Author Topic: Going against v-reg datasheet recommendation for more robust design  (Read 834 times)

0 Members and 1 Guest are viewing this topic.

Offline sebmadgwickTopic starter

  • Regular Contributor
  • *
  • Posts: 129
  • Country: gb
    • YouTube Channel
I am using the TPS63000 to generate 2.2V and 3.3V supply rails.  The TPS63000 output voltage is set by feedback from a potential divider as shown in the attached schematic.

The datasheet (page 11) recommends the use of high resistor values, "...keep the value for this resistor in the range of 200 kΩ" and "...a 1-MΩ resistor should be chosen...".  TI's WEBENCH Designer recommends the specific values shown in the attached schematic.

I am uncomfortable using such high resistances because of the risk of contaminants or even handling of the PCB resulting in destructively high voltages.  My preference would be to use resistor values one or two orders of magnitude lower.  The total device current consumption is 100-300 mA so the increased quiescent loss is not of concern.

Is there any reason I should stick to TI's recommendation here?  The recommendation seem specific without justification.
« Last Edit: December 16, 2019, 08:48:17 pm by sebmadgwick »
 

Online iMo

  • Super Contributor
  • ***
  • Posts: 5570
  • Country: va
Re: Going against v-reg datasheet recommendation for more robust design
« Reply #1 on: December 16, 2019, 08:51:09 pm »
The divider is at the output, therefore no need to stick with such values.
Readers discretion is advised..
 

Offline wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Re: Going against v-reg datasheet recommendation for more robust design
« Reply #2 on: December 16, 2019, 08:59:27 pm »
Is there any reason I should stick to TI's recommendation here?  The recommendation seem specific without justification.
Justification is simple, quiescent current and intended operation from battery in portable devices. By using low resistance divider, it can easily consume more than IC itself.
 

Online iMo

  • Super Contributor
  • ***
  • Posts: 5570
  • Country: va
Re: Going against v-reg datasheet recommendation for more robust design
« Reply #3 on: December 16, 2019, 09:05:19 pm »
The DS says the current through the divider should be at least 100x higher than the current into the FB pin (10nA). As wraper wrote they want to show high efficiency and low quiescent current therefore they recommend the values at higher side.
Readers discretion is advised..
 

Offline sebmadgwickTopic starter

  • Regular Contributor
  • *
  • Posts: 129
  • Country: gb
    • YouTube Channel
Re: Going against v-reg datasheet recommendation for more robust design
« Reply #4 on: December 17, 2019, 12:37:30 am »
The quiescent current of the IC is ~5 mA without Power Save and it is rated for up to 1.5 A output.  It therefore seems strange that recommendations would optimise to save a few uA without qualifying this.  The datasheets sates, "...keep the value for this resistor in the range of 200 kΩ", unconditionally.

Anyway, it looks like I was reading too much into this.  Thank you for your comments.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22436
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Going against v-reg datasheet recommendation for more robust design
« Reply #5 on: December 17, 2019, 04:16:53 am »
Conversely, I've seen datasheets with:
- ca. 10k dividers recommended,
- no specification for leakage current of FB pin,
- no discussion of how to choose those resistors (the 10k range is simply assumed as "good enough")

Had to contact TI via E2E forum to get an answer; turns out it was ~nA, as you expect for CMOS inputs.  With the usual fluff about "we don't measure it because it takes too long", of course they didn't have anything to say about "why not at least 'max 1uA' or such?"

Probably the DS/AN author of the part I was looking at, should've swapped roles with the one you're looking at! ::)

Generally, you can expect CMOS chips to use CMOS input structures (i.e., ESD diodes and gates), having very little leakage (say, 10 or 100nA for a bad part on a hot day); likewise, bipolar has to supply base current, so tends to pull up (or down) with some ~uA.  But when you need to know -- better off measuring it, get a sample or dev board and check.  Second best, asking the manufacturer...

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 
The following users thanked this post: sebmadgwick


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf