Lelon electrolytic capacitors quality

[VEGETA]

Hello

I want to ask about Lelon brand as it seems to be getting more popular as a very good and very cheap solution, but is this true or not? to what degree?

anyone used these in a product that he sells? since they can be a lot cheaper than say Rubicon stuff.

this will, for sure, lead to brand names vs others elec. caps discussion but it is ok 

[TimNJ]

We don’t use them at my company. My guess is they’re probably decent for their high end series. Failures reported with these lower cost brands is probably compounded by the caps being underspec’ed to start with. If a manufacturer is really reaching for low cost, they might go for the 3,000 hour series instead of the 10,000 hour series, and voila, failure in 2 years.

[VEGETA]

We don’t use them at my company. My guess is they’re probably decent for their high end series. Failures reported with these lower cost brands is probably compounded by the caps being underspec’ed to start with. If a manufacturer is really reaching for low cost, they might go for the 3,000 hour series instead of the 10,000 hour series, and voila, failure in 2 years.

if one want to have 1000uF cap, 2 or 3 of it, in a relatively cheap product then Lelon probably good. since there is a big difference in price between it and Rubicon equivalent.

I am talking about SMD caps of course, but in through hole caps case the price difference is not so much.

so is it worth it to switch to TH caps and use rubicons or stick to smd lelon... ?

I would need some 1000uF (x2 of it) to act as pi filter (with a 33uH L) to filter out IRM-30-12 module output... then this smooth filtered output is delivered to a dual buck converter to output 3.3v and 5v. 3.3v being 3 amps while 5v < 1A.

for 3.3v and 5v rails, all caps recommended by manufacturer are ceramics. but i want to add small elec. caps to enhance ringing stuff and more stable capacitance... as well as having a 2nd LC filter stage with say 2.2uH inductor (similar to switcher one) with larger value elec caps at 220~1000u range depending on space and price.


for such PSU, should I go for Lelon SMD or Rubicon THH?

[Vovk_Z]

I would need some 1000uF (x2 of it) to act as pi filter (with a 33uH L) to filter out IRM-30-12 module output... then this smooth filtered output is delivered to a dual buck converter to output 3.3v and 5v. 3.3v being 3 amps while 5v < 1A.

1. You may encounter IRM-30-12 module instability with too large additional inductance at its output (>10 uH). Possibly, 1-10 uH may be enough.
2. It is unnecessary to smooth its output too much if buck converters are at the output. They will introduce their own ripple anyway.
3. Low ripple and noise are more about PCB design and appropriate parts but not about 'just large LC filters'.
4. Try polymer electrolytes and/or large X7R ceramics (>=1206, 1210).

[bdunham7]

for such PSU, should I go for Lelon SMD or Rubicon THH?

Is the capacitor stressed in this application (temp, ripple current)?  Are those your only two choices?

[mawyatt]

Vaguely recall Dave spotting the large capacitance TH Lelon versions in a Siglent product, if true this might hint they are acceptable.

One could setup an accelerated life test per the OEMs recommendations, usually over-voltage, current and temperature, which will still require some time to validate.

Of course getting actual feedback from a quantity user with long term history is best for a more informed decision, otherwise you'll "roll the dice" and need to wait a few years to confirm your decision

Best,   

[VEGETA]

1. You may encounter IRM-30-12 module instability with too large additional inductance at its output (>10 uH). Possibly, 1-10 uH may be enough.

Hmmm I assume it can tolerate very high capacitance (~10000u) but first time I know it cannot tolerate high inductance. anywhere to read about this?

so you suggest making 10uH or even 2.2uH inductor with 1000u caps -> bigger caps with smaller L?

2. It is unnecessary to smooth its output too much if buck converters are at the output. They will introduce their own ripple anyway.

yes, but the input ripple itself will be going through the output despite the PSRR of the dual buck. thus, smoothing as much as possible is always a plus. won't hurt, besides size and some extra cost.

