Torroid vs. SMPS

[ChocoHolic]

Hi Guys and Ladies!

Somehow I am not getting lucky with my PSU concept for my subs.
I planned to design a SMPS which would allow universal AC input and additional 12V car battery supply.
The part which makes me thinking is the line SMPS.
First I planned a PFC-Boost + Halfbridge Forward Converter.
With my lower power requirements (600W max. / 150W average), I started thinking about an Flyback, which would allow a simple single stage design.

BUT :
No matter which topology I choose there are several drawbacks already in theory.
-The safety standard EN 60065 (Table 11) would require enormous creepage distances for reinforced isolation. Safety class II is a must for me as I do not want to have a mandatory earth connection.
In case of a flyback I would end up in a minimum creepage of
12.6mm and in case of PFC-Boost + Halfbridge around 10mm.
...values which make a suitful small sized transformer difficult...
The leakage inductance for the flyback is critical anyway.
BTW: I found some cut core material, which can handle high
flux and HF operation (pic...).
-EMI filtering:
a) I would need massive X-Filtering.
b) To avoid HF-noise on audio signal ground, there would probably the need of screening layers in the trafo in order to catch the capacitive leaking currents from the fast switched high voltages.

Size and weight of all this will be at least the same like a 250VA torroid.
I will need large caps ( 2F ) for the 12V battery supply anyway.
Coming from this it looks quite reasonable to use a simple 120V/240V ==> 10V torroid and a rectifyer instead of all the trouble above.
2F caps will lead to an acceptable low rippled 12V supply and
a traditional regulated push pull SMPS should then give stable
+/- 50V rails for the amp itself (or any other desired voltage).

So in my application the only remaining advantage of the
line-SMPS seems to be the possibility to draw sinusodial currents from the mains.... hm, who cares at such low power !!

What do you think?

[Eva]

Don't let outrageous safety regulations pi$$ you off. They were written just for that purpose, in order to help big companies to monopolize SMPS development (and EMC actually states for 'Eliminate Minor Competitors').

Have you ever seen a 50Hz standard toroid with three layers of mylar tape between primary and secondary windings? or even with a faraday shield? No! Most of them use a single layer of a low quality non-adhesive tape with some overlapping, that melts at approx 90ºC. Furthermore, they suffer from high primary to secondary leakage currents due to thin insulation and huge capacitance, so earthing is always required. They don't met any leakage specification, most brand-new big toroids will shock you if you touch any wire from the 'isolated' secondary windings with your feet wet on the floor.

Have you ever seen a 50Hz 'EI' transformer witn 16mm or 10mm primary to secondary creepages? No! They use no creepages at all!

Everybody is using these 50Hz transformers without trouble and nobody is messing with their manufacturers.

I hope you got what I'm trying to explain.

PD: I use just 5mm creepages (2.5mm per layer) and some plastic tubing for my SMPS transformers, and this already makes them *far safer* than most 50Hz ones. Earthing is a must anyway.

[Danko]

What connect you to the wall socket's earth on the SMPS? Where do you connect it? To the heatsink?

[ChocoHolic]

Hi Eva!

Thanks for the clear words. And I believe that there are
many 50Hz trafos out in the field that have poor isolation, even
if you pick a certified one.
Sometimes it seems to be sufficient that the trafo just passes the
4kV isolation test, to get a safety certification.
I only wound off one E-core-trafo, 15 years back. And that one had triple isolation layers between primary and secondary. The creepages would have been to small from what I would expect, even when I take into account that for a 230V standard transformer 5mm should be sufficient. But that transformer also had a shielding layer to be connected to ground and with this I expect that basis isolation would do in that particular case and so half of the creepage would be OK.
The inconvinient high requirements for the creepage in my SMPS are resulting from the fact that the RMS and/or DC values (that's the criteria according EN60065) in my boost converter would be 480V compared to 230V in a standard 50Hz torroid....

With regard to EMI, I feel that a shielding layer in a 50Hz trafo is less important than in a SMPS. The SMPS has high voltages and excessive du/dt typically ranging between 100V/µs .... 10kV/µs, while a 50Hz trafo operates with 0.1V/µs .

Do you really always ground all components?
One of the earliest reasons for humm that I discovered were ground
loops between the power plug and antenna...
In the beginning I simply decoupled the antenna shield.
Since some years I decided to change to class II designs for my DIY projects... ...and up to now I am quite lucky with that...
Many HiFi manufacturers have done this step, too.
Just N & L at the mains plug, no earth.

Danko, I guess your question is also answered above.
For my shieldings I planned a pseudo EMI-earth formed by
two Y-caps between L&N. The center tap of these both
would have been connected to the shielding. Shielding
would be inside the SMPS trafo, between primary & secondary.
And around the entire SMPS unit .

Attached a picture of my first winding trial for the flyback solution. 6mm margin tape on each side ==> with the additional
thickness of the isolation between prim/sec, I would get around
13mm creepage... The wires are isolated with a shrinking sleeve, which gives 0.6mm isolating wall thickness.
This design did not have interleafed prim/sec and the leakage
inductance was around 7µH. Not funny if you go for a flyback with
20A peak in case of 120V/AC.....
Even with interleafed construction, I guess still half of it will remain.
And I would have to consider the inductance of the rectifier loop on the secondary. Let's guess around 10nH, resulting on reflected 2µH on the primary side (Trafo: 15 turns : 1 turn).

BTW:
Now I am between the seats. Yesterday I was tending more
to the simple way.
Now ....

[ChocoHolic]

...uhps, I forgot to say, that the pic only shows the primary and the
zero-detection-winding.
Then a triple isolation followed, which can be seen in the pic of my first post. And the secondary for measuring the leakage inductance was a
copper foil, full winding width and 0.3mm thickness.

[mzzj]

Originally posted by Eva
Don't let outrageous safety regulations pi$$ you off. They were written just for that purpose, in order to help big companies to monopolize SMPS development (and EMC actually states for 'Eliminate Minor Competitors').

Have you checked that you have right medication?

I have put couple of small E-trannies apart, and yes they seem to fullfill creepage distance requirements with rather clever structure. Not so much experience from toroids as they are pain in *** to rewind, but that one what i put in pieces was insulated with 15mm wide mylar strip, with 2/3 parts overlap so efectively 3 layers of insulation. Or check almost any chinese piece of sith ATX power supply to pieces, there is at least 6mm creepage distance(wich is enough if primary and secondary sides are grounded?)

Recently i tested all of my junkbox trafos with 5kV ac hipot, all passed without breaktrough. Some of bigger toroids were quite high on leakage, something like 10mA at 5kV, but thats ok. Some double insulated potted E-core was superior to anything else, 0.2mA .

[mzzj]

Forgot to mention that toroids are usually? only class-I approved so you need grounding of primary side also.
Edit: and with little extra money Class-II

[Lars Clausen]

EVA: I think you should be very careful to encourage people to ignore safety regulations. These voltages can be dangerous, so i think Chokoholic's respect for EN60022 (safety for IT equipment) is very well considered.

Chocoholic: Have you thought about a resonance mode push pull converter? IME they are suitable for high power jobs. But forget the 12 V input to the same converter, you will have too much copper on your main transformer. Make 2 separate converters and join them on the secondary side.

[jackinnj]

Originally posted by Eva
Don't let outrageous safety regulations pi$$ you off. They were written just for that purpose, in order to help big companies to monopolize SMPS development (and EMC actually states for 'Eliminate Minor Competitors').

All you Greens should look at it this way -- any noise created by an SMPS is wasted energy.

I am sure that GE, Siemens and Phillips would recommend that their CAT scanners be situated near unshielded ATX power supplies.

[Eva]

Oh, it's always very healthy to have some safety regulations, but something is going terribly wrong when they are routinely ignored for certain classic applications and by big companies without nobody complaining. I have to go now but I will put some examples later.

Same goes for emission and susceptibility regulations, it's healthy to have regulations but something is going terribly wrong when no small company can afford costly though simple approval procedures and they are not allowed to certify equipment by its own, thus then can't develop and sell switching equipment, even though they may actually met regulations.

Anyway, I think that Chocoholic got the point.