Author Topic: Voltage regulator question(s)  (Read 1735 times)

0 Members and 1 Guest are viewing this topic.

Offline dsharp02

  • Contributor
  • Posts: 26
  • Country: us
Voltage regulator question(s)
« on: March 17, 2016, 03:52:04 am »
If I were to use a switching regulator to set the voltage to a desired voltage +  the LM317 dropout voltage, and then let a LM317 regulate that voltage down to the desired voltage, how much of the switching noise would make it through to the circuit?  What specs on the datasheet should I be looking at to answer this question?

Thanks,
Dave
 

Offline NilByMouth

  • Regular Contributor
  • *
  • Posts: 71
  • Country: gb
Re: Voltage regulator question(s)
« Reply #1 on: March 17, 2016, 04:43:54 am »
Dumb question: Why are you using a switching regulator AND a linear regulator?
 

Offline suicidaleggroll

  • Super Contributor
  • ***
  • Posts: 1455
  • Country: us
Re: Voltage regulator question(s)
« Reply #2 on: March 17, 2016, 04:45:30 am »
If I were to use a switching regulator to set the voltage to a desired voltage +  the LM317 dropout voltage, and then let a LM317 regulate that voltage down to the desired voltage, how much of the switching noise would make it through to the circuit?  What specs on the datasheet should I be looking at to answer this question?

Thanks,
Dave

I believe that would be "Ripple rejection", which is on the order of 60dB for the LM317.

Dumb question: Why are you using a switching regulator AND a linear regulator?

The switcher gets it close to the desired voltage, then the linear cleans it up.  Better efficiency than a linear alone, less ripple than a switcher alone.  With an LDO you might drop your efficiency by a few percent, while reducing your output ripple by a couple orders of magnitude.  Depending on the application this can be a good tradeoff.
« Last Edit: March 17, 2016, 04:49:15 am by suicidaleggroll »
 

Offline Audioguru

  • Super Contributor
  • ***
  • Posts: 1266
  • Country: ca
Re: Voltage regulator question(s)
« Reply #3 on: March 17, 2016, 06:23:27 am »
I believe that would be "Ripple rejection", which is on the order of 60dB for the LM317.
Only if the switching regulator produces 120Hz ripple which I doubt. its frequency is much higher. But a graph of ripple rejection vs frequency is shown on the datasheet.
 

Offline suicidaleggroll

  • Super Contributor
  • ***
  • Posts: 1455
  • Country: us
Re: Voltage regulator question(s)
« Reply #4 on: March 17, 2016, 06:31:22 am »
I believe that would be "Ripple rejection", which is on the order of 60dB for the LM317.
Only if the switching regulator produces 120Hz ripple which I doubt. its frequency is much higher. But a graph of ripple rejection vs frequency is shown on the datasheet.

Missed that graph, thanks.  So it's around 60dB from DC through ~20kHz, then starts falling off at about 20-30dB/decade above that.
 

Offline NilByMouth

  • Regular Contributor
  • *
  • Posts: 71
  • Country: gb
Re: Voltage regulator question(s)
« Reply #5 on: March 17, 2016, 09:56:42 pm »
Dumb question: Why are you using a switching regulator AND a linear regulator?

The switcher gets it close to the desired voltage, then the linear cleans it up.  Better efficiency than a linear alone, less ripple than a switcher alone.  With an LDO you might drop your efficiency by a few percent, while reducing your output ripple by a couple orders of magnitude.  Depending on the application this can be a good tradeoff.

True, but it depends on the power loss. I read if it's less than 0.5 watt, the extra cost for the switching reg doesn't justify the improvement.
 

Online digsys

  • Supporter
  • ****
  • Posts: 1710
  • Country: au
    • DIGSYS
Re: Voltage regulator question(s)
« Reply #6 on: March 17, 2016, 10:13:35 pm »
I've done it MANY times over the years. It's a well accepted practice. Often, it has nothing to do with cost. I even think Dave has a video on it.
"We" assist the HF noise issue using various means  - CM chokes or filters, RF beads, LC filters etc etc
This method is extremely useful when you have a high(ish) DC rail - even up to 36V+ and need 3V3, 5V up to 1A etc without wasted shltloads of heat (energy).
Most pre-made modules have ultra-low ripple in the first place (and now efficiencies 95%+)
Hello <tap> <tap> .. is this thing on?
 

