Switch Mode ISOLATED Power supply with low ripple

[visunaik]

I want to design an SMPS with 2 outputs with different grounds isolated from each other.Kindly let me know any articles/papers/websites/tips for this.The high and low frequency ripple should be less than 50mV for +/-12V output.

Thank you

 

[jbartos]

Suggestion/question: Would not it be more economical to consider two SMPS, physically different? Or is there any specific constraint such as weight, volume, etc. since your SMPS might need customization or development, certification, etc., which is costly and time-consuming?

 

[visunaik]

Dear jbartos,
The SMPS has to be small in size,cost less and enclosed within the same enclosure.Since the form factor must not be greater than 70X70X140, it seems difficult to accomodate two different supplies.Also the currents involved are less, in the range of 500 to 700 mA only on all outputs.

Your suggestions please...

 

[don01]

hi visunalik
Jbartos is right here you want two grounds then you aint gonna have many common components after you stick the square wave into the torriod in fact you may use the same former but you will need to isolate the secondaries from here at least, to get a full isolation. Why not think in terms of 2 supplies in 1 box I can't think of a small package with 2 isolated supplies. but you could do it if you repeat the cct twice

Regards
Don

 

[jbartos]

Suggestions/question:
1. Please, provide the input voltage level
2. Visit

http://www.thomasregister.com

and type dc-dc converter
which will return various converter catagories and many companies to search for or inquire about the power supply

 

[h1961]

Hi visunaik.
The IC UC3842/4 by TI is adapted for this function.
Visit:

http://www.ti.com

Regards.
Hugo.

 

[jbartos]

Suggestions to the previous posting:
1. Please, could you be more specific with the web site address or any search clue.
2. What about the input voltage that is not identified in the original posting?

 

[pamcinto]

visunaik,

There are a set of assumptions that I must make to provide you any information on this matter. If any of these assumptions are incorrect, please reply to this post.

-The form factor of 70X70X140 is a measurement in millimeters.
-Since a switch-mode power supply can be both a rectifier (AC/DC) or a DC/DC converter, I will assume that you mean DC/DC.
-You mentioned a 12V output is required...I will assume that the other output is also low voltage (<24V).

While there are quite a few multi-output power supplies on the market (eg.

http://www.vicr.com/products/vipac/vipac.htm

), these are almost always separate modules within one single package. They are also almost always higher power than what you have specified. It is quite possible to use one common transormer core for both outputs, but to be realisitic, this does not provide that large a savings since the core must be sized larger to accomodate the increased flux. The other reason is that manufacturers can simplify their lives and maximize their flexibility by keeping this modular approach. (eg.

http://www.vicr.com/products/dc-dc/converter-modules/2nd-gen.htm

)

The &quot;2nd-gen&quot; modules shown in the previous link are the very highest power density modules available in the industry right now. Even the &quot;half-brick&quot; has way more power than you are talking about but it may be possible to squeeze two of these on a circuit card (along with mandatory filter components) into the form factor you've described. This unfortunately will require you to produce a circuit card and spec-in some filter components for the outputs.

Now, all this asside, there is another possibility to be considered. If it so happens that you can run off AC line voltage, a computer power supply delivers some of the most common DC outputs all within a decent form factor. These power supplies can be noisy, but you can easily throw a power filter on the output of the appropriate voltages and voila!...a very inexpensive, fairly small power supply for your application.
Now I realize that you mentioned a 12V output and the closest in a computer supply is 15V. I've suggested this because in alot of instrumentation applications, the electronics are able to cope with a large range of supply voltages and many times, either 12 or 15V rails can be used.
If the AC line voltage cannot be used (battery operated/backed system) then open up the supply and meter out some leads to see if you can just bypass the front-end recitification of the AC power.

OK, so if none of this helps, there is always the option of creating your own DC/DC power supplies and with the power levels you're talking about, the material cost of this option will be very inexpensive especially if you can salvage some surplus store parts. The down-side to this option is that you will need somebody who is trained in this to create the design and it will involve a large amount of time to get it right.

I hope that something in this attrociously long posting helps you. Please bear in mind that the more detail you can provide in your original post, the more likely you are to get an appropriate answer (application, input/output voltage levels and currents/power required, noise tolerances) but I also understand that this is sometimes not for public ears.

Cheers,
Peter (Power and Instrumentation Design Engineer)

 

[jbartos]

Suggestions:
1. Please, notice that two power supplies still may or may not guarantee the separate grounds since some galvanic connection and common mode currents may exist.
2. To be sure, one can look into isolation transformer incorporation with two separate secondary windings and keeping one power supply only.
3. Notice that the transformer may be custom made and may have rectifiers on the output.
4. Obviously, the device is dependent on nature of the input voltage. This was called for in my previous posting.

 

[pamcinto]

It seems as though visunaik may have flown the coop since it has been over a week since his/her last posting. Perhaps this is a dead string....

 

[visunaik]

Hi gurus,
First of all, Thank you for the overwhelming response to my query.
This string is not dead, it is live and cicking....pamcinto!Thanks for your long and enlighting (first) reply

I was waiting for collection of replies to form my own opinion.I have visited the sites mentioned above and found them very helpful.Anyway, the emerging scenario suggests that a two supply approach will be better.
For those interested in knowing the exact specs:
AC mains: Indian 160 to 270 V, 50 Hz

Outputs:
first isolated:
+12V,-12V, Analog gnd : approx. 500mA each
+5V, Digital Gnd : approx. 1 A

second isolated:
+5v,-5v: approx 500 mA each

isolation voltage: between AC to DC side and between both isolated sides at least 1.5 kV for 1 minute.

I am also thinking of deploying the well known technique of flyback convertor with top switches and optical isolation for feedback.

Keep on suggesting some more things...

Thanks again

 

[jbartos]

Question: How are you going to galvanically separate two outputs without separate transformer windings? The flyback converters without T1 and T2 transformers will be preserving the galvanic path to the source, 160-270V 50Hz. Visit

http://w4dfu.ufl.edu/listarchives/1998/january/msg00290.html

http://schof.colorado.edu/~pwrelect/book/contents/TOC.html

http://www.hills2.u-net.com/electron/smps.htm#fwdfly

(see for T1 transformer)
etc. for more info