3. Low ripple and noise are more about PCB design and appropriate parts but not about 'just large LC filters'.

I do have some experience making such PCB but this time I will use the AC-DC module instead of using a 12v DC wall adapter as input like previous designs.

according to available datasheet, the internal switching frequency of the module is 65khz which is too low. thus bigger caps and L can do good filtering it.

I know how to keep switching node very small, do proper grounding, keep stuff shorter.,, etc...

4. Try polymer electrolytes and/or large X7R ceramics (>=1206, 1210).

I used 22u 1206 ceramics everywhere in the design, plus the electrolytics mentioned. I got feedback that I should add smaller elec. caps in various stages of the buck design to dampen it in order to get rid of the ringing and making it more stable... 22uF was chosen. I had an idea to use small snubber of  say 10k ohm + 1uF but looks like using elec caps in each stage was a better idea.

can you recommend one polymer cap? also, what benefit over electrolytic?

Is the capacitor stressed in this application (temp, ripple current)?  Are those your only two choices?

heat is not a big issue, will be inside Dreamcast game console.

as for ripple, I explained above how it looks like from module... and buck regulator switching frequency is 700 khz. will use multiple caps in parallel and a 2nd LC stage as well. so I guess it won't be all vs one cap.

I am concerned about price as well as performance.

Vaguely recall Dave spotting the large capacitance TH Lelon versions in a Siglent product, if true this might hint they are acceptable.

One could setup an accelerated life test per the OEMs recommendations, usually over-voltage, current and temperature, which will still require some time to validate.

Of course getting actual feedback from a quantity user with long term history is best for a more informed decision, otherwise you'll "roll the dice" and need to wait a few years to confirm your decision

Best,   

yes, this is one example of products using Lelon which is why it is thought that it is a high quality brand, not regular no name Chinese stuff.

such test should be done by the likes of Dave not one like me xD.

getting user feedback is what this thread aims for, we'll wait until they spawn here xD.

[mariush]

Instead of so big electrolytics, use lower capacitance solid polymer capacitors.

For example, try going for a 470uF 16v rated polymer, or even lower ... 270uF 16v capacitors are very common as they're often used on motherboards and video cards, so their price is often much lower than others.

LCSC has Apaq 270uF/16v for 9 cents each :
https://www.lcsc.com/product-detail/Solid-Capacitors_APAQ-Tech-160AR5K271M0809C_C369918.html
https://www.lcsc.com/product-detail/Solid-Capacitors_APAQ-Tech-160AR5K271M0809B_C2919861.html

and panasonic ones are 30 cents : https://www.lcsc.com/product-detail/Solid-Capacitors_PANASONIC-16SEPC270MX_C128484.html

and lelon 470/16 are 40 cents : https://www.lcsc.com/product-detail/Solid-Capacitors_Lelon-OVZ471M1CTR-1013_C250115.html

If you don't like asian stores, tme.eu has X-con (samxon if I'm not mistaken) capacitors and kemet capacitors for under 50 cents

See link

ex 470/16: https://www.tme.eu/ro/en/details/ulr470_16/tht-polymer-capacitors/x-con/ulr477m1cg1b/

[VEGETA]

Instead of so big electrolytics, use lower capacitance solid polymer capacitors.

For example, try going for a 470uF 16v rated polymer, or even lower ... 270uF 16v capacitors are very common as they're often used on motherboards and video cards, so their price is often much lower than others.

LCSC has Apaq 270uF/16v for 9 cents each :
https://www.lcsc.com/product-detail/Solid-Capacitors_APAQ-Tech-160AR5K271M0809C_C369918.html
https://www.lcsc.com/product-detail/Solid-Capacitors_APAQ-Tech-160AR5K271M0809B_C2919861.html

and panasonic ones are 30 cents : https://www.lcsc.com/product-detail/Solid-Capacitors_PANASONIC-16SEPC270MX_C128484.html

and lelon 470/16 are 40 cents : https://www.lcsc.com/product-detail/Solid-Capacitors_Lelon-OVZ471M1CTR-1013_C250115.html

If you don't like asian stores, tme.eu has X-con (samxon if I'm not mistaken) capacitors and kemet capacitors for under 50 cents

See link

ex 470/16: https://www.tme.eu/ro/en/details/ulr470_16/tht-polymer-capacitors/x-con/ulr477m1cg1b/

well, the datasheet of evm suggests this inductor: https://www.digikey.com/en/products/detail/tdk-corporation/SPM6530T-2R2M/1993513

which is 2.2u with very high current capability and low esr of 19m, previously i used an inductor which has 2 times the esr. could this also affect ripple performance?

anyway, I found these 2 polymer caps:

SMD: https://www.lcsc.com/product-detail/Solid-Capacitors_Lelon-OVZ471M1CTR-1013_C250115.html
THH: https://www.lcsc.com/product-detail/Solid-Capacitors_Lelon-ORZ471M1CBK-0812_C311633.html -> labelled as "plugin", is this regular solderable through hole??

I would need the following:

input filtering -> one inductor (2.2uH above) + 2 470u caps
buck filtering -> one 470u cap is enough as bulk local storage + some 1206 22u + 100n ceramics very close to power pins. + 1 inductor === all x2 since it is dual buck.
buck output stage -> many ceramics + one small elec. cap of 22uF for its high value ESR for dampening.
buck 2nd stage -> ceramics + 2 of 470u caps + one 2.2u inductor.

total is: 5 inductors + 7 470u caps + 2 22u elec caps + ceramics.

for 100 boards:
inductors: 318$ ~ 3.2$ per board.
smd caps: 150$ ~ 1.5$ per board.
thh caps: 92$ ~ 0.92$ per board.

elec caps and ceramics are beyond this thread but would not cost a lot.

the design already has through hole components like plugs and sockets + IRM module itself... using such caps won't make much difference between 0.92$ and 1.5$ per board, don't you think? but it will make a lot more sense when using elec. nichicon caps where through hole is more cheaper than smd ones.

first time got recommended to use polymers instead of elec. so I got no data or previous experience.

what do you recommend?

______

on another side, IRM-30-12 module says it has internal block diagram showing input common-mode choke. do i need to put my own or just rely on it? couldn't find a data on this. would it be ok to put my own CMC on AC side or directly at it's DC side before the caps? to make sure nothing is coupled or minimize it at least.

[mariush]

Well, my question would be  WHY are you stuck on using a dual output dc-dc converter which is expensive and requires a lot of components around it? 
The chip costs $2.5 at digikey, if you buy 250 of them.

I would need some 1000uF (x2 of it) to act as pi filter (with a 33uH L) to filter out IRM-30-12 module output... then this smooth filtered output is delivered to a dual buck converter to output 3.3v and 5v. 3.3v being 3 amps while 5v < 1A.

This is your problem.

If the output of that module goes directly to the switching regulators, you don't need to filter that output so much. One good capacitor may be enough.

So there's some solutions:

solution 1. You could do two separate dc-dc converters, one that does 12v -> 5v <1A ,  one that does 12v -> 3.3v (3A+)

solution 2. You could do 12v -> 5v at more than 3A and then use a potentially cheaper dc-dc converter to do 5v -> 3.3v  OR

solution 3. You could do 12v -> 3.3v at more than 4A and then use a boost dc-dc converter to do 3.3v -> 5v as needed

solution 1 is most common.

solution 2 is not really used often. Generally a linear regulator is used after a dc-dc converter to produce a lower voltage at low current, but this is not the case for you, your voltages are the other way around

solution 3 ... not really worth it, step-up (boost) regulators are typically less efficient than buck regulators, and in your case your input voltage is low enough that it just makes more sense to use a 2nd buck (step-down) regulator to produce 5v directly from 12v.

