a small pre-load resister to bleed off the extra energy stored in the output caps. The value of pre-load you use in your final design will be a trade-off between no-load output voltage regulation, no-load input power, power supply efficiency, etc.


If you're going to be using this power supply with some kind of MOSFET load that is switching, you will want to consider/investigate the following:
- Depending on how high the FET switching frequency is, you might run into some conducted EMI problems. This can be addressed with local decoupling caps at the FET, some kind of pi filter arrangement, use of our E-Shield winding technology, modifying the gate drive for the FET to turn on/off a bit slower.


In comparison to a linear regulator, a SMPS will typically have worse noise performance. Luckily, the noise generated by the SMPS that your application sees will mainly be in the form of high-frequency ripple in the output voltage. This is typically easy to filter.


To best protect the LinkSwitch device, you will want to do some fairly standard PSU tests:
- Thermal characterization. Inexpensive ceramic caps can have problems at cold temperatures. Your bulk caps will also suffer from increased ESR at colder temperatures. At higher temperatures, the lifetime of your bulk caps will decrease. You will also want to verify the LinkSwitch device temperature at your maximum operating temperature.
- Verify that your clamp circuitry is working correctly. Drain voltage/current waveforms under different combinations of line voltage, load level, etc should be verified to make sure there isn't excessive ringing in the clamp, excessive dissipation in the clamp, to verify that the drain voltage rating of the MOSFET isn't being exceeded, etc.
- Load/Line regulation measurements are typically a good idea.
- Output ripple under different line/load combinations are also a good idea.
- No-load input power (if applicable), efficiency, etc.
- Transient response.

-The Traveler

I am using lnk623 for smps 5w i.e.5V 1A. But there is problem like at output capacitor is charged up to 6.44 V then it discharged when no load but whenever i connected LED across it it gives me 1.67v  so what should i do for load regulation plz tell me its urgent

I have attached my design file

Hi jagrutiraut3,

Thank you for choosing Power Integrations.

May I know the end application of this power supply? Will this be for some kind of LED bulb/lighting? If so, you may want to check out Power Integrations' line of  devices designed for LED drivers. I am attaching the link below:

https://led-driver.power.com/

Your LNK623 design is actually a constant voltage power supply and normally LED loads will need constant current power supplies. What happened here is that the output was clamped by the LED voltage which is why you saw 1.67V. If the LED you connected will only serve as an indicator light or something then you may want to put a resistor in series (typically 330ohms / 0.5W).

this power supply is for a board having pic microcontroller,xbee and their interfacing

for testing i used LED. I attached ma design i imlemented it with some changes like i hv removed inductor at input side and only one capacitor is used and also i m confused in transformer pins

Please reply me fast

Hi jagrutiraut3,

First of you may need to use an electronic load to test the regulation. Using an LED to simulate your load will not be correct because the output will always be clamped to the forward voltage of the LED. If an electronic load is not available you can make use of a resistor load. A 5ohm resistor will give you 5W which is equivalent to your design. I would suggest using a 15W resistor to prevent excessive heat on the resistor.

On the output voltage increasing to 6.44V when no load is connected, this is because there should be a minimal impedance connected to the output to prevent the output capacitor from charging to above your desired output voltage range. Try connecting a 300 ohm resistor at the output to act as preload. You can reduce or increase this value depending on whether you want to minimize no load input power or maintain a certain output voltage at no load.

Could you illustrate what difficulties are facing with your transformer pins?

ok

Transformer problem resolved now

Thank you

Helllo m using lnk625 till circuit work well but as i change the output capacitor to 47uF25 V (prev was 47uF 10 v)

drain of lnk get burnt please tell me where m wrong  i hv attached scemtic of mine

Hi jagrutiraut3,

Is this for the same project? Or is this a new design? May I have the input and output specifications for this power supply?

I have checked your schematic and have a few comments:

1. Using a 100W fusible resistor would cause very high power dissipation. Can you use a 10ohms fusible resistor instead?

2. Your 47k snubber resistor will also cause high dissipation. Do you really need 4k for this?

3. Could you send me your transformer specification and construction?

4. Please also send the PIexpert design file you used so I can help you check.