Author Topic: My PSU design ripple and noise with picture measurements  (Read 41077 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 #275 on: August 15, 2023, 09:11:31 pm »
By "pulsating load" you mean what exactly? the 50-60 hz stuff is the console as mentioned.
Yes, the pulsating load is your console.

Ok, but you saw the pictures right?

I mean, I tested it on the console with different capacitances and this is the result which is about 6600u is more than enough.

the original power supply (schematic here) shows similar result but higher amplitude. Meaning no noise or spikes...etc which means the setup or power grid is not an issue.

However, original one for 3.3v rail has 1800u + 1000u + 470u + 6.8uH in between for PAL PSU, linked schematic is for NTSC PSU. this is a bit different than mine since I used 1uH inductors and higher frequency.

so the key was higher bulk caps since the pulsating load is very low frequency. Previously, it had another problem of high frequency noise and common-mode noise, this got fixed after using very good power module AC-DC with very good CMC + nice layout.

I have made the changes to add 2x 3300u caps + 2x470u poly to 3.3v which should be a lot better than currently added bulk caps (crappy chinese + ugly solder job).

Such very low ESR bulk caps should give very fast response... I am hoping that it could enhance it even further, we'll see.


3d image is attached, those are very huge caps for sure xD. 16mm diameter.



as for further reading, yes will do it in future. but i feel this psu is already very good in terms of noise and ripple don't you think so? further readings will be for my own development and knowledge.

Offline Vovk_Z

  • Super Contributor
  • ***
  • Posts: 1449
  • Country: ua
Re: My PSU design ripple and noise with picture measurements
« Reply #276 on: August 16, 2023, 11:40:25 am »
I see your 3300 uF caps are much larger then 2200 uF (16 mm vs 10 mm?) So, then 3x2200 uF may have sense. I have used 2200 uF x 6.3 V (for 3V3 rail) and they have 10 mm diameter. That's much smaller then 2 x 16 mm 3300 uF caps.
 

Offline Vovk_Z

  • Super Contributor
  • ***
  • Posts: 1449
  • Country: ua
Re: My PSU design ripple and noise with picture measurements
« Reply #277 on: August 16, 2023, 11:44:15 am »
but i feel this psu is already very good in terms of noise and ripple don't you think so? further readings will be for my own development and knowledge.
If it is now better than original then ok. With proper knowledge taken from those books you may do even better.  :-+
 

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 #278 on: August 16, 2023, 04:49:35 pm »
I see your 3300 uF caps are much larger then 2200 uF (16 mm vs 10 mm?) So, then 3x2200 uF may have sense. I have used 2200 uF x 6.3 V (for 3V3 rail) and they have 10 mm diameter. That's much smaller then 2 x 16 mm 3300 uF caps.


after reading a lot and changing my mind a lot, I came to love Samyoung NXH series of elec. caps. they do have 2200u ones but they are 12.5mm, having 3 of them costs more than double the 2 of 3300u. 3300u now fits ok as seen. LCSC does not have much of this series unfortunately.

I do have 2x470u polymer ones which will assist a lot, so all and all it should be better than I got now.

Quote
If it is now better than original then ok. With proper knowledge taken from those books you may do even better.  :-+

I will continue to learn for sure. the noise book from 1988 seem nice but old, surprise to see nothing new written on the subject.

Online T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 22001
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: My PSU design ripple and noise with picture measurements
« Reply #279 on: August 16, 2023, 05:23:30 pm »
I will continue to learn for sure. the noise book from 1988 seem nice but old, surprise to see nothing new written on the subject.

Age means nothing -- the theory of electromagnetism was well developed within a handful of decades of its discovery; the only thing that has changed is the ease with which, and the means by which, we are able to interact with the EM field.

There is very little in this book that wouldn't be familiar to, say, the eggheads at Bell Labs wholly a century ago (Zobel, Campbell, et al.), or to even more rarefied groups almost as long again prior -- physicists, or more generally "natural philosophers" as it was at the time: Hertz, Helmholtz, Maxwell, etc..  Even Faraday, despite his reluctance (or perhaps ignorance) for mathematics, probably would've understood much of the concept, purpose and method of the book, perhaps inbetween marveling at the depth of modern technology of course (that we have to deal with waves on a regular basis, not just the E and M separately, that concerned his experiments; not to mention the ease with which we generate said waves).

In other words -- the enduring value of books like this, are the fields (literally, in this case) they deal with.  EM itself does not change, only the technology we use to interact with it; and our analytical methods to approach it.

If you compare to a book like Terman's Radio Engineer's Handbook, or F. Langford-Smith's Radiotron Designer's Handbook, the mathematical approximations used were much cruder -- the engineer might be working on pencil and paper, slide rule in hand -- and plotted data (nomographs) and empirical measurements were the tools of choice when blindingly-simple equations didn't suffice.  Whereas today, not only are more complicated formulas feasible (consider the formulas in common circulation for microstrip impedance, for example), but algorithms entirely, since everyone and their dog has a computer available at all times.

Which are still excellent references, by the way.  The simplicity with which they approach problems is refreshing, and the results are no less effective -- we just might have tighter performance demands nowadays (squeezing out every dB of SNR and DR, while minimizing circuit size and power consumption, say), which sometimes suggest alternative solutions, but generally the conventional topologies work fine, or at the very least as starting points.

A contemporary case being the development of electric wave filters, at first single-pole (RC or LR), then k-derived (using a transformation that's easier to work in, resulting in circuits containing poles and zeroes which hampered performance but made them easier to work with), then m-derived (a more general method), and eventually analytical theory (by the... 40s or so, I forget? Crowned by Zverev's Handbook of Filter Synthesis published finally in 1967), which in sense fully and completely perfected the art.  Since then, optimization tools have addressed minor and more practical aspects of filter design: accommodating component selection by commercially available values; ensuring performance over component tolerances; and reducing engineer time by automating much of the process.  Not to mention incorporating transmission lines or full field simulation into the process, a critical step for microwave circuitry.

Or for a more direct case: not that exceptionally low power supply impedance was much of a concern back then (tube circuits are fine with some ohms), but most any of these authors would see your optimization problem, and a system diagram, and ask: why not simply increase error amp gain?  Or since you don't have control over it (internal to the IC), why not choose a regulator with higher gain?  It's surely cheaper and more compact than trying to stuff oversized capacitors into the box!

They would also have the wisdom to ask: do you really need this?  An intrepid one might even perform measurements on your load to see how much ripple it can tolerate before malfunction occurs, and set a maximum at the power supply of a modest fraction below the critical level.

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

Offline Sniper1

  • Regular Contributor
  • *
  • Posts: 81
  • Country: ro
Re: My PSU design ripple and noise with picture measurements
« Reply #280 on: August 16, 2023, 05:40:22 pm »
Maybe the ERS to to low and it is not properly compensated for modern caps .
SMPS need some compensation, well basically if the output caps are very low ESR they may respond to fast and that messes up with the FB loop and it picks random noise and amplifies it (or potentially osculates)
 

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 #281 on: August 16, 2023, 05:55:45 pm »
Maybe the ERS to to low and it is not properly compensated for modern caps .
SMPS need some compensation, well basically if the output caps are very low ESR they may respond to fast and that messes up with the FB loop and it picks random noise and amplifies it (or potentially osculates)

all of these are for 2nd stage filtering. I do have first stage filtering after switcher which has a 22u elec cap (high esr) to act as damper as agreed in previous pages.

so, having low esr at final output should not affect anything badly in terms of stability. feedback loop is taken from 1st stage not end of 2nd stage.

Quote
Or for a more direct case:

well, as you saw in waveforms everything looks fine now with 2mv p-p ripple except that 50-60hz power sips which as it seems can't be fixed. I have no problem with it also since it exists on the original power supply itself, mine is actually better. those caps now fit nicely, the system seems to need such caps due to its power sipping at low frequency, original psu also has big caps and bigger inductor as well.

Online Someone

  • Super Contributor
  • ***
  • Posts: 4685
  • Country: au
    • send complaints here
Re: My PSU design ripple and noise with picture measurements
« Reply #282 on: August 16, 2023, 10:52:15 pm »
except that 50-60hz power sips which as it seems can't be fixed.
Which is trivial to fix, and the answers have been provided several times in the thread.
 

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 #283 on: August 16, 2023, 11:39:15 pm »
except that 50-60hz power sips which as it seems can't be fixed.
Which is trivial to fix, and the answers have been provided several times in the thread.

I believe previously we mentioned that an LDO can fix this issue, but I didn't want such solution and still don't.

Besides that I assume I have followed everything else. Like, first I changed the entire design and used a complete solution with AC-DC power module + good CMC to ensure input power is good since this was an issue before or at least uncertainty.

previously it had ringing + noise, we reached a conclusion of using 22uF elec. caps on everywhere possible to dampen the system... which I did and looks like it proved its worth. Now there are zero noise spikes or ringing of any kind. especially after adding bulk capacitance on 3.3v rail. the 5v rail is fine without anything extra.

adding the elec. caps (with higher ESR) is the replacement solution of adding snubber circuits, as it does the same job. but a quick rough calculations on 3.3v rail for the new modification: it has a total of 7690uF capacitance and 2uH inductance for the total of both stages assuming no capacitance dc bias.

for that, Zo= sqrt(L/C) -> sqrt (2/7690) = 0.016R. Previously got mentioned that after calculating this, it should be added in series with one of the caps at the output. I can add it in series with an extra 22uF elec cap but since it is so low, I assumed that elec. cap ESR will do the job just fine and this was the recommended solution.

for the recent test, I put 3x2200u bulk caps which did the trick of removing the last power sip spike and now it is just regular 50-60hz consumption.

I want to emphasize that this 50-60hz thing exists in original psu itself despite being made professionally by sega themselves. do you think they could have fixed this or is this the way to go? actually, its amplitude in original psu is worse than mine.

so in your opinion, I need to do what besides all that? should I specifically add external snubber circuit using the data from the newly chosen elec. caps?



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 #284 on: September 24, 2023, 05:11:59 pm »
pics: https://imgur.com/a/Cyt5WCp

measurements: https://imgur.com/a/IbU18zA

new and final model arrived, tested to be fine and according to expectations.


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf