2
\$\begingroup\$

I am implementing a Closed loop Current Mode Control Buck Converter on a dsPIC Microcontroller which has an output of 3.3V and unregulated input of 5V.

I am doing digital control, so I implemented the PI controller as 2p2z controller.

Someone asked me why I am doing all this? Is there not an ASIC chip which would already have code for 3.3V output and an unregulated input voltage range? The ASIC chip would have the control loop already implemented with all the interrupts and stuff.

Is there such a thing in market? If so, why is ASIC preferable or why not?

\$\endgroup\$
6
  • \$\begingroup\$ You're using a dsPIC as a dc-dc converter? \$\endgroup\$ Nov 29 '19 at 14:49
  • \$\begingroup\$ ASIC = great performance. no flexibility or control over what you want unless you have millions of dollars to get your own developed. \$\endgroup\$
    – DKNguyen
    Nov 29 '19 at 14:53
  • \$\begingroup\$ @DKNguyen You don’t need that much. $100 thousand is enough. \$\endgroup\$
    – user110971
    Nov 29 '19 at 14:59
  • \$\begingroup\$ @user110971 good to know, point still stands though, of course \$\endgroup\$
    – DKNguyen
    Nov 29 '19 at 15:00
  • \$\begingroup\$ @DKNguyen Yes, of course. \$\endgroup\$
    – user110971
    Nov 29 '19 at 15:01
2
\$\begingroup\$

A quick search of Texas Instrument's products shows there are some 222 components between buck regulator controllers, integrated-switch converters, and fully-integrated modules that meet the requirements of 4.5-5.5 V in and 3.3 V regulated out. While there is some commonality between these components such as reuse of silicon dies in both converters and modules, it's fairly safe to say that there are over a hundred ASIC chips from just this one manufacturer that implement the control and sensing for a switching power converter. Yes, there is a market for these ASICs.

There are a wide array of power systems, starting with simple buck converters and going up to AC/DC supplies at kilowatts. You need to evaluate your design goals with perspective on the time/cost/complexity trade offs to achieve it, particularly if your end result is a salable product.

From my design experience, here are some reasons to use them over a custom-built solution:

  • Ease of integration. Someone's done all the stability and accuracy calculations for the core device and the designer just needs to choose values to plug into equations provided by the datasheet to meet the design goals (For TI, often provided in the "Applications and Implementation" section). There's no concern about loop rates, ADC speeds/resolutions, input limits, or architectures. Those are already decided.
  • Conservation of board area. Some of the design solutions are 10s of mm-square. At the board level, it would be impossible to achieve that small of a board area with a discrete microcontroller, FETs, resistors, inductor, and capacitor.
  • Reduction in component count. The fully-integrated modules may only require three components: module, input capacitor, output capacitor.
  • Supply chain assurance. Purchasing a device from a reputable manufacturer means you're less worried about end-of-life components which can affect how long you can manufacture the design and that is more risk the more different components you use.
  • Option of flexibility. The controllers leave compensation, switching speed, FET/diode, inductor, capacitor, and more available for selection or modification by the end user. This allows for cost optimization and tuning component selection to meet the varied requirements of the application.
  • Potentially increased performance. A lot of engineering effort has already gone into the design of the chip/module. The increased integration and already-tested specifications may be very advantageous in achieving performance goals.

The advantages of a fully custom solution include:

  • Knowledge and control over (more of) how the design works. There are no hidden coefficients stored in an inaccessible mask layer or EEPROM inside the chip. You know and control exact algorithm managing the implementation and thus can experiment or innovate on the control system.
  • Full design flexibility. If you want to choose every component (controller, gate drive, FETs, inductors, capacitors) and the architecture, you can do so. In fact, there are even microcontrollers from TI for this exact purpose.
  • Meeting your needs when ASICs can't. Particularly at high power, high voltage, or some other extreme characteristic, the ASIC designers may be unable to meet your needs. The engineering effort of completing and proving out the design may provide you with a competitive advantage by being novel on the market.

Overall, my recommendation: if this is a learning project, go ahead with it. Someone has to know how the inside of the system works and this may be a skill employers desire. If this is you needing 3.3 V at 2 A for something else, consider one of the ASIC options or even purchase an off-the-shelf buck converter circuit board.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.