Power supply tank cap

Your ideas of going about things aren't suitable for even the cushy living room environment.

If you're planning on using a pricey toroid transformer then that changes the equation considerably, you would want to be more conservative in the rating. Anyone who is using a supply like this doesn't care about space, weight, and heat; so put the whole thing in a nice suitcase sized steel box using that old perforated decoration radiator screen as appropriate for ventilation of this 15% efficiency thing.

"Fred Bloggs"

(snip abuse)

** Go eat rat bait - f*****ad.

....... Phil

"Ban" schreef in bericht news:s8U_f.16769$ snipped-for-privacy@news.edisontel.com...

[snip]

To keep this thread alive for another while, what about adding a nice 50-100mH inductor + 2nd filter cap. Perhaps someone can do some rule-of-thumb math on that ;)

--
Thanks, Frank.
(remove \'q\' and \'.invalid\' when replying by email)

"Fred Bloggs"

(snip abuse)

** Go eat rat bait - f*****ad.

....... Phil

Stick with your wussy battery powered stuff.

"Fred Bloggs"

(snip more abuse)

** Go eat a whole packet rat bait.

Enjoy the results.

....... Phil

I'm not planning anything. I was just using the old formulas from my classes

35yrs. back. And my results come very close to what I spiced up today. Nice that the program calculates the rms value of those charging pulses.

I found one thing that amazed me a lot. The better the parts, the more the ripple goes up. So a really lossy tranny makes the best performance, the same is true for the Diodes.

So lets take the 12.7V/90W tranny with a loss of 20%. The open circuit Vrms would then be 12.7 * 1.2 = 15.4V This is nice, only one measurement needed if the manufacturer gives the right data for Pn and Vn. With 90VA/12.7V=7.09A and Rn=V^2/p=1.792R. Now we can calculate the interiour resistance, that contains all the losses in one component.

20% of 1.792 = 0.358R With a 22mF ESR=30mR capacitor rated for 5.3Aripple (Rubicon 5$) and 10A diodes we get:

conducting time for tranny 2*4.04ms at 60Hz

100%=115V Vout= 14.57Vrms; Vmin=14.04V; Vripple=1.035Vpp I tranny = 6.438Arms; 13.01Apeak I cap= 5.035Arms ; -4A +9.02Apeak

-15%=97.25V Vout=11.99V min.=11.47V I tranny =6.321Arms Icap= 4.859rms

+15%= 132.35V Vout=17.52Vrms; 16.98V min I tranny=6.607A Icap =5.242Arms Ploss=0.83W I try Bemelmans inductor later. I see everything in the green range exept the output at brownout. :-(

--
ciao Ban
Apricale, Italy

The only thing you're neglecting is the asymptotic behavior of VDC/Vopenckt as a function of WRC for a give Rtransformer/Rload. It's best to overdo WRC to make the VDC insensitive to changes in C with temperature and aging. There is also an extreme sensitivity to conduction angle with Ipeak and Irms versus Idc, and that diode 0.75V seems rather weak with possibly +20A Ipeak into C. This is why all that simplified arithmetic is worthless. People who have studied all these parameters and analysis in depth know enough not to waste time with it and go with standardized plots of all that mess, then confirm with SPICE run over a range of conditions. There are not so many components here that you can't worst case it out on your own without relying on that dumb SPICE shot in the dark.

How do you know that I neglected it? I set the simulation with a voltage on C, and I measure with the cursors only after 100cycles. For the diodes I use MUR105 model. And I told you it was 35yrs ago, that's why I am able to do it. I do not need any dumb and unrealistic thumb-rules like you, that might work in some cases, but not here, where we have a constant current load and not a resistor. And I'm one of those people who knows, that's why I confirm with spice. I can easily admit when I was wrong and I learn from those errors. I do not need to creep around and justify anything. My self esteem doesn't rely on putting anybody down to feel better, rather I feel already good and do not need to make up an image. I'm worse than what others think about me, but I can admit it. That way we can laugh together about me! BTW lets close this thread now, I'm gonna report about the simulations in a new one.

--
ciao Ban
Apricale, Italy

These so-called dumb "thumb-rules" relating to what is properly called "input filter design" were a topic of research and peer reviewed publication over 65 years ago. All related considerations have since been worked out, settled upon, expanded and expounded in extreme detail, and published and re-published numerous times since, in all the major power supply handbooks such as those from GE, RCA, Motorola, and several of the *rectifier* transformer manufacturers. Like I always tell many of the transgendered personalities on USENET and their little queen games of playing the coquettish can't-be-convinced types, go take a leap if you don't believe it. This discussion is over.

of

sorry, typo, meant "Gasperi"

actually, its to neglect to mention typical power levels.

Yes, when you work on little things that include a lousy transformer (>

10%) the transformer dominates. most little transformers are lousy, they have a large surface-area to volume ratio so dissipate heat well - in which case, to minimise cost, they are made very lossy. As stuff gets bigger, its SA/V decreases, and it must be more efficient. this is why, for example, little SCIMs are shitty, but efficiency gets a whole lot better above 10kW.

for line-operated equipment, that is NOT the case (well OK it is, but the transformer power rating is 3-6 orders of magnitude higher than, say, a wall-wart.

most off-line SMPS fall into this category; there is an optimum level of capacitance

I'm interested.

Cheers Terry

absolutely. So when you start designing for extremely high efficiency, all sorts of evil problems appear that earlier, lossier designs managed to ignore. Another good example is filter damping, often no problem with say tantalums, but if you go for ultra-low ESR ceramics or OSCONs, Q can skyrocket.