TinySwitch-III Line Undervoltage Sense False Triggering

Hello evanpower,

thank you for considering using Power Integrations product.

Before we dig in, I may need to learn the problem better. I have several questions about this probelm:

1. what's your working condition, Input AC voltage, load condition and how much time is the removal of the power before you re-apply? Is there any difference if the off time of AC changes?

2. if possible, can you also upload the schmatic and waveform of the IC off of 2.5s.

Best regards

Hi YY,

Thanks for your response.

My input voltage is 120 to 277VAC. I see this issue at both 120V and 277V input. There are two output rails, one is 24V and one is 5V. The two power rails have loads of about 50mA each.

I have attached a schematic. Waveforms would be difficult to capture because connecting a scope probe to the EN/UV pin prevents the problem from happening. I presume this is because the current that would otherwise flow into the EN/UV pin flows through the oscilloscope probe instead.

To reproduce the problem power has to be disconnected for about 100 to 500ms and then reconnected. If it is disconnected for more time than that the power supply starts up as expected.  

Hello evanpower,

the power was disconnected about 100ms-500ms and how much is time you see for the IC stopping swicthing? is it latch off all the time? you mention more than 2.5s. how long time is it exactly?

will it come back itself or you need to re-power it?

If you can provide the waveform you saw, it will be easy for us to understand and explain it. i think we don't need EN/UV pin signal. if you can provide Bus voltgae(voltage across C1 positive and GND-HV), BP/M voltage and output voltage. it will be good enough for us to see what happens.

By the way, why you use 10uF BP capacitor for this device? the default value should be 1uF for reduce current limit.

Best Regards

Hi YY,

The IC stops switching until the next power cycle. I have left it for over 10 minutes and it has not resumed switching. I mention 2.5s only because that is the delay before the TinySwitch will restart after a fault condition.

If I remove power, leave it removed for a few seconds, and then reapply power the power supply will work.

I've attached two waveform images. In each one Channel 1 (in yellow) is the voltage from C1 positive to GND-HV and channel 2 (in green) is the voltage on the BP/M pin. In one of these waveforms the power supply started up properly. In the other it did not start up at all.

Do you think that the value of my BP capacitor could be related to my problem?

Thanks,

Hello evanpower,

Thank you very much for the waveform. it is very helpful.

if you see the not-working waveform, the BP voltage has a reverse current from BP 10uF capacitor into BP pin. this causes the BP voltage has a sudden drop. if you can zoom in that portion and detect how long time it drops, you can calculate how much current go into the BP pin. if it is higher than 9mA, it will casue the latch off option which means you need to cycle power to restart the power supply.

As why you get the negtive spike at BP pin, this is really weird. Did you connect any probe when you initially find this problem? if possible, can you try with 1uF capacitor and do the same test? 

Best Regards

Hi YY,

I zoomed in on what looked like a negative spike on the non-working waveform. Zoomed further in it did not seem to be a negative spike at all. It looked more like some small ringing. I've attached an image of it.

I first observed this probelm without any probes atached to the circuit. I do not need any probes attached to the circuit to reproduce the problem.

I repalced the 10uF capacitor with a 1uf capacitor and the problem still ocurred.

Hello evanpower,

thank you for the waveform. Can you share the layout of your design. Is the BP capacitor close the BP and Source pin. I am wondering why we get the ringing at BP pin during startup. Can you try one test please, solder the BP cap 1uF directly to the BP pin and Source. i am thinking if there is any ground noise caused by the layout problem.

Best Regards 

Hi YY,

I have attached the critical part of the layout in the file "Layout.png". The traces to the BP capacitor are longer than is ideal, so I did put the BP cap closer. See the atached image "NewCapLocation.png" for the new location of the 1uF BP cap.

With the capacitor in the new location I still saw the problem of the device sometimes not starting up.

Thanks

Hello evanpower,

It seems that new cap location improve the performance.

I think the new location is still not the best. It seems that you connect to FB pin. You can take thre refence from our datasheet to see how to make the BP pin layout. For a test prupose, you can use a through whole capacitor and solder it direct to BP pin and Source pin to see if the problem will be solved or not. Make sure the lead is as short as possible.

Best Regards

Hi YY,

As you advised I replaced the 1uF surface mount cap that I was using with a 1uF through-hole cap. I cut the capacitor legs as short as possible and mounted it directly over the TinySwitch IC. The behavior did not change. There were still some power-ups where the IC would not run. I've attached oscillisope images of the BP/M pin during a working start-up and a non-working start-up.

Best regards,

Evan

Hello evanpower,

can you cut the BP pin related traces and just leave with the BP pin pad. Otherwise it will pick up noise. Is the capacitor you use is X7R type?

Best Regards