# Input rectifier cap

Discussion in 'Electronic Design' started by Hammy, Jun 9, 2010.

1. ### HammyGuest

If I calculated a 120Hz, 1.3Arms ripple current for an input rectifier
cap for a flyback, how could I calculate the core temp based on a
given maximum ambient temperature (50C)?

I'm interested in using one Panasonic 270uf 105C cap rated at 1.42
Arms @120Hz.

http://search.digikey.com/scripts/DkSearch/dksus.dll?vendor=0&keywords=EETED2E271BA

There's no way that the commercial SMPS's I've looked at are using the
105C rms rateing to select the input cap. Most I've taken apart have
either one cap or two in series for a switch doubler including the
350W SMPS powering my computer. They must be increasing the rms
handling capability of a cap based on a reduced operating temperature
then the 105C specified in the datasheet.

Basso's book mentions a multiplier for determining max rms current for
reduced operating temperature. I've looked at several datasheets for
caps and haven't seen one. Is there a rule of thumb? Also is there a
reasonable estimate for including the converters high frequency rms
contribution to the capacitor?

2. ### GrantGuest

A lot of them only want 12 months life -- don't care what happens after
warranty expires
Odd you don't have datasheet numbers for this, it was on datasheets
I looked at a few weeks ago. Look for a table of multipliers buried
in the details, the tables are small, perhaps easy to miss?

An example. Recently I worked through ripple current adjusted for
both temp. and frequency for two different brands of caps and the
multipliers are a little different. These are 85'C caps.

I built a bank of caps for about 600A ripple current for <45'C, but
it's down to about 350A at 85'C (~500A at 60), bank is 20 x 10mF/50
plus 60 x 4700/50.

First brand is 1.73x < 45'C and 1.5x at 60'C, second is 1.55x <45'C
and 1.30x at 60'c (both 1.0x at 85'C).

Multipliers for frequency varied more, first is 1.08x for 10kHz,
second is 1.20x for 10kHz (both 1.0x at 120Hz).

Grant.

3. ### HammyGuest

Mine has been going for 8yrs now. I doubt it could deliver 350W
continuous power.

I also noticed with mine when cleaning it , it has no Common mode
choke!! This explained why a TV operating off the same line would get
snowy when the PC was on simultaneously. Oddly enough it has all these
regulatory approval stickers and it's a Antec supposedly.
Your right it is there. I can double the rms current for 70C ambient.
I always thought the 105C specification was for core temperature not
ambient. I have been grossly derateing my caps! Oh well live and
learn.
Christ what's it for an industrial egg friar. ;-)

4. ### GrantGuest

A small Splat! (stored energy) welder, only ~25 Ws max. Built for
occasional use with the cheapest caps I could find Yet to make the
power dump switch, I'm thinking 16 x 75A MOSFETs in parallel might
survive the experience, otherwise I could play safe and use a big
stud mounted SCR I got in a box here.

Grant.

5. ### Phil AllisonGuest

"Hammy the Halfwit "
** Look at the data.

** The ripple current data for lower ambients is in the table on page 1 of
the pdf - it doubles at 70C .

But a 3000 hour life is no good to anyone - so the usual practice is to use
the rated rms ripple for 105C and keep the ambient down to no more than 65C.

With just a bit of luck, the electro will then outlast the equipment.

..... Phil

6. ### Phil AllisonGuest

"MooseFET"
** HUH ??????

The impedance of a high voltage ceramic cap is waaaaayy more than an electro
at SMPS frequencies.

..... Phil

7. ### GrantGuest

Very good Pics here: http://grrr.id.au/splat/

Shows how much this is a hobby for me these days, not go near CAD software
yet this century. I used to use pcad stuff in the mid-80s, on a new 25k
PC-AT box with fancy graphics card, the 'full' 640kB installed we had to
memory on the mobo; it was lots and lots of double decker chips, each a
20 or so pin DIP. PC-AT complete with the first run 21MB HDDs that were
prone failed due to contamination, we were told. Monitor was from Canada,
with five coax leads from graphics card -- 2 colours plus the two syncs.

Company paid \$19k for a program that came on a 5 1/4 floppy disk, and the
pcad parts library was total crap, we had to make all our own. When one
used a large X-Y chart plotter because local PCB house didn't do NC
drilling, no Gerber machine either.

Grant.

8. ### GrantGuest

You lumping in the tan whatever losses with ESR? Likely be an issue
for me soon, so I plan to put temperature probes in a capacitor bank
to derate or shutdown at high temperature -- besides, I'm curious...

Grant.

9. ### HammyGuest

CORNELL DUBILIER has some Java calculators for calculateing lifetime
based on ambient temp and ripple current. Its here for anyone who
wants to try it.

http://www.cde.com/calc/

The site seems a bit slow mainly when useing firefox. Has anyone else
noticed CDE is slow.

Here is an application note on how they calculate the values and how
to use the tool.

http://www.digikey.com/Web Export/Supplier Content/CDE_338/PDF/CDE_SelectingCaps.pdf

I plugged in my ripple both 1.3Arms @ 120Hz and 0.8Arms @100kHz, for
an inferior capacitor then the panasonic. The results were 0.44W
dissapation, Avg core temp 63C Expected life hours 84,000 hr's!! In a
max ambient of 50C, reasonable its only a 50W flyback no major
heatsources and I'm mounting a small fan in the enclosure.

10. ### JoergGuest

It isn't too slow here on Firefox. But the applets won't work even
though I did enable Java and all that for this site :-(

[...]

11. ### GrantGuest

Same here Download + reinstalled Java JRE for it too, no joy.

Unless it wants the full Java development kit installed, I lost interest...

Grant.

12. ### HammyGuest

Yea I should have mentioned it doesn't work with Fire Fox. It does
work with IE...... I know Blah

I found it handy even if I'm not using the caps they have loaded in
the tool it gives you an idea quickly on any similar cap.

Or if you couldn't be bothered the application note shows how they
make the calculations and you can just do it up in Math CAD or excel.

I have noticed the site doesn't seem Fire Fox friendly, even
compatable. I'm sure they did rigorous testing after all CDE is a
huge multimillion \$ corporation. So the only logical conclusion is we
are doing something wrong;-)

13. ### HammyGuest

For the record I do have JRE installed and it still doesn't work with
FF, it does work with IE.

They do recommend using either IE or Netscape for the tool. The site
does however claim to be compatible with FF.

14. ### BaronGuest

Hammy Inscribed thus:
Mmm ! Works fine for me in "Opera, Firefox & Konqueror"...

15. ### JosephKKGuest

No joy here. Not in wine nor in native linux. Sun Java 1.6.x both
cases.

16. ### JosephKKGuest

BZZZt. The executive managers making such declaration have no such
direct knowledge. Including the CITO and the CIO.

17. ### JosephKKGuest

Geez, Netscape "died" many years ago. What is in their heads?

18. ### JosephKKGuest

Possibly they are on my cookie shit list.

19. ### JosephKKGuest

Ah, in the early transition era. When serious EDA was trying to
understand how to migrate to cheap mass produced HW instead of expensive
*nix workstations and EDA capable PC class HW was only comparatively
economical compared to *nix workstations.
Who else remembers TIGA cards?

20. ### JosephKKGuest

<snip>

Dunno, that might be in the hockey puck range of SCRs.