TNY276G Production ready design failing -- Need help
Hello,
For a new product we are launching, we have seen issues with the current limit detection via the capacitor on the BP/M pin.
The problem is that the chip appears to start off in the wrong current limit mode.
We set it with 10uF to Limit+1.
However, during power cycle testing, it appears that the supply may be operating on LIMIT-1 mode instead or possibly LIMIT.
The capacitors fitted measure between 8.8 and 9.8uF which meets the specification for tolerance of 10uF -50%
And so I am not sure what could be causing this problem since no information is given as to how the state machine for selecting the current mode operates.
The symptoms of the problem are that the supply will regulate correctly when I apply a step load to the 12V output. There is then a ~100mV drop (as expected) which is immediately compensated for and the voltage returns to 12V.
Then, when we power cycle it, it may or may not start in a different mode and when I then apply the same step load, the output voltage drops to 9V and I can see the supply trying to regulate by the voltage rising back up. However, the output never reaches 12V before the ~40ms is up and the supply will reset and restart.
I am able to get the same behaviour occurring constantly by changing the capacitor for a 4.7uF (outside the -50% tolerance).
Could you please advise what could be causing this supply to detect / start in the incorrect current limit mode?
Despite the correct value capacitor fitted?
Comments
Hi, can you try putting a 0.1uF ceramic capacitor across BP and SOURCE pin. This is to rule out noise coupling to BP pin.
Hi, we have just tried the 0.1uF capacitor from BP to S. It appears to make no difference.
The 10uF capacitor we have in the design is also ceramic X5R.
The waveform on pin 2 looks identical for power-up in correct mode and power up in incorrect mode.
We really do not understand the method of TNY I-limit mode detection and thus cannot solve this issue.
Can we please get some more advanced information on this?
Interestingly, adding another 10uF (Electrolytic) capacitor in parallel with the ceramic (20uF together), does seem to fix the problem but again, we do not understand the mechanism as the original capacitor is within the datasheet specification.
Here are some details...
| Attachment | Size |
|---|---|
| Good Mode +1 BP waveform (81.5 KB) | 81.5 KB |
| Good Mode +1 Waveform (79.06 KB) | 79.06 KB |
| BP_Bad.jpg (71.77 KB) | 71.77 KB |
Schematic...
| Attachment | Size |
|---|---|
| PSU1.jpg (437.64 KB) | 437.64 KB |
That's what I was going to ask next to use electrolytic capacitor for the BP supply. Electrolytic capacitors have stable capacitance tolerance over DC voltage. Where as ceramic capacitor depending on the quality, its capacitance can drop almost 90% when impressed by 5-6V DC. I suggest to ask your capacitor supplier for the DC BIAS CHARACTERISTICS of the ceramic capacitor you are using. If you remove the 10uF ceramic capacitor and leave the 10uF electrolytic, for sure it operate normally.
That is great information. We could have saved a lot of time if this was mentioned in the datasheet.
This is the datasheet recommendation for ceramic:
"BYPASS/MULTI-FUNCTION Pin Capacitor
The BYPASS/MULTI-FUNCTION pin can use a ceramic
capacitor as small as 0.1 μF for decoupling the internal power
supply of the device."
So it appears that the ceramic caps are withing the specification. I did several DC bias on these capacitors and found them ok for Cx > 10uF -50%. This is also substantiated by the waveform on BP pin looking the same for good and bad mode startup.
I am still not understanding when / how the chip determines the mode and why that is not a stable detection.
Hi, It looks like the culprit is the ceramic capacitor based on your testing, since when the electrolytic capacitor was used the problem went away. Please contact the FAE/Sales support assigned in your region for further information about the IC operation beyond the datasheet and applications notes have specified. (https://ac-dc.power.com/sales/)
Thank you.
I was hoping that this forum was the best way to obtain the correct knowledge. We do not have an FAE listed in our region.
Hi Power Integrations Team: I really need help with this. We have several 100 boards coming out in the next week.