Through hole capacitors vs SMD capacitors

[VEGETA]

Hello,

in my latest psu design which finally worked as intended, I needed to put more bulk caps on 3.3v rail output. long story short: space is small and eventually required 3x2200uF (besides 2x470u polymer + 22uF 1210 ceramics).

I previously used just SMD caps while the board already has some through hole components. I figured out why not see elec. caps which are through hole. I decided to use Lelon caps due to performance vs price.

I came to the following foundings:

1- through hole caps are better in terms of ESR, I figured maybe due to SMD base leads having more resistance and inductance. mostly they are 2 times better.
2- through hole caps are cheaper, not always cheaper and not always by big margin.

here is a simple example of directly comparable caps... specs -> 470uF, 16v, diameter of 8mm, rated for 5000 hours:

TH (RXW471M1CBK-0811): price is 49.7$ for 1k quantity, ESR is 85 mOhm.
SMD (VZH471M1CTR-0810): price is 63.8$ for 1k quantity, ESR is 170 mOhm.

but for my design I went with RXW222M1ABKF1020 which has 35 mOhm which is the cheapest of RXW series (low ESR series) of about 134$ for 1k, cheapest SMD low ESR VZH series is VZH222M1ATR-1316 which has 60 mOhm and its size is bigger + 243$ for 1k.

adding to that, most teardowns I see of PSU equipment especially high end ones always use through hole, power modules like Recom and meanwell use through hole.

as for polymer caps... through hole Lelon 6.3v 330uF cap has 8 mOhm ESR vs 15 mOhm SMD counterpart while through hole is a bit cheaper.


are my findings correct?
what is your opinion and choices for this?

what drives designers to using through hole vs smd?

[T3sl4co1l]

"Needed" is putting rather a lot of weight on that word, there.

SMT vs. THT: I wouldn't say that's true in general. SMT caps are just regular electrolytics with, usually the can painted differently (lacquer vs. shrink wrap?), and the leads coined flat and bent to the side, trapping a support/base piece.

SMT caps are generally smaller, and certainly can't be as tall, so there's that.  You will find lower ESRs in larger and taller parts, but they're not unavailable in SMT because of the ESR, they're unavailable because of the size and shape.

Polymers are also available in molded chip form, which can have lower ESL (wider terminals, lower profile to board).  I don't think ESR is affected per se / in general, just that it tends to be low as these are typically low-voltage parts, e.g. 4V 330uF for CPU bypass.

Tim

[VEGETA]

"Needed" is putting rather a lot of weight on that word, there.

well it is a heavy word fore sure but the last ripple\noise due to 50-60hz power sips didn't go until I added 3x2200uF bulk caps. I tested it and it worked perfectly fine, I am ok with that. 

You will find lower ESRs in larger and taller parts, but they're not unavailable in SMT because of the ESR, they're unavailable because of the size and shape.

doesn't negate my findings. for another example, I looked at 50v rated 1000uF cap from RXW through-hole low ESR series vs it SMD counter part from VHZ series... the through hole is 34 mOhm while the SMD one is 73 mOhm... more that 2 times.

however, as you said... SMDs are not tall enough to have that bigger size and lower ESR. this RXW cap is 16x25 vs 16x16.5 for SMD, an 8.5mm difference.

this also does not negate what I said and noticed, which is in general and equal comparable parts through hole is lower ESR.  the height is the key factor.


I find through hole in every youtube teardown, yes smd exist but not as much.


EDIT: I found good reviews about Samyoung caps, and found  NXQ10VB2200M10*25 LO 2200u cap to be 24 mOhm ESR at 100$ per 1k vs my previously preferred Lelon RXW222M1ABKF1020 which is 35 mOhm at 134$ per 1k. anyone can recommend Samyoung over Lelon?

[JustMeHere]

When you look at ESR, you also have to consider the test frequency.

[VEGETA]

When you look at ESR, you also have to consider the test frequency.

Both Samyoung and Lelon specify those figures based on 100 Khz based on their datasheet.

[Vovk_Z]

Just use any polymer TH caps. They have quite easy job here in your desing, I mean there isn't high riple current or too high temperatures etc. So any should work fine and will have a long life.
You have a lot of space (height etc) and don't have to make a large quanty so you may use TH.
SMD parts are fine for small crowded places and pick-and-place machines.
For example, you may look at ATX/microATX vs notebook motherboards: the former use TH electrolytes and the latter - SMD.

[VEGETA]

Just use any polymer TH caps. They have quite easy job here in your desing, I mean there isn't high riple current or too high temperatures etc. So any should work fine and will have a long life.
You have a lot of space (height etc) and don't have to make a large quanty so you may use TH.
SMD parts are fine for small crowded places and pick-and-place machines.
For example, you may look at ATX/microATX vs notebook motherboards: the former use TH electrolytes and the latter - SMD.

I have choosen:

3x 2200u (Samyoung NXQ10VB2200M10*25 LO) - 24m Ohm 10000 hours, in case no stock in LCSC then its Lelon counterpart is 35 mOhm.
2x 470u polymer (Lelon ORK471M0JBK-0608) - 8 mOhm 5000 hours.

I feel these are most suitable and cheap enough... although I do like Samyoung NXH series for its lower ESR and very high ripple capability but the price is high. if I need to do the 6600uF then I would need to put 2 of (NXH 25V3300 16*25), each one is 13 mOhm 10000 hours and it is used in apple charger. However, I do know that putting 3 elec. caps is better than 2 in case of reliability.

Plus, I keep reading that one shouldn't get Nichicon, Panasonic, etc... of Japanese branded elec. and poly caps from LCSC due to being old stock and so on. therefore limited my self to good non-jap brands like Lelon and Samyoung.


as for input caps to the switchers (which are fed from 12v output of power module), I came to choose Lelon RXW221M1EBK-0611 which is very good in terms of ripple handling and low ESR despite really not needed that much performance. it is 6.3mm which matches the current design, I need 5 per board where 3 of them form pi filter as main filter directly after module, and the other 2 for switchers input.

Am I overthinking this? over-elitism? 

[Vovk_Z]

Am I overthinking this? over-elitism?

Even if you are overthinking then it's not bad itself if you combain thinking with measurements.

[VEGETA]

Am I overthinking this? over-elitism?

Even if you are overthinking then it's not bad itself if you combain thinking with measurements.

ok, I have done measurements and findings in the original design thread. I measured the behavior with various caps numerous times until I found the necessary limit to be 6600uF or so.

now I shifted my focus on if I should get 3x 2200uF caps or 2x3300uF. Both are good type as mentioned above, but the 3300uF ones are clearly better choice from all sides except the fact that 3 in parallel are better than 2. despite this, I am leaning more towards 2 since the output will also feature 2x 470u low ESR polymer ones and 2x22u 1210 X7R caps which will contribute... which means ripple won't be all on elec caps. I do like the NHX series mentioned, 10000 hours and very very low ESR with +3.3A of ripple handling which is way more than needed.

what measurements you think I still need to do after doing all this and verifying on actual end-goal usage?

[Vovk_Z]

Two caps in parallel in your case is better than three because I don't see any reason to have third cap. Third cap will just take additional place onto a PCB.
If 3300 uF caps are not rare then use them instead of 2200 uF. It's that simple.

Upd.: 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 it has 10 mm diameter. That's much smaller then 2 x 16 mm 3300 uF caps.