Offline Alex Trofimov

  • Contributor
  • Posts: 43
  • Country: ru
Re: Voltage regulator question(s)
« Reply #7 on: March 17, 2016, 10:53:48 pm »
True, but it depends on the power loss. I read if it's less than 0.5 watt, the extra cost for the switching reg doesn't justify the improvement.

I depends again. :-) For example, if you have a battery powered equipment, wasting 0.5 W may cause to using a huge and expensive battery, which will be expensive to deliver, your equipment is heavy and too big with it and that sort of stuff. There's a lot of little thigies when you make the design, and when you put all them together the device is being born. :-) The statement you claim is just too simple.

 

Offline suicidaleggroll

  • Super Contributor
  • ***
  • Posts: 1455
  • Country: us
Re: Voltage regulator question(s)
« Reply #8 on: March 18, 2016, 02:08:15 am »
Dumb question: Why are you using a switching regulator AND a linear regulator?

The switcher gets it close to the desired voltage, then the linear cleans it up.  Better efficiency than a linear alone, less ripple than a switcher alone.  With an LDO you might drop your efficiency by a few percent, while reducing your output ripple by a couple orders of magnitude.  Depending on the application this can be a good tradeoff.

True, but it depends on the power loss. I read if it's less than 0.5 watt, the extra cost for the switching reg doesn't justify the improvement.

Says who?  Whoever is willing to make such a general statement like that doesn't know what they're talking about (or it was simply taken out of context).  There are MANY applications where an extra dollar or two on the switching reg and associated components is peanuts compared to a 50+% reduction in power consumption with equivalent ripple.  Wouldn't you rather spend a dollar more on your phone if it meant it could last 18 hours instead of 12 between charges?  Who cares if it saved less than 0.5W if it made a marked improvement in power consumption?  It's about percent improvement, not raw power numbers.

For any given application, you should make a table (even if only in your head), with columns for cost, board space, power consumption, and ripple.  Your rows would be switcher alone, linear alone, switcher+linear, and maybe switcher+ldo (cutting it closer than with a regular linear reg).  Compare the numbers and decide for yourself what the best approach is for that application.

Anywhere you have a big delta between the input voltage and the operating voltage, and ripple is of importance (ADCs, DACs, etc), the switcher+linear combo is going to make a really good case.
« Last Edit: March 18, 2016, 02:28:44 am by suicidaleggroll »
 

Offline NilByMouth

  • Regular Contributor
  • *
  • Posts: 71
  • Country: gb
Re: Voltage regulator question(s)
« Reply #9 on: March 18, 2016, 02:39:16 am »
Yes, sorry that was a bit too much of a generalisation - based on a low cost project you'd need to weigh up the benefits of using a £10+ regulator, which might double the cost of parts.
 

Offline Evan.Cornell

  • Regular Contributor
  • *
  • Posts: 51
  • Country: us
Re: Voltage regulator question(s)
« Reply #10 on: March 18, 2016, 02:46:24 am »
I believe that would be "Ripple rejection", which is on the order of 60dB for the LM317.
Only if the switching regulator produces 120Hz ripple which I doubt. its frequency is much higher. But a graph of ripple rejection vs frequency is shown on the datasheet.

Missed that graph, thanks.  So it's around 60dB from DC through ~20kHz, then starts falling off at about 20-30dB/decade above that.

This is a critical point when using a linear regulator to reduce switching ripple. One must know the switching frequency of the DC-DC converter (is it fixed frequency, or will frequency jump all over the place depending on load current?), and then ensure that the linear regulator (LDO is preferable to reduce power loss; i.e. 5V in, 4.75V out) has sufficient PSRR at the switching frequency of the DC-DC.

This is a common technique as another poster stated, particularly in sub-systems that require low noise (ADC, DAC circuits).
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf