Author Topic: My PSU design ripple and noise with picture measurements  (Read 40889 times)

0 Members and 1 Guest are viewing this topic.

Offline VEGETATopic starter

  • Super Contributor
  • ***
  • Posts: 1992
  • Country: jo
  • I am the cult of personality
    • Thundertronics
Re: My PSU design ripple and noise with picture measurements
« Reply #200 on: November 15, 2022, 12:51:42 pm »
Quote
imo your load is DC (Digital Current), something is demanding current (or pumping it) at 50-100Hz (that mVpp square wave). have you point out what causing it?

by DC in that reply i meant dreamcast.

I pointed out that I hooked the PSU i made to the dreamcast, started the dreamcast and waited until gdemu title screen appears (or a game, same result), then took the measurements on 3.3v and 5v as mentioned above (spring ground, etc..).

what did you expect or assume?

the waveform is taken by ac coupling 1x probe setting.



Quote
thats what we call, R&D one-time investment. it will worthy to buy if you think you are going to design many other PSUs in the future

yes, i will certainly get a nice unit for this purpose but not immediately now... will need some time.



___

so my application being dreamcast is taking big current at 50-60 hz periods which caused this waveform performance on 3.3v rail... correct? how can we fix this if it is fixable?


Offline Mechatrommer

  • Super Contributor
  • ***
  • Posts: 11700
  • Country: my
  • reassessing directives...
Re: My PSU design ripple and noise with picture measurements
« Reply #201 on: November 15, 2022, 01:45:36 pm »
so my application being dreamcast is taking big current at 50-60 hz periods which caused this waveform performance on 3.3v rail... correct? how can we fix this if it is fixable?
not necessarily big.. to know if its fixable, we need to sim, to sim... we need to know...
1) average current consumption by dreamcast on both 3.3 and 5V rail
2) your smps ic average switching frequency? Ton and Toff? if you got some picture.
3) better if we have ballpark value of your PSU's output impedance, you can build a test rig as mentioned, and give it a good stress loading..

low side switching.. on your left side..


then i think we can snap some simulation.. as i remember my Snr.Eng scolded me 20yrs ago its still fresh in my mind.... nothing is impossible!... but implicitly in behind... its just a matter of cost and space ;) i can think of one "active" way to deal with this, if you really want to fight it... cheers.
Nature: Evolution and the Illusion of Randomness (Stephen L. Talbott): Its now indisputable that... organisms “expertise” contextualizes its genome, and its nonsense to say that these powers are under the control of the genome being contextualized - Barbara McClintock
 

Offline VEGETATopic starter

  • Super Contributor
  • ***
  • Posts: 1992
  • Country: jo
  • I am the cult of personality
    • Thundertronics
Re: My PSU design ripple and noise with picture measurements
« Reply #202 on: November 15, 2022, 02:29:14 pm »
Quote
1) average current consumption by dreamcast on both 3.3 and 5V rail

can't be for sure since i didn't measure myself... but 3.3v is about 3 amps or a little less. someone posted it is 2.8 amps. 5v is < 0.5 amps.

Quote
2) your smps ic average switching frequency? Ton and Toff? if you got some picture.

it is fixed 700 khz.

i guess ton and toff are taken assuming input and output voltage. input voltage is 12v, output 1 = 3.3v, output 2 = 5v. now we know the duty cycle for each rail and can conclude timings.

Quote
3) better if we have ballpark value of your PSU's output impedance, you can build a test rig as mentioned, and give it a good stress loading..

i don't know how to measure it really. what test rig are you referring to?



remember that this device will be used with different types of 12v power bricks, each its own quality. what solution do you have in mind?
i don't mind little cost up if it is gonna significantly enhance the output.


Offline Mechatrommer

  • Super Contributor
  • ***
  • Posts: 11700
  • Country: my
  • reassessing directives...
Re: My PSU design ripple and noise with picture measurements
« Reply #203 on: November 15, 2022, 03:52:30 pm »
Quote
3) better if we have ballpark value of your PSU's output impedance, you can build a test rig as mentioned, and give it a good stress loading..
i don't know how to measure it really. what test rig are you referring to?
test rig picture shown above. by using voltage divider (psu internal resistance + Rload + mosfet Rds on), we can work out what is the "virtual" internal resistance.
https://www.science-campus.com/engineering/electrical/dc_theory/chapter6/dctheory_6_7.html

remember that this device will be used with different types of 12v power bricks, each its own quality. what solution do you have in mind?
i don't mind little cost up if it is gonna significantly enhance the output.
the solution to isolate input power source's noise has been proposed to you earlier by using slotted gnd plan. not much we can do if variant is the input source as different user will use different source hence different noise profile. maybe another easy fix is wind the input's wire into toroidal ferrite core/ring, ymmv.
https://my.rs-online.com/web/c/passive-components/emi-filters-protection/ferrite-rings/

other thing i noticed... analyzing your schematic and pcb, it seems you installed C37 and also maybe C38? (C22, C23 in earlier design? not sure which pcb you are testing now) are they important and recommended by manufacturer? i cant see it in tps62913 datasheet other than Cff = N/P (not populated) have you tried removing them? sometime bypass across either feedback voltage divider can make regulation worse. ymmv. reference: https://imgur.com/a/6ANuRwu
« Last Edit: November 15, 2022, 08:15:37 pm by Mechatrommer »
Nature: Evolution and the Illusion of Randomness (Stephen L. Talbott): Its now indisputable that... organisms “expertise” contextualizes its genome, and its nonsense to say that these powers are under the control of the genome being contextualized - Barbara McClintock
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21962
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: My PSU design ripple and noise with picture measurements
« Reply #204 on: November 15, 2022, 05:43:17 pm »
yes games are either ntsc or pal for it, but i never thought this would reflect back to the power supply itself! i mean as long as we provide enough capacitance and filtering, it should not dip much.

It necessarily reflects to the power supply, of course!

Indeed, if you've ever had the experience of a poor quality or failing adapter, you may've noticed this as the consumption of your PC/laptop/phone/widget varies with function.  Or, I had one occasion I was running my laptop off this shitty inverter thing I made,



The output filter inductors and/or capacitors vibrate slightly when switching (it's a "modified sine" output), and more when under load.  So the laptop's variable load is almost immediately (within 10ms or so) reflected in the sound of the inverter.  It's audible, responding faster than Task Manager's CPU gauge does.

Indeed, a power converter is a lowpass filter of sorts; it might add some noise, too, but to the extent we can null that out or otherwise ignore it, its behavior is dominated by the energy storage components that make it up.  So, input and output filters, and whatever the conversion mechanism is (usually a switched inductor, but sometimes capacitors are used, or a combination thereof).

An ideal power converter, I suppose you could say, stores no energy and transforms input to output (or vice versa) instantaneously.  Converters have certainly improved along such a trajectory, over the years: going from low frequency switchers and huge inductors (low ripple fraction), to higher ripple fractions (peak current mode, resonant) and higher frequencies.

Note that ripple fraction basically says how much energy is stored, or how many cycles it takes to build up / release the full energy of that element.  A ripple fraction of 30% means that about 30% of the energy is circulated each cycle, so takes about 1/x or 3 times that (3 cycles) to fully dis/charge that energy.

And doing that at higher frequencies, then, simply goes faster.  Your regulator might be around this level.  So, at 700kHz, that's a power bandwidth, so to speak, of ~230kHz.  The input and output filters cut generally lower than this, but this would be the limiting case, if you for some reason needed fastest response from load to source.

Some converters intentionally store energy, for safe keeping (hold-up time), or because the source itself is intermittent (single phase AC).  Note that 3-phase AC needs very little energy storage, because power is available continuously, and indeed many industrial converters only use modest filtering, to take advantage of this.

In any case, sooner or later, change in load must be reflected by change in draw from the source.  Even a battery isolated system must be charged sooner or later, and the amount of charge equals (well, modestly exceeds, due to losses) the load consumption over that time.

This is even relevant for SIGINT purposes.  The power consumption of a building or room can be monitored, and information can be leaked out as a result.  (A battery-isolated design might be required to help mitigate such attacks, even if power is generally available at the location!)

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline VEGETATopic starter

  • Super Contributor
  • ***
  • Posts: 1992
  • Country: jo
  • I am the cult of personality
    • Thundertronics
Re: My PSU design ripple and noise with picture measurements
« Reply #205 on: November 15, 2022, 09:27:58 pm »
well I have got some updated measurements, this time at no load to see if the spikes and ripple really exist in the power supply itself or from load.

here are the images: https://slow.pics/c/BIaB5avO

older design is the one used tps62913 and newer design using tps54394, competitor using lm2596 with 33uH inductor as main filter, 220u cap as input, 100u cap as output... very simple.

notes on new pics:

1- older design have about 3mv total p-p, no load.

2- competitor have about 2-3mv p-p which is very good. at no load of course, when loaded it makes those 50-60 hz ripple.

3- new design have good input ripple, about 7-10mv p-p but with huge spikes... but output ripple increased in magnitude. why is this?

4- there is some ringing effects as seen in images. I can hear very tiny sound from the board when my ears are like very close. could this be that switcher is not stable enough?

for your opinions.

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21962
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: My PSU design ripple and noise with picture measurements
« Reply #206 on: November 15, 2022, 10:08:37 pm »
Burst mode operation.  Looks like the filter is very poorly damped, hence the ringing.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline VEGETATopic starter

  • Super Contributor
  • ***
  • Posts: 1992
  • Country: jo
  • I am the cult of personality
    • Thundertronics
Re: My PSU design ripple and noise with picture measurements
« Reply #207 on: November 15, 2022, 10:20:33 pm »
Burst mode operation.  Looks like the filter is very poorly damped, hence the ringing.

Tim

I have followed design guidance, also it has internal compensation.

I didn't find burst mode but rather eco-mode in datasheet: https://www.ti.com/lit/ds/symlink/tps54394.pdf

it says it is for light loads, which our latest measurements are.

so what are the damping needed and where?

I have followed their datasheet and eval module precisely, especially in placement of caps and L. how could that go wrong?


Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21962
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: My PSU design ripple and noise with picture measurements
« Reply #208 on: November 16, 2022, 12:15:06 am »
I have followed their datasheet and eval module precisely, especially in placement of caps and L. how could that go wrong?

How?  Appnotes aren't generally well written, mediocrity is the norm.  Very little understanding is promoted, or used in the writing process.

In particular, take the inductance of the ferrite bead.  Resonate it with the capacitors on either side of it, in series.  Get the impedance Zo = sqrt(L/C).  Ideally, add this as a resistor in series.  But that would worsen DC performance.  So instead, put it in an R+C in parallel with one of the caps (probably the output side).  Use C > 2.5 times the total output capacitance.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline VEGETATopic starter

  • Super Contributor
  • ***
  • Posts: 1992
  • Country: jo
  • I am the cult of personality
    • Thundertronics
Re: My PSU design ripple and noise with picture measurements
« Reply #209 on: November 16, 2022, 05:25:17 am »
I have followed their datasheet and eval module precisely, especially in placement of caps and L. how could that go wrong?

How?  Appnotes aren't generally well written, mediocrity is the norm.  Very little understanding is promoted, or used in the writing process.

In particular, take the inductance of the ferrite bead.  Resonate it with the capacitors on either side of it, in series.  Get the impedance Zo = sqrt(L/C).  Ideally, add this as a resistor in series.  But that would worsen DC performance.  So instead, put it in an R+C in parallel with one of the caps (probably the output side).  Use C > 2.5 times the total output capacitance.

