Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Paralleling SMPS modules and common mode choke situation

Status
Not open for further replies.

Flyback

Well-Known Member
Hello,
Page 4 of this AC line filter datasheet rightly states that these line filters should not be connected up with their outputs in parallel with other AC line filters of the same make. (due to imbalance of current flow in the internal common mode chokes and subsequent saturation of these common mode chokes).
However, please confirm that if each line filter is connected to the input of a mains isolated SMPS, ( and not paralleled to another filter) then its perfectly fine to parallel the isolated outputs of these SMPSs. (obviously ensuring that the correct paralleling circuitry is included to ensure current share)

AC Line filter datasheet
**broken link removed**

-though you might expect a little worse Common mode problem in say one of the common mode chokes amongst the paralleled bank of smps modules, you surely wouldnt expect there to be a problem of common mode choke saturation. A power supply consultancy (not related to the module company) has said that we must pay them a big fee to do a special paralleled SMPS solution for us, using these synqor modules, because if we do it ourselves, by paralleling the smps module outputs, then we will suffer saturation of the common mode chokes. I believe this is nonsense, do you agree?
 
Just to make sure we understand your connections, a sketch would help/is important/is critical/is the end of the world. However, your description sounds ok. I've done parallel high density DC/DC converter projects with Astec, Vicor, and Synqor, and the vast majority of the problems have come from poorly or not-at-all documented "personality quirks" of the bricks, not from the basics like one brick/one filter.

EDIT - see next post.

ak
 
Last edited:
Just a moment... Just a moment...

Something about Synqor was nagging at me as I wrote that. There is a problem. With things connected as recommended by Synqor (independent input filter for each brick, no sharing except at the converter outputs) the system will not start up reliably. When power is applied to the inputs of all of the filters, each of the filter outputs comes up at a slightly different time, which means that each of the brick outputs comes up at a slightly different time, which means that the first brick up sees the entire load, doesn't know there are other bricks coming along to share the load, and goes into overcurrent shutdown. Etc. The Synqor current sharing system is based on digital data transferred among the bricks and an arbitration algorithm to determine which is the master brick and which are the slaves. This works fine if the bricks all come up together - really together, as in all inputs hard wired together. With separate filters that is not the case, and the algorithm does not handle this well. As personality quirks go, this is a biggie.

DISCLAIMER: This next part is a story, ***not*** a recommendation.

Against Synqor's recommendations, our solution was to tie together the outputs of the filters to create a common DC input bus to the bricks. This 100% solved the problem, and we have not seen any filter failures related to this connection.

ak
 
Just a moment... Just a moment...

Something about Synqor was nagging at me as I wrote that. There is a problem. With things connected as recommended by Synqor (independent input filter for each brick, no sharing except at the converter outputs) the system will not start up reliably. When power is applied to the inputs of all of the filters, each of the filter outputs comes up at a slightly different time, which means that each of the brick outputs comes up at a slightly different time, which means that the first brick up sees the entire load, doesn't know there are other bricks coming along to share the load, and goes into overcurrent shutdown. Etc. The Synqor current sharing system is based on digital data transferred among the bricks and an arbitration algorithm to determine which is the master brick and which are the slaves. This works fine if the bricks all come up together - really together, as in all inputs hard wired together. With separate filters that is not the case, and the algorithm does not handle this well. As personality quirks go, this is a biggie.

DISCLAIMER: This next part is a story, ***not*** a recommendation.

Against Synqor's recommendations, our solution was to tie together the outputs of the filters to create a common DC input bus to the bricks. This 100% solved the problem, and we have not seen any filter failures related to this connection.

ak

WOW AK,

I hadn't expected that problem. Something else I learned.

Would another approach be to fit a single higher current filter and parallel the power brick inputs after that?

spec
 
WOW AK, I hadn't expected that problem.
Neither did we. The system was done and the problem showed up in qualification testing. Scramble mode. Not counting that the problem was in their parts, Synqor's grade on helping us through it - C (barely).
Would another approach be to fit a single higher current filter and parallel the power brick inputs after that?
Of course, and if you don't need the super small form factor of a brick filter, there are tons to choose from. Corcom, Shaffner, Filter Concepts, Captor (MIL grade), Schurter ... Most of the old style big fat power line filters will have much better performance at the lower frequencies, like 10 kHz to 1 MHz.

Further clarification of the above story (brain is recalling things slowly today) - our system was 28 VDC input, so the filter devices were high power active filters, not all-passive AC power line filters. Hence the turn-on delay.

ak
 
When power is applied to the inputs of all of the filters, each of the filter outputs comes up at a slightly different time, which means that each of the brick outputs comes up at a slightly different time, which means that the first brick up sees the entire load, doesn't know there are other bricks coming along to share the load, and goes into overcurrent shutdown. Etc. The Synqor current sharing system is based on digital data transferred among the bricks and an arbitration algorithm to determine which is the master brick and which are the slaves. This works fine if the bricks all come up together - really together, as in all inputs hard wired together. With separate filters that is not the case, and the algorithm does not handle this well. As personality quirks go, this is a biggie.
Thankyou very much indeed, this is great info. The units are (or rather the quickest to come-on unit is) , as you imply, shutting down on overload protection, they think its a shorted or overloaded output...but i wonder if all started on totally no load, then it would be ok i presume as long as there was not too much output capacitance.?
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top