0
\$\begingroup\$

Following my question Powering up analog part of the video card (thanks everyone for their inputs and advices) I have chosen the devices, and even have drawn the circuit diagram of 12 V to 5 V power converter:

enter image description here

Output current is expected to be approx 250 mA, maximum 500 mA.

I was very happy, until I found the following article, which made me very upset. It is in Russian, but I translate key points.

The TPS562200 is having some serious drawbacks:

  1. Causing it to physically burn if layout is not followed - in particular there must be no switching current under the chip. Good to know before routing the board.
  2. Vias are required as VBST and SW pins are put at the different sides of the chip. Also good to know.
  3. Very high ripple under low load.

The last one seems actually being killing the whole design. According to calculations, the current when device goes into Advanced Eco-Mode Control mode (datasheet, chapter 7.3.2, page 12) if load consumes less than 500 mA => in my device the regulator will always be in this mode.

Article considers this case with very demonstrative pictures:

enter image description here Above is an output for the "nominal" operating conditions.

enter image description here Above is an output for the 200 mA supply current. Ripple frequency is 65 kHz.

enter image description here Above is an output for the 50 mA supply current. Ripple frequency is 1.2 kHz.

This noise would be a huge issue for the analog circuits. I see two ways to deal with it:

  1. change regulator with one rated for smaller current - for example, TPS560200, then "The transition point to the light load operation" will happen at 230 mA. However I suspect most of the times less current will be consumed from the supply, and it may easily happen that switcher will work in this featured ecomode.
  2. Keep this ADP3339 High Accuracy, Ultralow IQ, 1.5 A, anyCAP Low Dropout Regulator. It has very nice ripple rejection property:

enter image description here

and this ripple under 100 kHz is expected to be attenuated greatly.

Questions:

Is this ripple effect will take place under "lighter" loads for any switcher, or it is a "feature" of these TPS ones?

Do you think the design is acceptable for powering analog video circuits - would selected LDO cure the issue created by the TPS?

Maybe you can propose something better than I have drafted?

Update:

following Marcus's advice I tried Webench (previously I was using it in relation to TPS562200 only). I set requirements of 12 Vin, 6 Vout, max current 0.5A, and nominal current of 0.2A... and have got 53 results, one of them is this TPS562200!

I forgot one of the requirements I have: chips must be solderable by hand by not-so-experienced individual. Thus most appropriate packages are SO or SOT.

\$\endgroup\$
5
  • 1
    \$\begingroup\$ hm, if your design is always below the "this is a significant load" threshold, maybe that's simply not the right switcher – how did you end up with the tps56? \$\endgroup\$ – Marcus Müller Oct 2 '19 at 14:34
  • \$\begingroup\$ (by the way, ti.com has a pretty cool "webdesigner" thing for switch-mode power supplies that I can only recommend) \$\endgroup\$ – Marcus Müller Oct 2 '19 at 14:37
  • \$\begingroup\$ The answer is simple and stupid: this was the one I found available for purchase through performing channels. There're many devices of different models in the store, but I selected these ones as having best documentation and most fitting the requirements. Maybe I did it wrongly. I am not skilled and knowledgeable in the switcher market.... I used Webench and it is a fantastic tool, but still do not know how to select the right part for my task. I was sure many people have successfully implemented 12->5V conversion, but many designs I looked up on the web are unacceptable for various reasons! \$\endgroup\$ – Anonymous Oct 2 '19 at 14:39
  • \$\begingroup\$ Webench will recommend you a part! You actually don't need to know much :) \$\endgroup\$ – Marcus Müller Oct 2 '19 at 14:40
  • \$\begingroup\$ Updated question. Seems it is not that easy, and each Webench proposed option will require very close consideration anyway. This is an amount of information to consider... \$\endgroup\$ – Anonymous Oct 2 '19 at 14:49
2
\$\begingroup\$

The light load efficiency mode (PFM) greatly improves the efficiency in the range where it's operational, at the expense of higher ripple and variable switching frequency.

It's a hysteretic mode, so frequency depends on the load current.

There are other parts from TI and other manufacturers that are PWM only, or selectable between forced PWM and auto-PWM/PFM modes. One of those may be more appropriate if you're not concerned about the efficiency loss.

One example is the TPS562208, though a smaller device could be used based on your load requirements.

If you want to keep the switcher as-is, (for efficiency or convenience) the LDO will certainly help, but be sure you have enough headroom above the dropout voltage to get the PSRR you need. As you approach dropout, the PSRR will fall off rapidly. The datasheet doesn't do a good job of specifying this, so you may need to experiment.

\$\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.