Hello,
I needed to make a power supply using a switching regulator to drop voltage without wasting a lot of power, but I needed a extremely quiet supply.
The lt3042 voltage regulator is good, but the switching regulators edges were getting through.
I went on a development of useful filters to put between them.
I had forgotten that almost all ferrite beads are made for signals with no current and saturate their magnetic material with just a few mA's and turn into just a short bit of wire. It's a major pain in the *** to search for ferrite beads useful for supplying power to opamps.
They typically only have values at 100MHz and have no information for saturation behavior.
I looked at surface mount inductors, but their inter-winding capacitance let the switch edges through.
The laird 35F0121-1SR-10 was a lone surprise after a lot searching.
Says it's got a 10amp limit, but that is an overheating limit. Looks good to 100mA's.
Even has a good spice model.
I was able to make an astoundingly quiet supply.
To see the microvolts of noise I had to use multiple lt1028's to get a gain of 100,000 to see what was left.
The spectrum analyzer clearly saw fundamental and harmonics coming out of the switching regulator.
One bead ( and some caps) cut it down to just a little of the fundamental and no harmonics.
Over a thousand fold lower. After a second there was no trace of anything.
The output of the ultralow noise voltage regulator was noisier. And it's rated at a couple of micro volts.
BTW, the new acrylic film SMT caps go up to 10uF and seem extremely well behaved.
Hope this is useful.
I needed to make a power supply using a switching regulator to drop voltage without wasting a lot of power, but I needed a extremely quiet supply.
The lt3042 voltage regulator is good, but the switching regulators edges were getting through.
I went on a development of useful filters to put between them.
I had forgotten that almost all ferrite beads are made for signals with no current and saturate their magnetic material with just a few mA's and turn into just a short bit of wire. It's a major pain in the *** to search for ferrite beads useful for supplying power to opamps.
They typically only have values at 100MHz and have no information for saturation behavior.
I looked at surface mount inductors, but their inter-winding capacitance let the switch edges through.
The laird 35F0121-1SR-10 was a lone surprise after a lot searching.
Says it's got a 10amp limit, but that is an overheating limit. Looks good to 100mA's.
Even has a good spice model.
I was able to make an astoundingly quiet supply.
To see the microvolts of noise I had to use multiple lt1028's to get a gain of 100,000 to see what was left.
The spectrum analyzer clearly saw fundamental and harmonics coming out of the switching regulator.
One bead ( and some caps) cut it down to just a little of the fundamental and no harmonics.
Over a thousand fold lower. After a second there was no trace of anything.
The output of the ultralow noise voltage regulator was noisier. And it's rated at a couple of micro volts.
BTW, the new acrylic film SMT caps go up to 10uF and seem extremely well behaved.
Hope this is useful.
If both the 'send' and 'return' currents flow thru the ferrite, then it won't saturate.
There are about 6 common types/mixes of ferrites, some work at 1MHz and others at 100MHz.
Some of thes Jim Brown papers are an ferrites.
Audio Systems Group, Inc. Publications
* * * * * * *
and a chart:
https://static.dxengineering.com/global/images/chartsguides/d/dxe-acdcgen1-rfi_hc.pdf
the import part is at the bottom of the page about the mix numbers.
There are about 6 common types/mixes of ferrites, some work at 1MHz and others at 100MHz.
Some of thes Jim Brown papers are an ferrites.
Audio Systems Group, Inc. Publications
* * * * * * *
and a chart:
https://static.dxengineering.com/global/images/chartsguides/d/dxe-acdcgen1-rfi_hc.pdf
the import part is at the bottom of the page about the mix numbers.
Last edited:
Like SS above said, try using a common mode core design. The next best thing IMO, use an additional DM bead on the return leg. This technique lifts the common regulators ( input) noisy ground from the output or SA measurement ground.
Any series impedance will offer massive improvement over a trace or wire. heck a piece of chewing gum on bailing wire might make all the difference. The trick is the measurement across the smallish ceramic cap. `~ see it is simply a voltage divider with a good HF output cap !!
EDIT>
more parts here try Fair-Rite type 73 materiel the best of these are lossy ferrite mixtures.
Last edited:
Jim Williams' application notes for LT are a goldmine of helpful hints to achieve lowest noise. AN-70 is one and more specifically for this case, AN-101 : http://www.analog.com/media/en/technical-documentation/application-notes/an101f.pdf
Common mode chokes work like that but that is most often not what is needed.
They stop a signal (noise or otherwise) that is on both the signal/power and rtn/grnd.
In the case of a switching power supply it is not very workable or effective.
The noise is differential and not common mode, so the two wire choke is not the answer.
I do sometime put a bead in the power/signal line and the rtn/grnd line but that is to block noise from multiple sources.
Take a better look at the beads you reference. I have been looking at this stuff for decades and I find few beads that even mention saturation points. Most only specify performance at 100 MHz and are not lossy below 10 MHz.
Let me know if you find anything that actually has specifications that compare.
I would love to add another part as useful to my toolbox.
You might be interested in this on when to use a common mode choke and when to use beads.
murata.com/~/media/webrenewal/products/emc/emifil/knowhow/26to30.ashx
They stop a signal (noise or otherwise) that is on both the signal/power and rtn/grnd.
In the case of a switching power supply it is not very workable or effective.
The noise is differential and not common mode, so the two wire choke is not the answer.
I do sometime put a bead in the power/signal line and the rtn/grnd line but that is to block noise from multiple sources.
Take a better look at the beads you reference. I have been looking at this stuff for decades and I find few beads that even mention saturation points. Most only specify performance at 100 MHz and are not lossy below 10 MHz.
Let me know if you find anything that actually has specifications that compare.
I would love to add another part as useful to my toolbox.
You might be interested in this on when to use a common mode choke and when to use beads.
murata.com/~/media/webrenewal/products/emc/emifil/knowhow/26to30.ashx
The Murata BLM21PG range is meant specifically for high currents, but what happens to inductance is indeed not specified:
http://www.farnell.com/datasheets/1...79.290923316.1540014843-1597443229.1494277915
Q is about 2 at 10 MHz.
http://www.farnell.com/datasheets/1...79.290923316.1540014843-1597443229.1494277915
Q is about 2 at 10 MHz.
The majority of RFI / EMI beads have soft 'saturation' using sort of a distributed "air" gap or additional RF 'lossy stuff' in the mix , so they only specify Zmax . , There are no other discrete points to spec... The link I provided has a white paper about this, did you see it? (AFAIK Fair-Rite Inc.wrote the book on these type/s of small beads) The folks that design SMPS know a thing or two about filter inductors using powered iron cores on high DC. So looking at the B/H characteristics, one could argue they are even lossy approaching DC.
Last edited:
Which white paper are you referring to? I didn't see any on that webpage, but links to technical papers have such but none I found applicable or referring to type 73 material.
If you do a search on 35F0121-1SR-10.pdf you will find one of the only data sheets I found that actually covers what happens as dc current rises in a bead. It dies fast. And this is the best I found. That's why I started this post. It really is useful.
We have a power supply group. They are damn good. But their concerns are not similar to audio needs. What I had to do is very applicable. And the results are very useful if you want to use a switching power supply and make sure the noise generated does not cause primary or secondary problems.
I am reasonably sure of that because I measured with both a spectrum analyzer for the smps outputs and low noise opamps being needed for an oscilloscope to see the remaining noise level (that I needed to know).
If you do a search on 35F0121-1SR-10.pdf you will find one of the only data sheets I found that actually covers what happens as dc current rises in a bead. It dies fast. And this is the best I found. That's why I started this post. It really is useful.
We have a power supply group. They are damn good. But their concerns are not similar to audio needs. What I had to do is very applicable. And the results are very useful if you want to use a switching power supply and make sure the noise generated does not cause primary or secondary problems.
I am reasonably sure of that because I measured with both a spectrum analyzer for the smps outputs and low noise opamps being needed for an oscilloscope to see the remaining noise level (that I needed to know).
I will nearly always find benefits from using a ferrite on incoming ac, ac secondaries, on interconnects in front of preamplifier, amplifiers. Also on digital circuits to keep hf noise from propagating.
The ac usually has several turns, accompanied by a film X-cap, while the signal lines are a much smaller core, single pass-through. Always careful to mess around with the signal vs the power supply, where things are much more forgiving.
The ac usually has several turns, accompanied by a film X-cap, while the signal lines are a much smaller core, single pass-through. Always careful to mess around with the signal vs the power supply, where things are much more forgiving.
Gotcha. That has about the same response to dc current as the one I used.
But, it seems intended for frequencies around 100MHz.
Doesn't even show values below 10MHz. Might not do much to block switch mode supplies fundamental frequency.
The 35F0121-1SR-10 inductance peaks about 3MHz and the resisitive impedance peaks about 20MHz and stays useful to past 1GHz.
Just really good for switch mode power supply use for audio.
Not something I expected to find.
But, it seems intended for frequencies around 100MHz.
Doesn't even show values below 10MHz. Might not do much to block switch mode supplies fundamental frequency.
The 35F0121-1SR-10 inductance peaks about 3MHz and the resisitive impedance peaks about 20MHz and stays useful to past 1GHz.
Just really good for switch mode power supply use for audio.
Not something I expected to find.
yeah from DC to light lol
it's the same dang 'jelly bean' part there's no majick here
don't rely on dem catalogs so much
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.