AP62600 is under 50 cents at LCSC, and it can do 3.3v or 5v up to 6A , at up to 1.2 mHz switching frequencies : https://www.lcsc.com/product-detail/DC-DC-Converters_Diodes-Incorporated-AP62600SJ-7_C4748917.html

It's so cheap you could use two such regulators, one for 3.3v, one for 5v, and not care about optimizing and saving pennies.
You could also use it to output only 3.3v and then use a step-up regulator to produce 5v


Leave it at the default 800 kHz, leave it on USM mode(leave mode pin floating or add a jumper option to set it to pwm only or pfm if you want) The datasheet has example circuit and has recommended layout example ... it's a funky footprint that may scare you but really it's easy.

The datasheet says " An output capacitor with large capacitance and low ESR is the best option. For most applications, a 22μF to 68μF ceramic capacitor is sufficient. "

What I would probably do is have a 22uF 16v-25v ceramic in parallel with maybe a couple 10uF 25v+ ceramic, all 0805 or lower, and I'd probably add a 100-270uF 6.3v-10v solid (polymer ) capacitor as well.  As the switching frequency is so high, you don't have to go overboard with capacitance.
For inductor, follow the instructions in the datasheet, pick one with low resistance and peak current in the range specified in datasheet.

other potential chips you could use

TPS564208 : https://www.lcsc.com/product-detail/DC-DC-Converters_Texas-Instruments-TPS564208DDCR_C486062.html

 ... can do up to 4A of current, runs at 560kHz, simpler footprint, detailed datasheet, has example layout (but if you make your board 2 layers, have the inductor closer to the SW pin, the recommended layout uses internal layer for direct connection to inductor), has suggested values, formulas how to calculate if you want to .. could use same inductor value for both 5v out and 3.3v out ( 2.2..4.7uH recommended for 3.3v, 3.3..4.7uH recommended for 5v, so you could just buy 2  4.7uH inductors and be done with it.

Same deal with ceramic capacitors on output - 22 to 68uF suggested as enough, 3 ceramics in parallel would work and I'd personally add a polymer capacitor as well.


For the 5v regulator with less than 1A current, you could save a few pennies by going for a regulator "optimized" for lower currents and maybe even fixed output voltage so you don't need to have feedback resistors.

Some examples

AP63205 (fixed 5v, up to 2A, 1.1 mhz switching frequency) : https://www.lcsc.com/product-detail/DC-DC-Converters_Diodes-Incorporated-AP63205WU-7_C2071056.html

AP62200 (super cheap, 10-15 cents, max 2A, adjustable up to 7v) : https://www.lcsc.com/product-detail/DC-DC-Converters_Diodes-Incorporated-AP62200TWU-7_C2895288.html


 
For AP62600 a 50 cent 1.2uH inductor like this one would probably be fine (but double check everything) : https://www.lcsc.com/product-detail/Inductors-SMD_Murata-Electronics-LQH5BPN1R2NT0L_C1331882.html
(edit: no, the linked on is rated for only 3.1A current, recommended to be max output current + at least 35% of output current, so if you want 3A out, should be rated for at last 4A)

For the others, looking at 3.3uH - 4.7uH there's hundreds, just have to filter using the current and resistance
a random pick  that may work but again, pay attention to datasheets : https://www.lcsc.com/product-detail/Inductors-SMD_TDK-SPM6530T-4R7M-HZR_C2991837.html

[VEGETA]

Well, my question would be  WHY are you stuck on using a dual output dc-dc converter which is expensive and requires a lot of components around it? 

I was using other single output ones but ended up with this as I was choosing expensive ones which are +3$ each, since I was going low noise route. ultimately didn't achieve <10 mv p-p figure I wanted due to many stuff... one major thing being the wall adapter used and layout. first I choose TPS62913 which is very new and promises ultra low noise but has low PSRR which requires smooth input...etc.

This is your problem.

If the output of that module goes directly to the switching regulators, you don't need to filter that output so much. One good capacitor may be enough.

being a dual buck won't make difference here since ultimately smooth output is needed right?

ok, so assume small Pi filter is used, will that be ok?


please check the module test report here: https://www.meanwell-web.com/content/files/pdfs/productPdfs/MW/IRM-30/IRM-30-12-rpt.pdf

spec claims 150mV p-p which is a lot, but test result indicates mere 36mv max which is nice. datasheet spec claims this result was achieved with 47u + 0.1u caps only.

what made me want bigger caps is the internal switching frequency of 56 khz. I assume 470u + small L + 470u can do the job in this case. reducing total ripple to below 50mV is good enough since regulators will have some PSRR and output filtering will do some too. my hope was to reach 10mv p-p.

but since the system is ntsc or pal, it takes regular sips of power at 60 or 50hz which always present at power supply. even if the output of the power supply is <1mV ripple, the 50 60hz cycles are there when system powers on completely (5 seconds). this is the case of original PSU as well.

AP62600 is under 50 cents at LCSC, and it can do 3.3v or 5v up to 6A , at up to 1.2 mHz switching frequencies

yes and available at digikey, don't know if lcsc can be trusted with such ICs. I believe Diodes inc is reputable as TI, so one can trust their noise ripple figures. Quick glance in datasheet I saw it promises about 10-15 mv p-p at 6 amps which is great... with extra filtering it could be enhanced more. Layout does not seem hard either.

What I would probably do is have a 22uF 16v-25v ceramic in parallel with maybe a couple 10uF 25v+ ceramic, all 0805 or lower

why not 1206 22u in parallel with some smaller one like 1u and 100n? why 0805 or less? bigger is better for dc bias right?

For AP62600 a 50 cent 1.2uH inductor like this one would probably be fine

since I planned on having a 2nd stage LC filtering, I must know the exact inductor DC resistance to compensate for it in regulator output voltage. 19mR meaning 3.48v regulator output to achieve 3.30v at output pin @ 3 amps. if DCR does not affect ripple performance, then any typical value is ok as long as it is from good honest manufacturer where you know for sure it is correct. please help if you know about this.

assume I want 1.2 MHz, then datasheet suggests 1uH for 5v and 0.82uH for 3.3v... aren't they a bit low? maybe bumping to 2.2uH could be better?

evaluation module used these inductors for 3.3v and 5v respectively: 7443330082 and 7443330100. they have about 2.65mOhm dc resistance but very expensive!

my previous L was about half the price but still not very cheap.

I asked before about if dc resistance can actually affect performance in terms of ripple. I have no problem getting L from LCSC but is their DCR trust-able? for example this one is from TDK and also available at digikey, 4.6mDCR and relatively cheap...

having to buy 5 per board, one for IRM output pi filter, 2 for switcher and 2 for 2nd LC stage... or could remove the one for IRM and get any cheapy in its place.

___

instead of adding 1uF+10kR snubber on buck output, does my 22uF elec. cap suggestion ok? assume having the very fast 22u ceramics available in good quantity.

[mariush]

Evaluation board picks an inductor that's can handle 6A output current, you say your circuit will only go up to 3A

So they went with a 1 µH  20 A 2.65mOhm inductor, you could probably go with as little as 5-6A and it would still be fine.. lots of inductors with resistance under 10mOhm and suitable current values under 1$

You say it's gonna end up in a console in which case i seriously doubt you need ultra low output voltage ripple and noise filtering.. have you actually made tests, like for example supplying 3.3v 3A with a linear regulator from voltage coming from rectified AC voltage from a classic transformer vs switching power supply?

> why not 1206 22u in parallel with some smaller one like 1u and 100n? why 0805 or less? bigger is better for dc bias right?

bigger, higher risk of microphonics, sensitivity to vibration etc etc ... plus probably expensive unless you go for lower grades like x5r
10uF in 0805 are much more common and in higher voltage ratings - with 3.3v out, a 25v rated 10uF would still have 7+ uF of capacitance.

ex there's 285  22uF 16v or higher capacitors,  1 0603, 30 0805 , 51 1206, 102 1210 ... 
Cheapest x7r grade is 1210 16v rated at 0.17$ if you get 100 , x5r is cheaper at 0.09$ 0805 25v rated
You can get x7r grade 10uF 0805 or 1206 25v for 8 cents if you get 100 ...


i don't think a 1nF would have any point on the output of a capacitor, 100nF almost equally pointless.. on the input maybe makes sense, but 1uF would be just as good on input.

[VEGETA]

So they went with a 1 µH  20 A 2.65mOhm inductor, you could probably go with as little as 5-6A and it would still be fine.. lots of inductors with resistance under 10mOhm and suitable current values under 1$

yes lower capability is ok. can you suggest some? I will search thoroughly later on.

ou say it's gonna end up in a console in which case i seriously doubt you need ultra low output voltage ripple and noise filtering..

yes it does not need ultra low, but going as low as possible is always nice. analog video depends on lower noise but how low is enough? I don't know.

have you actually made tests, like for example supplying 3.3v 3A with a linear regulator from voltage coming from rectified AC voltage from a classic transformer vs switching power supply?

I've done a lot of tests here with multiple designs and changes but no linear PSU used (don't have one). the main point is original PSU also shows the 50 60 sips, and i have another psu which has the same output shape.

can linear regulator smooth the 50 60 hz sips? don't know but right now i am ok with having very low ripple since my PSU is the only one which has universal AC input on the market.

I've read a document saying lower DCR L is good for ripple but not sure about it.

[Vovk_Z]

1. You may encounter IRM-30-12 module instability with too large additional inductance at its output (>10 uH). Possibly, 1-10 uH may be enough.

Hmmm I assume it can tolerate very high capacitance (~10000u) but first time I know it cannot tolerate high inductance. anywhere to read about this?

I don't have a literature or an article about it. That's just a fact from my own experience with  DC/DC ICs.

22 uF wet electrolyte may help to dampen excessive oscillation. I guess you may have different wet electrolytes at your hands (10...100 uF) and test and compare performance with each value. The same with SMD ceramics size - it is very easy to test 0805 vs 1206 vs 1210, vs a bit different capacity values.

[VEGETA]

1. You may encounter IRM-30-12 module instability with too large additional inductance at its output (>10 uH). Possibly, 1-10 uH may be enough.

Hmmm I assume it can tolerate very high capacitance (~10000u) but first time I know it cannot tolerate high inductance. anywhere to read about this?

I don't have a literature or an article about it. That's just a fact from my own experience with  DC/DC ICs.

22 uF wet electrolyte may help to dampen excessive oscillation. I guess you may have different wet electrolytes at your hands (10...100 uF) and test and compare performance with each value. The same with SMD ceramics size - it is very easy to test 0805 vs 1206 vs 1210, vs a bit different capacity values.

from your experience, can you inform me on the output ripple and noise of such module? plus, do I need a common-mode choke with it or not?

Yes, I think 22uF is a nice choice here. I hope adding it to the output won't make ripple worse by being higher esr and slower to react, assuming multiple enough ceramics.

and this begs the question of do i need to use 2nd stage LC filter?

[T3sl4co1l]

Relevant context for those out of the loop --
https://www.eevblog.com/forum/projects/my-psu-design-ripple-and-noise-with-picture-measurements/

At least, I'm assuming this is the load. Sounds like it.

As I recall, you've got enough input capacitance on there already not to need more.  Ripple at the input isn't very important.  Ripple at the output (there was a 12V direct wired output, wasn't there--?) might deserve its own filter, which I recall mentioning but don't recall if it was actually added or not.

Far more important will be the mains noise, especially common mode.  Modules generally aren't very good; even medical rated ones tend to need external filtering to pass.  I have tested a few others in labs before, but I don't know about IRM-30 specifically.

