Where to get coil formers....

Cosmo?

Nice folks.

John

EI187 bobbins with that many pins would be rare now, outside of the era of signal transformers.

You'd be better off using a regular pin-free bobbin, and mounting it on your own platform. A platform might be kludged using a wide-body IC socket or 0.1 inch terminal strips, and there are dynamite epoxies in the local hardware nowadays.

If 14 terminations on a commodity bobbin are critical, find a suitably sized bobbin first, then go looking for the cores....

EI187 is unlikely to be offered with more than 8 or 10 terminals. (eg magnetics PCB180881 - 8pin horiz, Lodestone B-EI187V-10-1-PH3 - 10 pin vertical)

There are small SMD structures with up to 14pins. Examples of these are offered prewound for 'multi-app' applications by Miller (multi-6), TDK (multiwinding PCA) and Cooper/Coiltronics (versa-pac).Check Digi-Key for these.

RL

What do other people do for, say, multi output DC-DC converters? Just spam it with transformers?

Oh for the days, when an electronic organ's power transformer had something like 5 secondaries, or an o-scope with more. 'Course, they didn't have to put all those connections in a square inch of PCB.

Ugly though. Hoping for something molded.

The board isn't laid out for an SMD transformer, but it's tempting.

Oh, another option is planar, though that really requires a lot of layers in the board to do right. I think the extra area would require a bigger board.

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms

Make the few that remain "flying lead" configuration.

One can begin one's own transformer business with them as a single raw goods source.

There aee very few apps that require multiple 'isolated' outputs. Normally a multi-output psu cinsiders the requirements of an integrated system with maybe two or three reference return requirements.

That determined - you use the hardware available, even if it means deviating from some idealized peconceptions. Transformer pins cost money, particularly if their count exceeds the easily obtainable; flying leads are an anathema to production personnel.

So you can theoretically get one output (or more) per pin, with a common return, if you pick the proper topology..

The source reqires a pin and, if isolated another pin for a return.

Safety standards may eat up potential pins due to spacing requirements.

Screens may require a pin (or two).

Figuring out who or what you're serving and how important each issue is will be part of the design process. Somebody's paying and has expectations that will need to be met. So far, end-of-life disposition, or the hobby after-market don't weigh large in making these decisions.

RL

Well, I'm certainly not multiplexing anything here.

I've got a flyback converter, so it's got 2 pins for the primary, 2 pins for feedback*, and 3 pins for each secondary. There are two secondaries right now. I suppose I could imagine using 4 for some purposes, but I would like to avoid that tedium.

The outputs are +10/-5V, so the only way to multiplex them would be a funny duty cycle. Which would require forward converter action, and would make regulation tricky. So the whole circuit starts looking funny if I do that, and maybe performance suffers.

*If you balk at having an extra feedback winding, instead of optoisolators, my excuse is, I don't have any reason to regulate one output tighter than the other, so instead I'm just regulating both poorly. If that's not good enough, then imagine I said it's a supply winding instead, like for a TOPSwitch.

I'm not using a TOPSwitch right now, but I may on a future version. That'll be even nicer, because then I can replace the (iron cored) power transformer with _yet another_ secondary on the converter.

I suppose the best solution in that case will be, use three seperate transformers, put down a shitload of turns for the primary on each, and wire the primaries in parallel. That's one transformer for each isolated supply, and one for the 12V output, which will be regulated by TL431 + opto, and will have the TOPSwitch's supply winding.

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms

legg wrote in news: snipped-for-privacy@4ax.com:

Good its not just me.:-)

I read through that twice and I dont know what the hell he is doing?

Shame on me for not drinking coffee, nor drawing a diagram:

For the DC-DC converter.

For the TOPSwitch model, add SEC3, a single winding of 12V, and change PRI to ~120V.

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms

Could you say that again? Even in print, I can't determine what number of isolated secondary returns you are trying to get, and what outputs are associated with each.

RL

Correct. But it needs its auxiliary supply still, so the pins are still required.

Or did they make one that uses so little current that it's acceptable to drive from a dropping resistor?

Possible. Leakage screws that up a bit, so regulation will take a hit. It's also on the wrong side of 12V, which requires, say, another amp.

Hmm, I lied. It's not 12V input, it's unregulated power that goes into a

7812. More like 18V nominal. I keep forgetting that it comes from a "12V" transformer that doesn't produce anywhere near 12VAC.

I've already done it, using the "flying wire" approach. I only need a couple of watts. EI187 is "too big", in that it takes very little gap (none, IIRC) and only a few turns to handle the flux, like 6+3T for the secondaries. So all the tape and insulation completely dominates the size of the thing.

That would be excellent for the secondaries, though.

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms

So thats ten pins plus creepage. Without the fb winding it's 8pins plus creepage and could fit on the commodity 10 pin bobbin.

- If the Topswitch will require an optocoupler, you might as well get used to one; that would get rid of the fb winding soonest.

- If you take your feedback from the primary, you could dispense with the feedback winding. The flyback voltage (wr to the +12V input) will be ~ representative of the output, after filtering and peak detection. An RCD snubber would produce this voltage, for feedback purposes, free. You'll have to keep loading into account, and maybe clip the actual leakage mag energy through a parallel path ( it's decay time is much less than that of the functional mag energy you're trying to use).

I'm not sure how much isolated output power you're expecting to pass through an EI187 transformer. I've actually seen them used as CM filters in 50W off-line products, though I doubt much thought was behind it, save that the parts were available, or used elsewhere.

If the primary is as simple as it can be, it might benefit from 3xfep wire, to remove margin and layer insulation from the bobbin fill, but large turn-counts argue against it's use in most places.

With internal safety margins and external body creepage issues, there's not much winding space left, inside an EI187.

RL

"Tim Williams" wrote in news:i7ubul$hp0$ snipped-for-privacy@news.eternal-september.org:

The NCP12XX series has a integrated HV current source operates directly off the HVDC bus its good upto about 10 to 20W depends on your fets drive requirments (Qg).

I'm pretty sure their are other controllers with an integrated HV current source similiar to Onsemis.

For rectified and filtered 12VAC LT has a no opto series of controllers. Integrated 60V switch and current sense. It can operate directly off upto 40V dc.

I've never used that one though. [snip]

Their app notes all have a housekeeping winding. It should be possible to run the thing off the bottom end of a current snubber - somewhat like Lambda tended to do, (but only for the sake of recycling energy). If the start-up bleeder were across the current snubber's capacitor, it would reduce forward overvoltage in the current snubber's diode - but I keep forgeting that Top switches don't need an external bleeder. I assume they also discharge the bulk cap safely.

You would't really be using the circuit as a snubber (though it would do that job), it's just one way of powering the circuit without another winding. There are all sorts of hijinks you can get up to at low power levels but you don't want to populate the board with gimmicks at the expense of efficiency or size.

What is the 7812 going to be doing in later higher voltage instances ?

You must be running the thing above 300KHz, then.

Maybe a bit too fussy. The benefit on a primary, with fewer pins, is that isolation to the core may be less difficult to build up after you've 'solved' the internal margin issue. If you were going to have a flying lead pair, let it be the primary and get its terminations the hell out of there.

RL

Hmm?

Present setup is: iron cored power transformer > FWB (18V unreg) > 7812 (12Vreg for stuff). DC-DC runs from 18Vunreg.

7812 won't be needed with a regulated 12V winding, of course.

Nah, just a UC3842 at 120kHz or something like that.

Let's see, 5us on time at 20V is 100uWb, into 6 turns is 16.7uWb/t, at

0.4T Bmax is 41.7 mm^2 minimum area. 0.187" is 4.75mm, so a square is 22.5 mm^2. Yeah, the on-time will have to be about half, or the turns about double. I may be misremembering the actual primary turns I used. I may've also used a core slightly larger than 0.187", but I know it's not by much.

At any rate, I can recalculate the exact turns when I select a core.

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms