Small SMD xformer for a forward converter with double isolation

Reinforced at only 300V??? I've never heard of that before. Do you really mean a peak rating over 2kV, for line-to-operator insulation?

Ethernet transformers are hi-potted at 500V or so, and the average pulse transformer maybe 1000.

Tim

-- Deep Friar: a very philosophical monk. Website:

Check with Murata:

If you ask around mention medical grade (EN60601), that's where most of the highly isolating ones are marketed.

In Europe Wuerth is a really good place to start. I know they make offline flyback transformers with several kV of isolation. Those contain gapped ferrite cores which is not so nice in your case but can work. You could ignore the large primary winding and go in via the LV helper winding (which is isolation-wise on the primary side) and then the turns ratio won't be so extreme. Not sure about your size requirments though.

--
Regards, Joerg

http://www.analogconsultants.com/

something here that fits?

-Lasse

Does it have to be a forward converter, at this power level? If you dump the output inductor, you can combine real estate to allow for a larger isolation component.

RL

Other options might might involve picking a non-regulating topology, no matter how configured, and post regulating. You might even get this kind of power level by diac-pulsing a suitable signal transformer at a reasonable rep rate.

PE-68629 and PE-68630 (pulse engineering) are reinforced to he coordinated EN60950 standard.

RL

really

300Vac is actually 230Vac, but during the entire lifetime. So the insulation must not break down due to long time exposure. Many of the xformers rated for 1500VRMS tests has no double insulation and the xformer is only designed for long time exposure to low voltage (like below 60V and the occational spike). Murata 786J Series is an example of that, rated for 1000Vrms, but long time exposure is 60V:

(see the last page)

Regards

Klaus

This part is certainly better, but pretty sure not meant for long endurance if I ask about it. It is also not double insulated, and marketing would surely add that marking if it was approved for that ;-)

Regards

Klaus

Nice, it has the goverment approvals, looks promishing. Thanks :-)

Regards

Klaus

I have a very tight power budget, aiming for above 90% efficiency for the converter, so the hard driven converter is not an option I think. The converter will run with 20V in and 3.3V out, but I will be using

1:1 xformer, a low duty, to have less loss in the output diodes (only forward voltage related)

But, it does not need to be a forward, can be anything, but sofar the single switch forward looks like a good choice. Has full reset of the core and easy access to the current sense resistor for current mode control.

I will be testing if I gain anything from running it bang-bang. Full duty, large output cap, and shutting down the PWM when the output is high (cycle skipping to reduce standby losses)

Regards

Klaus

That would be great, but I could not find the information on the standard they are approved to. The 1500Vrms is just a 5sec test, but I may be overly pessimistic...

Regards

Klaus

Yes, I've seen some of their datasheets before -- for instance, they have a line of 3kV (peak) DC-DC isolators, but they're still only rated for SELV. I suspect it's just because they're cheap, or chicken, and don't want to get the approvals done. Which really just means to us, their 1.5kV or 3kV or whatever rating is meaningless.

But you require reinforced, rather than functional or basic?

I've usually seen things like, 300VAC working, 2.5kV hi-pot (AC RMS or DC peak, 1-60 seconds, etc.), maybe with a pulse peak rating up to 5kV or what have you. And numbers being numbers, the manufacturers prefer to display them by the highest number they can dig up, so a device that's listed at 5kV might *really* be good for 240VAC line use. So it looks funny when you said you're looking for a 300V device, when it's usually called 2.5kV or however much.

Tim

-- Deep Friar: a very philosophical monk. Website:

300Vac is actually 230Vac, but during the entire lifetime. So the insulation must not break down due to long time exposure. Many of the xformers rated for 1500VRMS tests has no double insulation and the xformer is only designed for long time exposure to low voltage (like below 60V and the occational spike). Murata 786J Series is an example of that, rated for 1000Vrms, but long time exposure is 60V:

(see the last page)

Regards

Klaus

They don't always advertise a rating, sometimes only the breakdown voltage because otherwise they must go through an expensive agency review. If you need transformers with certificates I'd ask such companies for med grade. Those come with "the papers", for example med grade LAN transformers and maybe by now even PoE transformers.

Not sure what small means in this context but a flyback transformer using just the helper winding and secondary might work. Like this one, which is stated as reinforced and 400VDC working voltage (4500VAC breakdown):

So you'd go in at 3/4 and tap off at 6/8. Just warn people not to touch pins 1 and 2 when operation or ... bzzzt ... ouch ...

--
Regards, Joerg

http://www.analogconsultants.com/

What standard are YOU refering to, when you ask for double-isolation in a transformer?

The 60950 coordinated standard (IEC60950 - EN60950 - UL60950 - CSA-C22.2 No.60950-00) for 'Information Technology Equipment' was in it's 3rd edition for most participants around Y2K, after much hammering out. Unless you're in biomedical, avionics, military, or some really sticky telecom, it'll one of the safety standards you'll find most commonly referenced in specs.

It defines (among other things) the physical requirements of isolation barriers, materials, clearance and creepage distances and their proof-type testing in certification procedures for specific operating voltage classes in specific operating environments. It also defines these operating voltage class groupings and environments.

There are also transformer standards (UL and other), but you're not selling a transformer. It's the 60950 requirements that ride rough-shod over internal construction of your product, and all components used in it. In small parts, creepage can be a sticking point - so pre-approval is verrrry useful.

If you're expecting more than generic 30 degree rises in the transformer, then UL will step in and ask for a listed insulation system to prove suitability for the higher temperatures expected. A pre-aproved part from a vendor (like Pulse) is very likely using their own listed insulation system. Feel free to inquire.

RL

A flyback offers current sensing, just not output current sensing. Other methods may offer other means of limitation. With 20V in there are many ways to skin the cat.

The efficiency isn't likely to be 90%, at this 100mW power level or

3.3V output voltage. Look at the control circuit's power consumption, for a start.

RL

Highly doubtful you'll get 90% efficiency at this power level. That's only

10mW of loss!

3.3V @ 100mW is around 30mA. That times a 0.33V shottky output diode is

10mW right there.

you could get fancy with sync rect driven by a transformer but even driving the gates of the transistor consumes SOME power.

Just saying you have a tough spec to meet.

Maybe I misunderstood your post, but I think it stated that there was a reference to the pulse part as approved according to 60950, but I simply could not find the reference in the datasheet.

We must conform to 61010 and UL508, and the product har double insulated marking, Class II

The 30 degree rise can be tough to achieve, but we will normally use only UL listed materials

Thank you for your good comments :-)

Regards

Klaus

Yes, it is a tough one. Sofar I will be running the PWM with a 4047 CMOS, low power comparator and error amp. The single switch forward will not have a reset winding, but resonate on the VDS capacitace of the switch FET. Secondary conduction losses is reduce, running at low dutycycle and 1:1 xformer ratio (low VF/transformer winding ratio)

Using bang-bang, could then employ the use of syncronous rectification on the secondary side, since duty cycle is fixed at close to 50% a self-running sync FET option is used (winding voltage turns sync FET on)

Regards

Klaus

n

only

s

ng

-

Yes, is a tough one. It is however only a target :-)

OK, that's strange. The 60950 listing seems to have been removed from the parts between 1996 and 2011. The original sheet I was refering to is obsolete. Temporarily at:

Mark with XXXX if you file it somewhere.

I'm unfamiliar with either, so I don't know where the insulation class is pulled from. There may be completely different definitions for insulation thickness, creepage and clearance that you can use. The UL508 standard, when coordinated, will apparently be UL60947, UL61010 or UL61800, depending on what the product actually does. RL