Tim

[VEGETA]

bigger, higher risk of microphonics, sensitivity to vibration etc etc ... plus probably expensive unless you go for lower grades like x5r
10uF in 0805 are much more common and in higher voltage ratings - with 3.3v out, a 25v rated 10uF would still have 7+ uF of capacitance.

Although I tried such bigger 1206 x5r caps and got no buzzing or similar. going for 10uf 0805 would need lots of space. do you mean higher value makes bigger risk of microphonics or bigger size?

assume i would need 20x 22u per board and 100 boards, here are some comparisons:

1210 X7R 16v CL32B226KOJNNNE -> 2000 for 146.6$.

1206 10u 25v X7R CL31B106KAHNNNE -> 4000 for 60$. for 5000 to compensate for dc bias it is 76$.


0805 X7R 16v EMK212BB7106KG -> 4000 for 156$. assume 5000 to compensate for dc bias then it is 195.5$.

0805 X7R 16v CL21B106KOQNNNE -> 4000 for 71$, 5000 for 90$.

all of those are X7R, for X5R it is going to be a lot cheaper.

thus, maybe can use: 2x 22u 1210 + 6x 1206 10u (or 0805?) + 1x 22u electrolytic (for damping) + 1 220u or 470u polymer at the output. however, getting 220 or 470u capacitance is well above the recommended value.

i don't think a 1nF would have any point on the output of a capacitor, 100nF almost equally pointless.. on the input maybe makes sense, but 1uF would be just as good on input.

but why? input has 12v and output is 3.3v and 5v which is better in terms of dc bias.

however, are you speaking about the snubber cap with the 10k resistor or just a regular filter cap?

Relevant context for those out of the loop --
https://www.eevblog.com/forum/projects/my-psu-design-ripple-and-noise-with-picture-measurements/

At least, I'm assuming this is the load. Sounds like it.

As I recall, you've got enough input capacitance on there already not to need more.  Ripple at the input isn't very important.  Ripple at the output (there was a 12V direct wired output, wasn't there--?) might deserve its own filter, which I recall mentioning but don't recall if it was actually added or not.

Far more important will be the mains noise, especially common mode.  Modules generally aren't very good; even medical rated ones tend to need external filtering to pass.  I have tested a few others in labs before, but I don't know about IRM-30 specifically.

Tim

yes in those designs i had enough input capacitance but it didn't tame the ripple down a lot.

there is a 12v output pin as you mentioned, but that is not so important to have lower ripple than 3.3v and 5v because 12v is not for video circuitry but rather disc drive motor and similar stuff. someone suggested making a ground return path which is isolated from the rest for it... by making a no copper zone.

so you suggest adding a CMC on the AC input side, which is the case of the original PSU itself. would you suggest a good part or type?

till now we discussed capacitance types to see which are best... still no clear winner 

[T3sl4co1l]

There is no clear winner because there is no definition of application requirements (ripple, impedance, temperature, life, size..).  Once those are defined, the reason for dozens of capacitor types being available becomes more apparent.

That includes applications with very relaxed requirements say for life or ripple, which are well served by cheaper types, and often, cheaper brands.  But cheaper brands also have a range of products, some which serve very well indeed.

Similarly, without an EMI environment and setup specified, without spectra / test results, and without other application requirements (current and voltage rating, temperature, size, allowable ground leakage current, etc.), determining an EMI filter is impossible.  The best I can suggest is to construct such a setup (a LISN will be required; there are many LISN design threads on this forum) and measure it as well as you can yourself, and adjust things until it improves.  Then whatever configuration you settle on, build that on PCB and call it done.

Tim

[VEGETA]

There is no clear winner because there is no definition of application requirements (ripple, impedance, temperature, life, size..).  Once those are defined, the reason for dozens of capacitor types being available becomes more apparent.

That includes applications with very relaxed requirements say for life or ripple, which are well served by cheaper types, and often, cheaper brands.  But cheaper brands also have a range of products, some which serve very well indeed.

Similarly, without an EMI environment and setup specified, without spectra / test results, and without other application requirements (current and voltage rating, temperature, size, allowable ground leakage current, etc.), determining an EMI filter is impossible.  The best I can suggest is to construct such a setup (a LISN will be required; there are many LISN design threads on this forum) and measure it as well as you can yourself, and adjust things until it improves.  Then whatever configuration you settle on, build that on PCB and call it done.

Tim

since i chosen to add an ac-dc module, meaning the psu will work directly from mains, then i can get away without having really ultra low single digit final ripple value. since this will be a highly demanded product in the market, no other full direct ac psu available.

i settled on meanwell irm-30-12 for many reasons: cheap + has 12v output + size is ok + ripple output at least from manufacturer seems ok + can tolerate some extra output capacitance + has safety features.

poster her thankfully suggested AP62600 which I liked, cheap enough alternative to my dual buck.

will try to get cheap and good inductors to be used to filter different stages + using different types of caps as needed.

maybe i can first settle on a basic circuit like this and maybe build it and see the result.



as for LISN stuff, i still need to learn a lot about this, so prefer letting it be for future. can't find cheap lisn available too. don't have spec analyzer as well xD

[Vovk_Z]

There is no clear winner because there is no definition of application requirements (ripple, impedance, temperature, life, size..).  Once those are defined, the reason for dozens of capacitor types being available becomes more apparent.

That includes applications with very relaxed requirements say for life or ripple, which are well served by cheaper types, and often, cheaper brands.  But cheaper brands also have a range of products, some which serve very well indeed.

Similarly, without an EMI environment and setup specified, without spectra / test results, and without other application requirements (current and voltage rating, temperature, size, allowable ground leakage current, etc.), determining an EMI filter is impossible.  The best I can suggest is to construct such a setup (a LISN will be required; there are many LISN design threads on this forum) and measure it as well as you can yourself, and adjust things until it improves.  Then whatever configuration you settle on, build that on PCB and call it done.

Tim said it all. Nothing to add.
When I design my DC/DC (I have some small experience) - I experiment a lot with all available in local shops types and values.

[VEGETA]

There is no clear winner because there is no definition of application requirements (ripple, impedance, temperature, life, size..).  Once those are defined, the reason for dozens of capacitor types being available becomes more apparent.

That includes applications with very relaxed requirements say for life or ripple, which are well served by cheaper types, and often, cheaper brands.  But cheaper brands also have a range of products, some which serve very well indeed.

Similarly, without an EMI environment and setup specified, without spectra / test results, and without other application requirements (current and voltage rating, temperature, size, allowable ground leakage current, etc.), determining an EMI filter is impossible.  The best I can suggest is to construct such a setup (a LISN will be required; there are many LISN design threads on this forum) and measure it as well as you can yourself, and adjust things until it improves.  Then whatever configuration you settle on, build that on PCB and call it done.

Tim said it all. Nothing to add.
When I design my DC/DC (I have some small experience) - I experiment a lot with all available in local shops types and values.

for me i don't have such local market, meaning not like US where you can get stuff from digikey and mouser locally and so on. but i can for sure make a design taking all into consideration then make a small prototype and see how it performs. will keep you posted about it when schematic is available.

but my requirements are not that hard to achieve actually. just low ripple and noise, but not down to single digit since it could be not so realistic.

[Vovk_Z]

then make a small prototype and see how it performs. will keep you posted about it when schematic is available.

Sure. You may buy and use cheap Chinese capacitor kits for prototyping. After sucesfull prototyping then you'll know exact requrements and you'll know which type of cap is better for some place and you'll know exact parts used so then you may order good and more expencive reputable parts from reputable suppliers (E.g. Mouser etc).
For example, I have three or four different Chinese resistor kits (TH and SMD), mostly for prototyping but they are good for any simple job too. I dont buy Chinese capacitor kits for prototyping only because I have local shops with a large range of different caps. I'm talking about kits because you get a large range of caps for cheap.