Author Topic: Possible to reduce output voltage ripple below 1mV?  (Read 13179 times)

0 Members and 1 Guest are viewing this topic.

Online onemilimeter

  • Frequent Contributor
  • **
  • Posts: 314
Possible to reduce output voltage ripple below 1mV?
« on: June 16, 2011, 04:30:42 am »
I've several DC/DC converters (Traco TEL 3-0523). Some of the specs are as follows:

Vin: 4.5V ~ 9V (nominal 5V)
Vout: +/- 15V @ +/-100mA
Ripple and noise (20 MHz Bandwidth): <60 mVpk-pk
http://www.tracopower.com/fileadmin/medien/dokumente/pdf/datasheets/tel3.pdf

Assuming that the output ripple is 60 mVpk-pk. Is it possible to reduce the ripple below than 1mV? Please share if you know how to achieve it. Thanks.
 

alm

  • Guest
Re: Possible to reduce output voltage ripple below 1mV?
« Reply #1 on: June 16, 2011, 07:42:12 am »
Sure, just add tons of filtering. Not drawing much current also helps. It may be more economical to get a cleaner regulator, though.
 

Online onemilimeter

  • Frequent Contributor
  • **
  • Posts: 314
Re: Possible to reduce output voltage ripple below 1mV?
« Reply #2 on: June 16, 2011, 08:45:08 am »
Sure, just add tons of filtering. Not drawing much current also helps. It may be more economical to get a cleaner regulator, though.

Thanks. Do you mean a linear regulator? In my application, I need isolation and +/-15V.
 

Online Zero999

  • Super Contributor
  • ***
  • Posts: 12157
  • Country: gb
  • Hero999
Re: Possible to reduce output voltage ripple below 1mV?
« Reply #3 on: June 16, 2011, 09:26:57 am »
You could choose a hybrid design, in this case a +/-16V switching regulator, with lots of filtering, followed by a +/-15V low drop out linear regulator.
 

Offline scrat

  • Frequent Contributor
  • **
  • Posts: 606
  • Country: it
Re: Possible to reduce output voltage ripple below 1mV?
« Reply #4 on: June 16, 2011, 09:28:18 am »
You could choose a hybrid design, in this case a +/-16V switching regulator, with lots of filtering, followed by a +/-15V low drop out linear regulator.
Agree. A switching for coarse regulation and a linear for noise suppression.
The problem could still remain in the switching emissions.
One machine can do the work of fifty ordinary men. No machine can do the work of one extraordinary man. - Elbert Hubbard
 

alm

  • Guest
Re: Possible to reduce output voltage ripple below 1mV?
« Reply #5 on: June 16, 2011, 09:45:46 am »
Make sure you pick a linear regulator with good ripple rejection at higher frequencies in that case. For example, the ubiquitous 78xx has quite poor ripple rejection above a few kHz or so, and possibly zero at the frequency of the harmonics. There are some linear regulators specifically designed as post regulator after a switching regulator, but they may not go to 15V, let alone negative voltages (who uses these high voltages in this day and age?). This appnote by the recently deceased Jim Willams may be of interest.

Even within switching regulators, there is a significant variation in ripple. A higher switching frequency reduces the ripple, inductor selection is also significant, more output capacitance will reduce ripple, and some switching regulators are just noisier than others. If you want low ripple, you probably want a modern switcher with a high switching frequency, not something ancient like an MC34063.
 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 28128
  • Country: au
    • EEVblog
Re: Possible to reduce output voltage ripple below 1mV?
« Reply #6 on: June 16, 2011, 11:09:53 am »
You could choose a hybrid design, in this case a +/-16V switching regulator, with lots of filtering, followed by a +/-15V low drop out linear regulator.

Yep, that would have been my suggestion.

Dave.
 

Offline ejeffrey

  • Super Contributor
  • ***
  • Posts: 1736
  • Country: us
Re: Possible to reduce output voltage ripple below 1mV?
« Reply #7 on: June 17, 2011, 12:26:41 am »
The best part about going to a hybrid setup is that you can allow voltage drop across your SMPS output filter while still staying above the dropout voltage of your linear regulator.  This makes it possible to use multi-stage RC filters or an active capacitance multiplier circuit without compromising the load regulation.

Unless you have a very tight power budget I would go a bit higher than 16 V.  Make a +20 SMPS output for a 15 volt supply.  Now you can put a 1 uF capacitor, a 20 ohm series resistor, and another 1 uF capacitor before your linear regulator.  At 100 mA current load you now have 18 V input to your linear regulator (In this case you can use a standard, rather than LDO regulator).  Or, if you prefer, drop the switch mode supply down to +18 V and use an LDO.

A capacitance multiplier looks a lot like a standard voltage regulator, but it lacks a reference voltage -- it tracks the input at low frequency.  However, they can have excellent ripple rejection at high frequency, much better than a standard linear regulator.  It only takes 4 components: an NPN transistor, two resistors, and a capacitor.
 

Offline slburris

  • Frequent Contributor
  • **
  • Posts: 520
Re: Possible to reduce output voltage ripple below 1mV?
« Reply #8 on: June 17, 2011, 12:55:23 am »
If you are into reducing noise into the nV range, some ideas from
Wenzel may be of interest:

http://www.wenzel.com/documents/finesse.html

Scott
 

Offline hannobisschoff

  • Contributor
  • Posts: 38
  • I get high on solder fumes.
Re: Possible to reduce output voltage ripple below 1mV?
« Reply #9 on: June 17, 2011, 08:39:08 pm »
Hi all. I would like to use this topic as an opportunity to ask a question.

I just built a "lab" power supply, with 50V output @2 Amps.
The output ripple at full load is about 10mV
Is that any danger to projects possibly being damaged?
In which cases would you want a supply with less than a 1mV ripple?

 

Online onemilimeter

  • Frequent Contributor
  • **
  • Posts: 314
Re: Possible to reduce output voltage ripple below 1mV?
« Reply #10 on: June 17, 2011, 09:18:53 pm »
Hi all. I would like to use this topic as an opportunity to ask a question.

I just built a "lab" power supply, with 50V output @2 Amps.
The output ripple at full load is about 10mV
Is that any danger to projects possibly being damaged?
In which cases would you want a supply with less than a 1mV ripple?

I think output ripple of 10mV is very good considering that full output voltage is 50V (i.e. the ripple is only 0.02% of the FS output voltage). Mind to share your design here? :)

In my case, I wish to provide a clean supply to analog circuit (e.g. signal conditioning and ADC).
 

Offline ejeffrey

  • Super Contributor
  • ***
  • Posts: 1736
  • Country: us
Re: Possible to reduce output voltage ripple below 1mV?
« Reply #11 on: June 18, 2011, 05:07:35 am »
10 mV ripple won't damage any circuit I have ever heard of.  The normal reason to need a very quiet supply are to reduce noise in analog circuits.

For instance, a typical op-amp has around 100 dB of power supply rejection at DC and low frequency.  Assuming you don't screw that up in your circuit, 10 mV of supply ripple gives you 100 nV of ripple on your signal.  A good opamp has ~5 nV noise voltage in a 1 Hz bandwidth, so 100 nV can completely dominate your system noise.  PSRR is worse at higher frequencies used in switch-mode supplies and in discrete transistor circuit, so in those cases the supply restrictions are even worse.

The worst case is when the supply noise actually directly enters your signal path.  For instance, excitation voltages for sensors or reference voltages for ADCs.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf