Solution Finder Get Tech Support

TOP266 NEW vs OLD

Posted by: mjscottinnc on

We have a 55 Watt TOP266 design which has worked flawlessly since 2014. However the last production run is seeing a 30% failure rate from the field. When we started to examine the failed units we noticed that the units that failed all have parts with the new Logo on them and a date code of 1650. Our design has not changed, our layout has not changed, our vendor has not changed. The failures seem to be cumulative, we test and burn-in each system prior to shipment. The failures occur days to months later.

Can you explain?

On a chance I downloaded the newest version of PI Expert, loaded the original design, restrict the transformer core the original one EER28 and then optimize it (just like we did originally). What comes out is an identical design with different component values. All of the transformer primary protection and snubbing components have change. Of particular interest is VR1 which is the protection diode for the TOP266 D terminal. In the design of 2014 PI Expert suggested 180 Volt TVS, but now that value has been changed to 130 Volt. Also the number of turns for each winding has changed by one or two turns.

I really need an answer.

Comments

Submitted by PI-Kenobi on 07/26/2017

You mentioned that the design has not changed: only the TOP266, correct? What kind of failure are you referring to? Does the TOP266 fail? Can you please send us your original PI Expert file so that we can evaluate it?

The original design was for a TOP266 and the parts used in production from 2014 - 2016 were the old PI logo. This design first went into production 1Q 2014. The latest production run 1Q 2017 used TOP266 with the new PI logo and these are the parts that have failed. What we see is the TOP266 shutting down and restarting. The shutdowns get more and more frequent which tells us that something is degrading over time. Original design file attached.

Attachment Size
Original Design File 325 KB

Please check the temperature of the device. It seems like the TOP266 might be performing a hysteretic restart which is typically seen when the device temperature has been exceeded.

Submitted by mjscottinnc on 07/27/2017

I do not believe that temperature has ever been an issue with this design. It has an overdesigned heatsink and more importantly this is a design from 2013 that went into production in 2014 and has only now failed. We will check temperature during one of our burn-in tests to confirm. So far we have shipped over 1000 units into the field. Had temperature been an issue, that would have been seen long ago. It was only when the TOP266 with the new logo was used that the units failed. Let's not overlook the obvious here! When taken with the fact that the new PI Expert now suggest different components that the NEW TOP266 becomes an even more likely suspect. When comparing the original design vs the new PI Expert suggestions... VR1 has changed from 180 V to 130 V, R3 has changed from 51 ohms to 5.1 ohms. This tells me that the parameters used by the NEW PI Expert have change so that the voltage at the D terminal is now clamped sooner and harder. Also taking into consideration that the specifications for the TOP266 max voltage on terminal D is now 735 V, it should actually work better but does not. My conclusion is that either the design, die, process, or prepackage testing has changed. This is not the same part we used from the prior 2 years.

Submitted by mjscottinnc on 07/27/2017

Sorry I misread my notes on R3, it was 5.1 ohm on the original design and on the on the new suggestions.

Submitted by mjscottinnc on 08/01/2017

I'm curious if anyone in Tech Support is working on my observations?

Can anyone explain the differences between different design recommendation from PI Expert when the designs are identical?

Can anyone explain the problems we are seeing with the new Logo'd parts vs old Logo? New fab mask? New Fab Facility? There has to be a reason.

The last communication was 6 days ago.

Submitted by PI-Kenobi on 08/02/2017

Sorry for the delayed response. For product reliability and quality inquiries, we have a QA hotline.  Please direct your request to CIM@power.com. Is it possible for you to send us your original PI Expert design file s that we can assess why the design changed. In the meanwhile, I would recommend that you not modify your original design based on the new recommendations by PI Expert.

The original design file was sent on this forum on Wed, 07/26/2017 - 11:32

Submitted by PI-Kenobi on 08/09/2017

Unless you re optimize your origingal design in PI Expert with the exact same parameters as your original optimization, you might end up with different results. In your case, it seems like you re-optimized an existing design (created in 2014) by restricting the transformer core: is this what you had done originally as well?