Solution Finder Get Tech Support

TOP226Y sporadic shutdown

Posted by: danw on

Hi, I've designed, --or rather copied a design given by Premier--, for a 24V supply for a client. This one:
http://www.premiermag.com/pdf/pol-24020.pdf
The supply is under-loaded rather than overloaded; --only using a fraction of the watts it is capable of, which might be part of the problem. The problem is that some of the units we produced occasionally shut down with no apparent cause, typically once every 2 or 3 days, but sometimes run for 10 days; sometimes shutdown after 1/2 hour.
We were able to force the TOP226Y to shutdown by using a heat gun on it and making it very hot, as it should be; but in this application it never gets hot enough to even feel warm to the touch. And there are no sudden spikes in loading or anything like that.
Even turning power off and back on often fails to revive the switcher, which at first lead us to believe that the chip was dead. But removing the power for several minutes and then turning it back on often restores normal functioning.
Looking at the specification for the TOP22x family, I find no mention of any feature or mechanism that could possibly lead to such a persistent shutdown, except for thermal shutdown. Anything else, as far as the spec describes, is subject to recovery via the auto-restart.
So, it would seem to me that the problem is a false triggering of the thermal shutdown circuits, possibly caused by EMI noise?
I do have the (TO-220) schottky standing parallel to the TOP226Y, which is probably not a good idea; but we tried bringing the two components closer together, as well as separating them, to see if it would affect the frequency of shutdown events, and it doesn't seem to have an effect.
So I'm wondering if this problem is known, and/or there is a workaround or recommendation.
T.I.A.

Comments

Submitted by danw on 03/02/2015

Update: One thing I did different when copying the design was I connected the Y safety capacitors between the primary positive voltage and secondary ground, instead of between the primary negative and secondary ground. Changing that over the weekend seems to have made a positive difference so far.
Could that cause the shutdowns?

Submitted by PI-KSAN on 03/03/2015

Thanks for Posting your question into forum. we understand that you copied a design from premier but when we looked at that document,it's just there a schematic. The problem that you are seeing might be because of Layout. Please go through our Top Switch Application Note 23 and Data sheet, there are some strongly recommended design guidelines and have to follow some of those design considerations for the lay out. Please go through it and see whether your lay out is according to the design guidelines. If so get back to us.

Thanks,
PI-KSAN

Submitted by danw on 03/05/2015

Your website is not working well; have had problems for about a week now, with "Server not found" messages.
Is this link to AN-23 (from your search results page) correct?
http://ac-dc.power.com/design-support/product-documents/application-not…
It just keeps giving me the "Server not found" message.
I'm using Firefox version 36.0.1 on Windows XP.

Getting back to the subject, I've gone through AN-15 and found no rules that I'm violating.
Besides, I should mention I've been designing power supplies using Power Integrations
products for so many years I've lost count. Probably about 50 designs or so, and I've
never had this problem before, of sporadic shutdowns.
The only thing I did different this time was connecting the common mode dampening Y
capacitors between output ground and primary POSITIVE, as opposed to the more
traditional connection to the NEGATIVE, for PCB layout convenience, which in my
mind should make no difference whatsoever, as both the positive and the negative are
"quiet" nodes (relative to earth ground), or at least neither better than the other, since
both are floating most of the time anyhow, and when not floating, each side may
connect to either line or neutral through the bridge, alternatingly.
And yet, changing the connection of the Y capacitor from the positive to the negative
so far appears to have solved the problem; but I can't think of a theoretical reason
for this to make any difference.

Submitted by danw on 03/05/2015

Got through and downloaded AN-23.
Lots of calculations, like for transformer windings and whatnot.
I have your design applications for that (not that we have ever
used them in an actual design, since the cost of ordering a
custom transformer is far outside most application budgets.
Pulling designs from Premier Magnetics is the only realistic
way to go, unless you are a military contractor, I suppose.)

((Coming to think of it, if your company is not happy about
people using 3rd party designs, perhaps what you should
do is come up with a rich set of design applications for
specific voltages and currents that use off-the-shelf parts,
including transformer. If you don't do that, then you have
to put up with the fact customers like us use Premier's
designs. It is just not realistic to expect everybody to
design and order a custom transformer for every app.
That is so prohibitively expensive we'd just have to find
some other solution altogether. Designing a transformer
and then looking to see if it can be bought off the shelf
is also not realistic, as there is no search tool allowing
one to find transformers by internal parameters, and
nobody has the time to browse through hundreds of
transformers for a suitable one.))

So, it seems there's nothing in it related to my problem in
AN-23, anyhow.
No warnings of shutdowns related to any components.
And in fact, AN-23 does not show or even mention Y caps
used for common mode damping between primary and
secondary sides. AN-15 does so, however, and it clearly
states that the primary positive rail is a "quiet" node. And
one of the schematics in it shows a 3-capacitor solution
with two of them in series across the primary DC, and from
the middle, another Y cap going to the output ground. No
preference stated; merely calls that an "option".

Submitted by PI-KSAN on 03/19/2015

Hi Thanks for getting back to us. You can still design Power supplies with our part with Ready made Magnetics not Custom Made. We have a design software Called PI Expert and PI XLS. PI Expert gives the optimised design solution for the specification that you enter. PI XLS is a transformer design software tool ,it designs transformer based on your specification. In Transformer design PI XLS Software you can choose whatever the core and bobbin material you want to use. It's not mandatory to use only custom made magnetics, our parts can use to design power supplies for any magnetics. Please use our PI Expert and PI XLS software.
Regarding the Lay out Issues and Rules , Please follow the below link which is for TopSwitch Design Guide:
http://ac-dc.power.com/design-support/product-documents/application-not…
Please download PI Expert and PI XLS.