Tim

no app note was used, only datasheet and eval board + direct support.

so you think the output meaning stage 1 or stage 2 after ferrite? the datasheet and eval only have stage 1, I added stage 2 for better filtering.

this is the ferrite bead: https://www.fair-rite.com/product/sm-beads-differential-mode-2773037447/

I could not get the inductance value. however it does have impedance curves. judging from the 2nd graph in "electrical" tab, the impedance is a little above 0 at 3 amps. first graph shows Z=38 at 1 MHz, our 700 khz would be maybe 35 but it is outside the drawing.

anyway, stage 1 has 7x10uF = 70uF (not calculating effective one due to dc bias), 2nd stage has 5x10uF + lower value ones = 50uF.

so I need something like 150uF elec. cap with a resistor in parallel? I read that elec. cap alone can do the job since it has high esr, but here may need more. how much resistance to add?

I actually added 220u elec cap at 3.3v output directly but it didn't enhance the performance during full load... however, this may be due to the 50-60 ntsc pal thing we talked about. I didn't try it without load though. that was for stage 2. also, each rail have a load on it, maybe 1k or so in parallel with output caps.

as for stage 1 they didn't recommend elect cap to begin with and they specified a maximum capacitance of 68u... so what do you think?

so by having Zo=sqrt(L/C) for stage 1 and 2, what are we looking to do?

I have different types of resistors I can use in parallel with 220u cap, what do you suggest?

and one more thing, does this R+C for damping actually act like other caps in terms of filtering ripple or it is just for damping because of very high R?


Offline Mechatrommer

  • Super Contributor
  • ***
  • Posts: 11700
  • Country: my
  • reassessing directives...
Re: My PSU design ripple and noise with picture measurements
« Reply #210 on: November 16, 2022, 05:43:14 am »
How about another revision? Trash all those redundant smd caps thats taking up spaces, smps only act as preregulator, nevermind worse performance.. put linear ldo regulator 3.3v and 5v at the front end and let the psrr do its magic, so if you are still not satisfy, nobody here will be able to help you anymore >:D
Nature: Evolution and the Illusion of Randomness (Stephen L. Talbott): Its now indisputable that... organisms “expertise” contextualizes its genome, and its nonsense to say that these powers are under the control of the genome being contextualized - Barbara McClintock
 

Offline VEGETATopic starter

  • Super Contributor
  • ***
  • Posts: 1992
  • Country: jo
  • I am the cult of personality
    • Thundertronics
Re: My PSU design ripple and noise with picture measurements
« Reply #211 on: November 16, 2022, 06:03:04 am »
How about another revision? Trash all those redundant smd caps thats taking up spaces, smps only act as preregulator, nevermind worse performance.. put linear ldo regulator 3.3v and 5v at the front end and let the psrr do its magic, so if you are still not satisfy, nobody here will be able to help you anymore >:D

ldo will have heat issues and will require me to put heatsinks and so on which originally i didn't want.

however, having too many 10uF caps is troublesome, I can use 22u caps instead. maybe one of those higher quality low esr ones i linked before.

also, ldo psrr is usually good for lower frequencies and won't do much for high or mid frequency noise. could they enhance the low frequency ripple though? i don't know.


if we are going this route, then we would need like really low drop voltage... mabye 100mv? i bet the ldos which could do that are very expensive xD. however at 100mv dropout for 3.3v at 3 amps = 3x100 = 300mW dissipation. it won't need a heatsink. having to install heatsinks are kinda no-go for me.

i am easily satisfied.. xD i just need that nice juicy <10mv p-p full load  :popcorn:


lol i found lt3033 which can do 100mv dropout but it is 9$ per one IC hhh.

Offline Mechatrommer

  • Super Contributor
  • ***
  • Posts: 11700
  • Country: my
  • reassessing directives...
Re: My PSU design ripple and noise with picture measurements
« Reply #212 on: November 16, 2022, 06:52:54 am »
Look again, your problem is 50hz, far from hi freq region.. its some low freq loading that either feedback control unable to compensate, or you have significant resistance in your output tracks.. increase your track width.. and add filter/capacitance.. very near to power connector.. thats what it takes if you want to improve things... more space, more components, more cost, more heatsink... thats engineering  >:D.. for 0.3w you can let the pcb spread the heat ymmv..
Nature: Evolution and the Illusion of Randomness (Stephen L. Talbott): Its now indisputable that... organisms “expertise” contextualizes its genome, and its nonsense to say that these powers are under the control of the genome being contextualized - Barbara McClintock
 

