• State Resolved
  • Locked Locked
  • Replies 5 replies
  • Subscribers 63 subscribers
  • Views 356 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

isolated ac/dc convertor selection guide?

Kevin Braswell
Intellectual 295 points
Other Parts Discussed in Thread: UCC28610, UCC28711

I'm new to AC/DC power conversion and need to design one that offers...

 - isolation

 - Vin = 120Vac(typ), "universal" AC input might be nice but not a hard requirement

 - Vout = 3.6~4.8V

 - Iout(max) = 0.55Adc (2.0Apeak) - load regulation can be relatively poor (~5%) during peak current

 - Higher efficiency is always nicer, but ~70% efficiency at full load is probably adequate

 - ~0.1W loss at no load/light load (Iout<0.02A) is probably adequate

 - There's no compelling need for PFC

 - Board area, height, and cost are concerns

I'm looking for basic guidance on choosing a topology, as well as selecting a controller, design guides, reference designs, etc.

I haven't found anything on the TI website for topology selection (but other resources have indicated a flyback topology may be best suited - and I see that TI offers multiple flyback controllers).  Does TI have a design guide that covers isolated AC/DC converter topology selection?

I tried using WeBench to select a controller, but it didn't give any options.  I also reviewed the Power Management Guide - PWM and Resonant Controllers Selection Guide.  It indicated that any controller that could support up to 2.0A could NOT support a "supply voltage" (Vout?) down to 4.2V.  So I'm concluding that TI does not have a controller suitable for this application.  But there seems to be a market for these controllers, so I'm concerned that I'm missing something.  Can someone either confirm that TI is not in this market, or set me straight on how to read the selection guide?

Thanks!

  • 0
    TI__Genius 15910 points

    Flyback would be my go-to choice here for cost, isolation, power level concerns.  Quasi resonant controller for the no-load power consumption is  also a good idea. The comment about the supply voltage is referring to the VDD voltage to bias the IC, usually around 10V to 20V depending upon the controller...it's the voltage to supply bias to the chip (= supply voltage).  I would point you towards the UCC28610 as there's a design calculator that is an excel worksheet that will hand you the components required...just be sure that the design is for 2A output current, not 0.55A as the supply needs to be designed for peak power (that might impact your small size but look at it anyway). http://www.ti.com/general/docs/lit/getliterature.tsp?literatureNumber=sluc113f&fileType=zip for the file but I started it out for you already in the attachment.  If you have MathCAD, i have a file for the UCC28610 that has comments in it that may help you walk through the design calculator.

    For a beginner, keep in mind that layout is as important as component selection...you can have all the right components and still not get the thing to turn on if you layout is bad so heed ALL of the recommentdations about close, fat traces, small loops, etc (we aren't kidding when we tell you these things!)

    There's plenty of power supply design seminars out there from TI/Unitrode that are good resources, http://www.smps.us/Unitrode.html is the best way to find them as even I can't locate them on the TI website (sad but so true).

    What is the end-product for this application?  Any idea as to the potential units per year?  If you have more specific requirements, there might be another controller that would be more suitable for the application.

  • 0 in reply to Lisa Dinwoodie
    TI__Genius 15910 points

    oops, forgot to add the attachment that I told you about....

    UCC28610 Design Calculator SLUC113.xlsx
  • 0 in reply to Lisa Dinwoodie
    Intellectual 295 points

    Hey Lisa,

    Thanks for the Unitrode link.  That does appear to have some decent background info.

    And thanks for the "supply voltage" clarification.  I've done plenty of DC/DC converters (so I'm aware of the importance of layout, and following the datasheet).  With DC/DC controllers, the output voltage (and possibly output current limit) is typically defined by the controller silicon.  Perhaps with most (all?) of these AC/DC converters, the output voltage is set by the transformer, so it's not readily defined, or limited by the controller.  So I'm getting used to that.

    As for the quantity / application...

    I've been contracted by a client to do the design.  They have not indicated quantity, so I can only guess.  It's a specialized application (not in the consumer market).  I would guess quantities are 1k~10k per year.

    As for application...

    Incoming power is US 120Vac - I don't think this customer is targeting any markets outside the US.

    Output power (~4.2V) is to power a cellular wireless module (for remote data connection) and possibly charge a battery (at up to ~100mA).  The wireless module still relies on 2G air interface, so the peak current can be up to 2.0A during Tx burst (2 slot max = 1.154ms, 25% duty cycle).  Most of the time, the wireless module is Rx-only and the load current will be <25mA.  Calls are infrequent and brief (maybe only 3 a day, each one lasting <10s), and that's when the AVG current can be up to 0.55Adc (with 2.0A peaks).

    Board area and height are notably limited.  Length can't exceed ~2 1/4".  Width can't exceed ~1 3/4".  Height can't exceed ~1".  This will also be surrounded on all 6 sides by PCB and mechanics, so there's little opportunity for convective heat loss.  Conductive heat loss (PCB heatsinking) will be used as much as is possible, but power loss (self-heating) needs to be kept under control.

    Is there a reference design / eval kit for the part you suggested?

    Thanks!

    Kevin Braswell

  • 0 in reply to Kevin Braswell
    TI__Genius 15910 points

    10 seconds is an eternity...just sayin'

    The UCC28610 does have an EVM but it is considerably larger than 2.25 x 1.75 inches because the layout was more intended for elbow-deep analysis of the waveforms and not designed to meet a customer's final form fit and function so that would be up to you to do.

    There is also a part that is targetted for USB charger applications, UCC28711, that may be perfect for you and it also has an NTC option (negative temp coefficient). I think you should look at that and if that looks like it may be a contender but there's a feature or something else that you are concerned you can send me an email and we can discuss a few other options.

  • 0 in reply to Lisa Dinwoodie
    Intellectual 295 points

    Well, the 10 second reference was just for a thermal perspective.  I.e., self-heating while on a call is _somewhat_ transient and there probably won't be enough time to reach thermal steady state while on a call.

    An EVM would be a great help, but only as a leg up for prototyping and evaluating board area requirements.  No way would I even consider (or expect to find) an EVM suited for a customer's application.

    I'll dig around and look at the UCC28xxx products.

    Thanks!