Mods to DER-107 ref design causing thermal IC problems at 277VAC input range.
I inherited a PS design based on DER-107 to debug and am having some thermal problems on the controller chip at the high voltage input range.
I am looking for some advice on what to look at or what to change.
See Details and Questions below:
DER-107 is a 10W design 90-300VAC using TOP245R.
http://www.powerint.com/sites/default/files/PDFFiles/der107.pdf
Details
We are pushing the upper voltage limit to run off of 277 mains +10% which is 305VAC
Our output is 5W ~ 1/2 the example and we are pushing XFMR to double insulation CATIII (UL61010-1) but otherwise following ref XFMR design.
The original engineer thougth he could use a cheaper DIP8 TOP244PN part since our power dissipation is less.
The ref design TOP245RN is a TO-263-7C Dpack with a lot of copper on the layout.
Our DIP8 has about 1" square of copper total top bottom of pcb tied into source pins and with vias linking pours together.
With ~ 4W load @ 277VAC the controller chip thermal rise is 55C!
At 120VAC in and the same load the IC has a rise of only 11C
Note that the RC snubber Diode is also running hot with a 22C rise
Questions
1) should I expect to see such a thermal rise increase based on input voltage?
2) Is this just a thermal dissipation issue between the two packages or the Rds of the switching FET?
3) Does the snubber blocking diode thermal rise indicate some problems with this snubber circuit as the main culpret? I will check for excessive ringing
4) Our increase of the XFMR isolation may have increased LL and Cs I will capture some waveforms of the switching
5) are there any higher performing DOP8 TOP parts that would drop in and help out here? Perhaps with lower Rds on.
6) Other issues to look into?
Thanks for reading this, look forward to your feedback
Comments
PI-Cochrane - thanks for responding!
Actually the primary problem turned out to be that the transformer build was very different than ref design
With another vendor's sample in the same target I get 19C rise instead of the 55C.
This is a big improvement but the load was only 3W in either case and our target is closer to 5W.
I have a board rev to address this.
After digging into this I do have some additional questions I am hoping you can address:
I do not see TOP245/244 in available products list on PI web site, I had to dig for the data sheets.
The data sheet dated 2005 shows the part is mature.
Q - Should I consider migrating to a more up to date part and if yes what are your suggestions?
I see that the TOP244PN DIP-8B package has reduced functionality over the TOP245R D pack with fewer leads.
One issues relates to emissions in that you can't slow it down to 1/2 speed so the conducted emissions may be a problem at the low end. Also there is not a dedicated current overload X pin, instead the M pin discussion I believe implies that one should use the thermal shut down control for current limit. This means the FET and package will reach ~ 140C and so there could be some thermal concerns from UL.
Q - Are there other compromises that hapen with this migration?
Attached is a spread sheet with detail on the problem transformer
Although the construction is very different I believe it could work.
I suspect that the high OCL of 2600uH is the culpret
Q - do you see any obvious base issue in the design that contributed to the heat?
Chris Wells
Eaton Electrical
It is true that options of different lackages are not same, and the DIP-8 package has no pin to control the frequency. If you really need the option to reduce your frequency, you have to use other packages.
For the thermal shut down, all the parts will be shut down once the junction temperature reaches 140 degree c. Because the thermal shut down only happens at abnotmal condtions, and140 degree c is the junction temperature, I do not think UL mark is a concern. Actually, PI parts used a lot by different customers for different applications, all these parts shut down at 140 degree c, and I believe they all can pass UL stadard. Or customer will complain about that.
I really TOP244/5 is a very good choice for you, and they should be available right now, so you do not have to use latest version of TOP switch.
I found that DER107 uses a 600uH inductance. For a 5w design, I really believe the 2600uH is too much for your application, because your output power is only half of DER10( even you use TOP244 instead of TOP245). The high inductance cause alot higher copper loss, higher core loss and huge difference of the operation mode of TOPSWITCH. Before you decide which transformer to be used, PLS run the calculation with PI expert, and figure out the right inductance and turn ratios.
With good transformer and good package, I believe the thermal issue can be solved.
Yes the parts are available "right now" however my designs live a long time in production.
My concern is about having to update it in a few years.
What is the expected sourcing support projection on this TOP244PN part?
Also is there a cost advantage towards using something newer?
I did some emission screens with vendor 2 transformer and the higher frequency is not an issue.
What does the belly band lend to the transformer performance?
Ok so the high OCL is driving the heat.
The engineer before me was apparently just following the reference design without digging in very deep.
I will try out PI expert like you suggested.
Thanks for your feedback again.
BTW I very much appreciate your application notes and reference designs.
Chris Wells
Eaton Electrical
Right now, we do not have any plan to obsolete this family TOP-JX and TOP HX are only several yesrs old, and thus obsoleting TOP24x families will cause a lot of trouble for our customers and lose a lot of opportunities for PI.
Could you caontact with the local FAE team for the sourcing support? I believe they will be very glad to help you.
The belly band is used to control the EMI.
Regards
Do you mean EMI ElectroMagnetic Immunity or ElectroMagnetic Emissions?
I was hoping to know what specifically it is doing.
I did some web searches and see that it has to do with reducing radiated emissions off the transformer
http://www.linkedin.com/groups/why-shield-are-used-in-3881004.S.89185028
Chris Wells
Eaton Electrical
The belly band is used to control the conductive EMI...
I am 99% sure that your problem is caused by the package. As can be seen in PI datasheet, R package and P package has huge difference for thermal and power capability, although they use same MOSFET. I suggest you using the TOP244R package and more copper heatsinkk on the PCB. The thermal should be improved a lot with these two solutions.
I do not think lower Rds,on part is necessary, since the issue is mainly caused by higher switching loss at high line input. But if TOP244R can not handle the thermal, then we have to try the TOP245R...
Regards