Offline VEGETATopic starter

  • Super Contributor
  • ***
  • Posts: 1992
  • Country: jo
  • I am the cult of personality
    • Thundertronics
Re: My PSU design ripple and noise with picture measurements
« Reply #213 on: November 16, 2022, 07:09:26 am »
Look again, your problem is 50hz, far from hi freq region.. its some low freq loading that either feedback control unable to compensate, or you have significant resistance in your output tracks.. increase your track width.. and add filter/capacitance.. very near to power connector.. thats what it takes if you want to improve things... more space, more components, more cost, more heatsink... thats engineering  >:D.. for 0.3w you can let the pcb spread the heat ymmv..

well, lt3033 won't need any heatsink but it is massively expensive. more realistic LDOs would be ones reaching 1-1.5w dropout dissipation, i guess heatsink needed for that right? gonna be very hot.

look at my tracks in some of the layout images above, i actually used zone copper filling so that it is way wider than tracks.

so the ldo can actually attenuate the 50-60hz ntsc pal power sipping pattern? i don't know about that. problem is that the smps ic here still even at no load has some ripple and noise which is not 50-60hz. maybe i should be trying a load resistor to see the actual performance of that ic since no or light load is using eco-mode.

hmm it is getting a bit more complicated and way expensive that i needed...  :--


I mean see that competitor I spoke about, using the very cheap switcher IC LM2596 operating at 150 khz... at no load it achieves very small ripple and noise <5mv but at full load still has that pattern. could this actually be the better alternative?? it uses very large 33u inductor as main inductor while my ic uses 4.7 max since 700khz.

Offline jonpaul

  • Super Contributor
  • ***
  • Posts: 3492
  • Country: fr
Re: My PSU design ripple and noise with picture measurements
« Reply #214 on: November 16, 2022, 07:34:51 am »
Bonjour, after 9 pages of comments, perhaps 100 messages, looking at the breadboard photo
I think the OP needs more research and experience to do this PSU.

The breadboard technique is not useable to debug these issues.

Jon
Jean-Paul  the Internet Dinosaur
 
The following users thanked this post: Someone

Offline VEGETATopic starter

  • Super Contributor
  • ***
  • Posts: 1992
  • Country: jo
  • I am the cult of personality
    • Thundertronics
Re: My PSU design ripple and noise with picture measurements
« Reply #215 on: November 16, 2022, 07:40:05 am »
Bonjour, after 9 pages of comments, perhaps 100 messages, looking at the breadboard photo
I think the OP needs more research and experience to do this PSU.

The breadboard technique is not useable to debug these issues.

Jon

what breadboard? i never used any breadboard in this project so far. all manufactured PCBs. are you referring to probing technique? still, i didn't post an image for it i think.

thanks

Offline Mechatrommer

  • Super Contributor
  • ***
  • Posts: 11700
  • Country: my
  • reassessing directives...
Re: My PSU design ripple and noise with picture measurements
« Reply #216 on: November 16, 2022, 08:09:00 am »
I mean see that competitor I spoke about, using the very cheap switcher IC LM2596 operating at 150 khz... at no load it achieves very small ripple and noise <5mv but at full load still has that pattern. could this actually be the better alternative?? it uses very large 33u inductor as main inductor while my ic uses 4.7 max since 700khz.
thats a strategy right? Start cheap, optimize later. if he revise and add post-reg, what hope do you have? Probably still cheaper or same cost as your latest version. Didnt we mentioned try larger L? dont make my mistake of following recommended L.. we'll never know..
Nature: Evolution and the Illusion of Randomness (Stephen L. Talbott): Its now indisputable that... organisms “expertise” contextualizes its genome, and its nonsense to say that these powers are under the control of the genome being contextualized - Barbara McClintock
 

Offline jonpaul

  • Super Contributor
  • ***
  • Posts: 3492
  • Country: fr
Re: My PSU design ripple and noise with picture measurements
« Reply #217 on: November 16, 2022, 09:09:58 am »
Mistook photo in the post of T3sl4co1
Sorry.

J
Jean-Paul  the Internet Dinosaur
 

Offline VEGETATopic starter

  • Super Contributor
  • ***
  • Posts: 1992
  • Country: jo
  • I am the cult of personality
    • Thundertronics
Re: My PSU design ripple and noise with picture measurements
« Reply #218 on: November 16, 2022, 03:29:27 pm »
I mean see that competitor I spoke about, using the very cheap switcher IC LM2596 operating at 150 khz... at no load it achieves very small ripple and noise <5mv but at full load still has that pattern. could this actually be the better alternative?? it uses very large 33u inductor as main inductor while my ic uses 4.7 max since 700khz.
thats a strategy right? Start cheap, optimize later. if he revise and add post-reg, what hope do you have? Probably still cheaper or same cost as your latest version. Didnt we mentioned try larger L? dont make my mistake of following recommended L.. we'll never know..

hmmm I always get warned about using more than recommended inductor size, it happened twice. so my 4.7u is the upper limit, i can now stack another one on top of it to be 2x 4.7u in parallel for the sake of testing. do you think it might work?

what is you test about this? you mentioned you had a mistake following recommended L.

hmmm... I would stick to pure switching design for now... really don't want to have extremely costly device as well as more complicated. I found LT3033 which can work for 3.3v and ADM7171 (ADM7171ACPZ-5.0-R7) for 5v. both can handle about 100mv dropout which is fantastic without heating. 5v rail can tolerate more since it has less current demand. However, these are not cheap. taking also that we need a switcher or 2 switchers... the problem is that I want this device to be sold at 40$ retail price, I can bump it up to 50$ at absolute most if it is really lower noise and ripple but choosing such high cost stuff will make that near impossible. There is TPS7A53 too but it also will increase cost.+ 91 weeks lead time xD

another approach is the cheaper LM1085 (3A) and LM1086 (1.5A) regulator but this would require a heatsink for sure, which makes it kinda worse that original 20-year old PSU which has a lot of heat.


Offline Mechatrommer

  • Super Contributor
  • ***
  • Posts: 11700
  • Country: my
  • reassessing directives...
Re: My PSU design ripple and noise with picture measurements
« Reply #219 on: November 16, 2022, 04:06:47 pm »
what is you test about this? you mentioned you had a mistake following recommended L.
the recommended L is 22uH, big spikes on output, changed to 470uH, spikes went down almost 100%. it wont hurt trying 10 minutes stacking them together in parallel to get 9.4uH, if it doesnt make any different call it failed and remove back to original 4.7uH

hmmm... I would stick to pure switching design for now...
sometime we have to pay the price to get better performance. after the other cheaper solutions failed. ymmv.
Nature: Evolution and the Illusion of Randomness (Stephen L. Talbott): Its now indisputable that... organisms “expertise” contextualizes its genome, and its nonsense to say that these powers are under the control of the genome being contextualized - Barbara McClintock
 

Offline VEGETATopic starter

  • Super Contributor
  • ***
  • Posts: 1992
  • Country: jo
  • I am the cult of personality
    • Thundertronics
Re: My PSU design ripple and noise with picture measurements
« Reply #220 on: November 16, 2022, 07:29:16 pm »
Quote
the recommended L is 22uH, big spikes on output, changed to 470uH, spikes went down almost 100%. it wont hurt trying 10 minutes stacking them together in parallel to get 9.4uH, if it doesnt make any different call it failed and remove back to original 4.7uH

oh so they told you 22u is the recommended, what was the upper maximum they told you about? but 470u is too much compared to 22... assuming modern switcher which has fast switching like 1-2 mhz, having smaller L is doable but it would not hurt to use very large L right?

what mV p-p you had before and after? do you have some waveforms?

Quote
sometime we have to pay the price to get better performance. after the other cheaper solutions failed. ymmv.

you are correct but also there are limits to how much you can increase price... the other cheaper solutions are working fine especially the not-so-crappy ones... I just wanted like a "premium" version which has truely better performance... so from 38$ for the best one to 50$ for my solution seem nice... just not much more.

testing L is one last hope I can cling to. but i try with others here to have another solution using the same parts if possible

Offline Mechatrommer

  • Super Contributor
  • ***
  • Posts: 11700
  • Country: my
  • reassessing directives...
Re: My PSU design ripple and noise with picture measurements
« Reply #221 on: November 16, 2022, 08:16:28 pm »
what mV p-p you had before and after? do you have some waveforms?
you can look my psu here https://www.eevblog.com/forum/projects/voltage-spikes-on-the-linear-output-in-split-rail-smps-pre-regulator/msg4495825/#msg4495825 you can read my tuning/debug process if you are interested. but different requirement as yours, i will be happy if it can do 100mVpp. and i start with the cheapest jelly bean 34063 so i dont expect too much when starting this project knowing people like to bash 34063... this is for personal use, and as an exercise and learning too. i didnt report my last result (by increasing L from 22 to 470uH) which is somewhere 60-100mVpp ripples/spikes on the linear post-regulator output (still nasty ripples on smps pre-regulator, dont ask!). in the mean time i will wait to send the revised pcb ver4 to jlcpcb... i'm doing other project now.
Nature: Evolution and the Illusion of Randomness (Stephen L. Talbott): Its now indisputable that... organisms “expertise” contextualizes its genome, and its nonsense to say that these powers are under the control of the genome being contextualized - Barbara McClintock
 

Offline VEGETATopic starter

  • Super Contributor
  • ***
  • Posts: 1992
  • Country: jo
  • I am the cult of personality
    • Thundertronics
Re: My PSU design ripple and noise with picture measurements
« Reply #222 on: November 16, 2022, 08:30:07 pm »
I tried putting 2 10uH inductors on top of the 4.7u one so now they are all in parallel not in series... so 10 | 10 | 4.7 not a full 26.7u. the result didn't really change. so i think maybe series is the way to go?

i am really confused to what i should do now...

Offline Mechatrommer

  • Super Contributor
  • ***
  • Posts: 11700
  • Country: my
  • reassessing directives...
Re: My PSU design ripple and noise with picture measurements
« Reply #223 on: November 16, 2022, 08:57:41 pm »
yup inductors should be in series to increase inductance, not parallel. i was having a brainfart moment :palm: you should be able to google this easily https://www.allaboutcircuits.com/textbook/direct-current/chpt-15/series-and-parallel-inductors/ so then you have to do dead bug style inductors on your shiny pcb ;D
Nature: Evolution and the Illusion of Randomness (Stephen L. Talbott): Its now indisputable that... organisms “expertise” contextualizes its genome, and its nonsense to say that these powers are under the control of the genome being contextualized - Barbara McClintock
 

Offline VEGETATopic starter

  • Super Contributor
  • ***
  • Posts: 1992
  • Country: jo
  • I am the cult of personality
    • Thundertronics
Re: My PSU design ripple and noise with picture measurements
« Reply #224 on: November 16, 2022, 09:21:26 pm »
yup inductors should be in series to increase inductance, not parallel. i was having a brainfart moment :palm: you should be able to google this easily https://www.allaboutcircuits.com/textbook/direct-current/chpt-15/series-and-parallel-inductors/ so then you have to do dead bug style inductors on your shiny pcb ;D

here is what i did: https://imgur.com/a/UD88Rdq

those are in parallel... I remember reading that in parallel does provide benefit of sharing the load for sure, but in our case the load is fine for one inductor. it is just the ripple performance that we need.

i will have to remove the 4.7uH inductor and replace it with 10uH one... it should provide some visible benefit being more than 2 times bigger. if so, then it makes sense to install a bigger one like 100uH or so but I actually don't have one. maybe salvage from other boards that have 33u


in your design you seem to use a very generic and low quality IC, having 60mv could be a good result.


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf