Author Topic: source needed for very low ripple isolated DC-DC converters  (Read 1641 times)

0 Members and 1 Guest are viewing this topic.

Offline graybeardTopic starter

  • Frequent Contributor
  • **
  • Posts: 431
  • Country: us
  • Consulting III-V RF/mixed signal/device engineer
    • Chris Grossman
I am working on a fast turn around project in which I need very low noise PCB mounted isolated 3.3V output DC to DC converters.  I need sub mV output ripple.  I can use multiple converters for lower noise.  Multiple converters with output of 200mA-1.1A or a single converter with an output of 650mA to 3.2A.   Based on DRE I choose an input voltage of 24V, but I am not married to that.  I only need 40V maximum of isolation.  Low output noise is my primary concern.

Linear regulators do not do a great job on noise in the 100 KHz to 100s of Mhz that comes out of the converters I have looked at.

I was hoping to buy something off-the-shelf since I have many design challenges and I do not want to spend time designing filters for the converters.  It seems I may have no choice but to devote time to filter design.

I have talked with the technical reps from several companies, but they just tell be I will need to design my own filters and refer me to their application notes.

Do you know of any companies that I should contact?

Thank you
Chris

Offline fourtytwo42

  • Super Contributor
  • ***
  • Posts: 1182
  • Country: gb
  • Interested in all things green/ECO NOT political
Re: source needed for very low ripple isolated DC-DC converters
« Reply #1 on: June 01, 2023, 09:26:49 am »
DC-DC modules always have limited internal filtering as that is sufficient for the bulk of applications, best get on and add the external filter components, I am sure the vendor will provide you with the required components to suit there module rather than you having to do the design. Don't forget input noise can be a problem with these devices too.
 

Offline tszaboo

  • Super Contributor
  • ***
  • Posts: 7281
  • Country: nl
  • Current job: ATEX product design
Re: source needed for very low ripple isolated DC-DC converters
« Reply #2 on: June 01, 2023, 11:15:55 am »
Those modules are not going to be low noise. The small transformer is driven really hard, sharp edges, and often driven into saturation. What you do is: take one which gives you higher voltage than needed, then burn the voltage with an LDO or linear regulator. Depending on your noise level, you may need some common mode noise filtering inductors.
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 26682
  • Country: nl
    • NCT Developments
Re: source needed for very low ripple isolated DC-DC converters
« Reply #3 on: June 01, 2023, 01:21:41 pm »
I am working on a fast turn around project in which I need very low noise PCB mounted isolated 3.3V output DC to DC converters.  I need sub mV output ripple.  I can use multiple converters for lower noise.  Multiple converters with output of 200mA-1.1A or a single converter with an output of 650mA to 3.2A.   Based on DRE I choose an input voltage of 24V, but I am not married to that.  I only need 40V maximum of isolation.  Low output noise is my primary concern.

Linear regulators do not do a great job on noise in the 100 KHz to 100s of Mhz that comes out of the converters I have looked at.

I was hoping to buy something off-the-shelf since I have many design challenges and I do not want to spend time designing filters for the converters.  It seems I may have no choice but to devote time to filter design.

I have talked with the technical reps from several companies, but they just tell be I will need to design my own filters and refer me to their application notes.
What I normally use is a common mode filter at the outputs of DC-DC converter modules (preferably shielded models) with a large electrolytic (say 330 uf) at the DC-DC side and an MLCC at the side of the circuit. Don't forget a ceramic capacitor between in / out of the DC-DC converter to short the leakage current. For sensitive circuits, I use linear regulators with an RC filter at the input so only load / line regulation are an issue.
« Last Edit: June 01, 2023, 01:23:45 pm by nctnico »
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 
The following users thanked this post: Someone

Offline Peabody

  • Super Contributor
  • ***
  • Posts: 1954
  • Country: us
Re: source needed for very low ripple isolated DC-DC converters
« Reply #4 on: June 01, 2023, 02:49:13 pm »
You might look at Dave's video #1116 on using a capacitance multiplier to reduce ripple.
 

Offline Marco

  • Super Contributor
  • ***
  • Posts: 6686
  • Country: nl
Re: source needed for very low ripple isolated DC-DC converters
« Reply #5 on: June 01, 2023, 03:49:41 pm »
Fast turn around? Just throw money at it, or rather throw money at Analog Devices. LT8640A + LT3071.

PS. oops missed the isolated, LT1683 and LT3071 then.

PPS. guess I underestimate what a good passive filter can do, from the LT1683 datasheet :
« Last Edit: June 01, 2023, 07:35:12 pm by Marco »
 

Offline Faringdon

  • Super Contributor
  • ***
  • Posts: 1944
  • Country: gb
Re: source needed for very low ripple isolated DC-DC converters
« Reply #6 on: June 01, 2023, 06:32:19 pm »
You need isolation, and you want sub-mv ripple...you could use say four interleave switched flybacks 90 degrees apart.....go high frequency so you can use small filter...the interleaving will of course allow you to reduce ripple of itself.
But really.....the mV ripple on an SMPS output , ayk, is not the "problem noise" from an SMPS...the "problem noise" from an SMPS is the noise that you cant see as ripple on the output...its the high frequency common mode noise that exudes due to the switching edges.....so common mode chokes, and proper use of Y caps for that.....If its in a metal enclosure...proper y cap'ing to the enclosure...and of course, shielding oF the cables to/from the SMPS.

You often find a dcdc converter in a metal enclosure wont actually work until you y cap to the enclosure from cct ground....and use comm mode chokes too.

also remember at 500khz+ switching frequencies, small value ceramic caps can give less ripple than larger value ceramics....since less strays
« Last Edit: June 01, 2023, 06:35:39 pm by Faringdon »
'Perfection' is the enemy of 'perfectly satisfactory'
 

Offline Someone

  • Super Contributor
  • ***
  • Posts: 4493
  • Country: au
    • send complaints here
Re: source needed for very low ripple isolated DC-DC converters
« Reply #7 on: June 02, 2023, 12:56:35 am »
Those modules are not going to be low noise. The small transformer is driven really hard, sharp edges, and often driven into saturation. What you do is: take one which gives you higher voltage than needed, then burn the voltage with an LDO or linear regulator. Depending on your noise level, you may need some common mode noise filtering inductors.
As expanded on by nctnico, there can be significant common mode noise so post regulator LDO's aren't always going to reduce "noise" or coupled noise.....  switchers can be annoying when chasing "low noise".

I need sub mV output ripple.
When it comes to noise the question immediately after amplitude is: what bandwidth?
 

Offline Faringdon

  • Super Contributor
  • ***
  • Posts: 1944
  • Country: gb
Re: source needed for very low ripple isolated DC-DC converters
« Reply #8 on: June 03, 2023, 10:48:05 am »
Another point that people dont realise often, is that  if you have a dcdc in a metal enclosure......you can make an internal copper plane, and connect it to the enclosure.....and then it will be capacitively coupled to the main cct ground through the pcb.....which can solve noise issues.....so its like a y capacitor....i think rarely is the output voltage ripple on a switcher a serious noise issue.
'Perfection' is the enemy of 'perfectly satisfactory'
 

Offline temperance

  • Frequent Contributor
  • **
  • Posts: 400
  • Country: 00
Re: source needed for very low ripple isolated DC-DC converters
« Reply #9 on: June 03, 2023, 01:15:58 pm »
Quote
You need isolation, and you want sub-mv ripple...you could use say four interleave switched flybacks 90 degrees apart.....go high frequency so you can use small filter...the interleaving will of course allow you to reduce ripple of itself.
But really.....the mV ripple on an SMPS output , ayk, is not the "problem noise" from an SMPS...the "problem noise" from an SMPS is the noise that you cant see as ripple on the output...its the high frequency common mode noise that exudes due to the switching edges.....so common mode chokes, and proper use of Y caps for that.....If its in a metal enclosure...proper y cap'ing to the enclosure...and of course, shielding oF the cables to/from the SMPS.

You often find a dcdc converter in a metal enclosure wont actually work until you y cap to the enclosure from cct ground....and use comm mode chokes too.

also remember at 500khz+ switching frequencies, small value ceramic caps can give less ripple than larger value ceramics....since less strays

An over complicated solution it seems and a fly back for a low noise power supply is just the worse choice you can make. Don't you know that?
 

Offline Faringdon

  • Super Contributor
  • ***
  • Posts: 1944
  • Country: gb
Re: source needed for very low ripple isolated DC-DC converters
« Reply #10 on: June 03, 2023, 05:25:40 pm »
Thanks, is the  original poster  able to tell us why  they  needs a low noise power supply?
Also, , the 24v input, where does that come from?, an offline PSU, may i ask?.....that might be the best place to begin low noise effectations.
'Perfection' is the enemy of 'perfectly satisfactory'
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 26682
  • Country: nl
    • NCT Developments
Re: source needed for very low ripple isolated DC-DC converters
« Reply #11 on: June 03, 2023, 06:25:05 pm »
Quote
You need isolation, and you want sub-mv ripple...you could use say four interleave switched flybacks 90 degrees apart.....go high frequency so you can use small filter...the interleaving will of course allow you to reduce ripple of itself.
But really.....the mV ripple on an SMPS output , ayk, is not the "problem noise" from an SMPS...the "problem noise" from an SMPS is the noise that you cant see as ripple on the output...its the high frequency common mode noise that exudes due to the switching edges.....so common mode chokes, and proper use of Y caps for that.....If its in a metal enclosure...proper y cap'ing to the enclosure...and of course, shielding oF the cables to/from the SMPS.

You often find a dcdc converter in a metal enclosure wont actually work until you y cap to the enclosure from cct ground....and use comm mode chokes too.

also remember at 500khz+ switching frequencies, small value ceramic caps can give less ripple than larger value ceramics....since less strays

An over complicated solution it seems and a fly back for a low noise power supply is just the worse choice you can make. Don't you know that?
For industrial equipment you are often left to deal with a 24V DC supply coming from somewhere and powering all kinds of other stuff. But it is perfectly possible to use switching DC-DC converters in highly sensitive circuits in such a situation. The trick is to keep the HF currents contained at the DC-DC converters.
« Last Edit: June 03, 2023, 06:29:59 pm by nctnico »
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline Faringdon

  • Super Contributor
  • ***
  • Posts: 1944
  • Country: gb
Re: source needed for very low ripple isolated DC-DC converters
« Reply #12 on: June 03, 2023, 08:25:55 pm »
Exactly...and i reckon common mode noise is far worse than vout ripple....generally speaking.
'Perfection' is the enemy of 'perfectly satisfactory'
 

Offline Marco

  • Super Contributor
  • ***
  • Posts: 6686
  • Country: nl
Re: source needed for very low ripple isolated DC-DC converters
« Reply #13 on: June 04, 2023, 11:06:42 am »
Without advanced planar transformers with the winding matching/shielding technique, the common mode effects of flybacks are exactly the problem.

Forward is quieter with less effort, with Analog devices having some slew rate limited devices ready to go ... for a price. LT1533 seems a better matched IC than LT1683 though in retrospect.
 

Offline Faringdon

  • Super Contributor
  • ***
  • Posts: 1944
  • Country: gb
Re: source needed for very low ripple isolated DC-DC converters
« Reply #14 on: June 04, 2023, 11:15:59 am »
Quote
Without advanced planar transformers with the winding matching/shielding technique, the common mode effects of flybacks are exactly the problem.
Thanks, now it is getting interesting for us all.....

Two transistor Forward generally is indeed quieter than flyback, since the drain node transition is generally less in voltage...but we can remember that its "dv/dt" that actually  causes the problem, so, as you well know,  forwards are not "necessarily" quiter than flybacks....."a flyback with a decent snubber, etc etc...."...also, flybacks that turn on in the "valley" are quieter than other flybacks, sometimes can even  be arranged to be quieter than some forwards.

And as you know, even if we have noise mitigating circuitry in with a flyback, the flyback still has the advantage of only needing one primary fet  and one secondary diode.....and no high side drive...so the overall situation is that a flyback can end up being the easiest way to get low noise.....certainly, its often the cheapest way to get "low enough" common mode noise.

Then going on to the two switch flyback.....two primary fets but still only a single secondary diode......which can compete more directly with 2 tran forward for common mode noise situation.....As you know, with two switch flyback, the fet vds  does not go above the  input rail....as with the 2 switch forward.

A one transistor forward can be worse than a flyback for common mode noise, because its drain voltage shoots up to "2 x vin" when the fet switches off.......considering a reset winding equal to NP.

....then we move on to LLC.....resonant.....but not necessarily super quiet....if the fet cds capacitances are low, then the switching node will have super fast dv/dt between the rails...so even that can be "common mode noisy"....but with LLC, as you know, people can add capacitance across the fets to reduce dv/dt.....and then just have more dead time....having said that, few people do this.  With LLC, if there is much magnetising current, ayk, then that will slew the switching node quicker, and (potentially) create more common mode  noise...unless mitigated.

High power, high voltage  PSFB's are known for a "modified version", whereby they have capacitance added across the FETs, so as to reduce the dv/dt of the FET off-going....and thence reduce common mode noise.

...This thread (eventuially) gets on to talk about adding caps  across the fets of a PSFB...
https://www.edaboard.com/threads/phase-shift-full-bridge-smps-is-massively-over-hyped.342663/

And of course, we all know that more common mode noise "generally" means more radiated noise....whereas more diff mode noise, doesnt always mean more radiated noise.

Anyway, apologies to OP.....please could you state what is it about the load that needs very low vout ripple? (if it does need that,  please excuse my assumption)
« Last Edit: June 04, 2023, 11:37:27 am by Faringdon »
'Perfection' is the enemy of 'perfectly satisfactory'
 

Offline mtwieg

  • Regular Contributor
  • *
  • Posts: 95
  • Country: us
Re: source needed for very low ripple isolated DC-DC converters
« Reply #15 on: June 12, 2023, 12:30:16 pm »
I am working on a fast turn around project in which I need very low noise PCB mounted isolated 3.3V output DC to DC converters.  I need sub mV output ripple.  I can use multiple converters for lower noise.  Multiple converters with output of 200mA-1.1A or a single converter with an output of 650mA to 3.2A.   Based on DRE I choose an input voltage of 24V, but I am not married to that.  I only need 40V maximum of isolation.  Low output noise is my primary concern.

Linear regulators do not do a great job on noise in the 100 KHz to 100s of Mhz that comes out of the converters I have looked at.

I was hoping to buy something off-the-shelf since I have many design challenges and I do not want to spend time designing filters for the converters.  It seems I may have no choice but to devote time to filter design.

I have talked with the technical reps from several companies, but they just tell be I will need to design my own filters and refer me to their application notes.

Do you know of any companies that I should contact?

Thank you
Chris
Maybe a bit late, but you requirements aren't clear. "sub mV output ripple" measured how? Peak to peak, or RMS? Then you refer to noise from LDOs in the 100kHz-100MHz bands, unclear how that relates to your requirements. Also if it's isolated then your conducted emissions will also have a common mode component, I'm assuming there's some requirement there. And that's not even getting into radiated emissions...

Even with all those requirements described coherently, it's unlikely that COTS products will have specifications which translate directly to your requirements. For example they may state that they pass specific tests for conducted/radiated emissions, but usually don't share anything beyond the pass/fail result.

Some will specify a maximum output ripple voltage (AFAIK there's no standards for output ripple) but I doubt you'll find anything specified at <1mVpp.

Basically there's no avoiding having to do some manual verification on things like this, whether you go fully custom or COTS.

In my experience, clients who want a "low noise" power supply don't actually know what they need, and give vague requirements. My approach is to suggest some requirements which can be verified, with existing standards if possible. Agree on that and put it in the list of deliverables in the contract. Then when you deliver, and they say "it's too noisy!!" you point to the contract and ask which requirement isn't being met.
« Last Edit: June 12, 2023, 12:35:41 pm by mtwieg »
 
The following users thanked this post: Wolfram


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf