LED Driver SMPS Flyback Design (Transformers/EMI Filters)

[David]

Hi all,

I am in the process of designing my very first SMPS flyback transformer. My knowledge on the subject has gone from almost zero to the point where I am confident enough to go ahead with the project (that's what I love about electronics). Now there is one subject that has been bugging myself and giving me sleepless nights for the past couple of weeks (one of the downsides to electronics!), and that is the safety requirements. I am a "young player" in the Electronics industry, as Dave would say, (not long out of university) so please bare with me.

First, a very brief summary of the project:

* The transformer is for an offline flyback (90Vac to 265Vac), constant current LED driver (<9.1W). This is a primary side regulated (PSR) design.
* Due to space constraints I am forced to use a relatively small core (EFD is suited due to low profile).
* At the moment it looks as if an EFD20 will do the trick (however low Ae and small winding width means I will need multiple primary layers which is fairly unavoidable with this design). I have another question on interleaving windings but I don't want to bombard everyone with lots of questions at the moment...
* The transformer must meet the safety requirements of EN61347-2-13.
* The LED driver is a class 2 (double-insulated) safety product.

So here goes the question(s)...

1. I want to use a margin-wound transformer construction to meet the appropriate creepage/clearance requirements. However I have seen many different methods of meeting the requirements. Some application notes/designs use margins on both sides of the bobbin whilst on the other hand some use margin tape purely on the secondary side?

2. If anyone is familiar with EN61347 then I would love to hear from you. I am confused about the actual distances required for the design? Being a class 2 product, I believe I require supplementary insulation? (According to a table I read in EN60598-1)

3. Am I right in saying that you should account for the inherent clearances of the bobbin itself?

4. Where exactly are the clearances/creepage distances measured to and from (in regards to transformers)? Is it purely from primary to secondary windings?

5. Can all of the above be avoided by simply using triple insulated wire (TIW) throughout?

I apologise in advance for all the questions. I would not normally ask in the forums but I think I have exhausted all other options.

Cheers,

Dave

P.S. I am sure some sort of beer payment can be made for any helpful answers!

[David]

Bump

[Neilm]

The first question I have is who is building the transformer? Do they have the equipment to test that the transformer?

I am not too familiar with that standard but I do know EN61010 and have designed SMPS to that standard.

1) As long as there is sufficient creepage and clearance between the primary and secondary any method is OK. Some may be easier to build in certain situations, other constructions will affect the performance of the flyback transformer (for instance increasing the leakage which will increase the turn off spike on the FET)

2) Pass - In 61010 it is 3mm for double clearance.

3) The bobbin will provide some clearance to the pins. The clearance is the shortest distance in air between the primary and the secondary.

4) Clearance and creepage is measured from anything conductive (or whose insulation properties are insufficient) that is connected to the primary, to anything conductive (or whose insulation properties are insufficient) that is connected to the secondary. The clearance is the total distance in air between the two. Please note the phrase "in air". I was once given a unit to approve that the manufacturers said had the correct clearance, but they put a metal screw head in the way so they had less than 10% of the clearance they thought. Practically, this means that if the primary and secondary is connected to the same side of the bobbin, you have to deduct the conductive bits (i.e the pins and associated pads) from the distance between the two.

5) TIW wire might be OK as long as the insulation is sufficient - I don't know what the standard says.

The creepage and clearances will have to be proved by flash testing the transformer. The standard should say what the required test voltage is.

Yours

Neil

[David]

Hi Nelim,

Thank you for your response! I now have a few more answers to my troubles.

There is so much conflicting information on Flyback transformers on the internet, it's hard to know who's using the correct formula! Almost every datasheet and app note has a different approach or method for calculating even 'simple' things such as the primary inductance.

One particular problem I have found is how to calculate the required air gap within the transformer core. Am I correct in saying that this can be achieved by using I2L (Output current squared x Inductance) and then finding a graph plotting this against the air gap? From there I would then select the nearest (or custom) AL value and then I can determine the number of turns required for the primary?

Thanks for the help,

Dave

[Neilm]

Dave,

I must confess that my knowledge of flyback transformers is more to do with the safety side of the application. The few times I have designed them they didn't come out anything like I expected. Last time I didn't bother and gave it to a guy in the office who is a a magnetics expert and let him design it while I did the control loop.

Yours

Neil

[David]

I must admit, that could be the way this project goes dependant upon time. I'll definetly give it a go though, managed to source some sample cores and bobbins so I will soon start winding and inevitably blowing things up! 

[David]

Cheers guys. I already had a good read through the TI notes which were fairly useful. However, they describe yet another method for calculating gaps etc! Transformer design is hard work!

[David]

Hi all,

Thank you for all the feedback! I am definitely on a steep learning curve here but enjoying it at the same time. I now have a preliminary transformer drawing and have sourced all the parts to start protoyping it soon 

I now have a couple of questions regarding EMI filters for flyback supplies...

1. In the flyback design I am working on there is no ground connection (class 2), only Live and Neutral. Am I correct in saying that any noise flowing in the Live will flow back out Neutral, therefore there will be no common mode noise? - Hence removing the need for a common-mode choke?

2. I have seen many different methods of implementing filters to reduce differential mode noise in the hundreds of application notes I have been looking through. The most common arrangement appears to be a simple PI filter, as you may expect. However I have seen two variations of the filter (please see attached image). What is the reason for using two inductors? (It what I would describe as a "stacked" PI filter). Is the inductor required on the Neutral as we are dealing with an AC input? Can only the single inductor on the Live side be used?

Again, apologies for so many questions. I have been scratching my head over this one for a while. 

Cheers,

Dave

[David]

I have also seen the use of Common-mode chokes for input filters which utilise the leakage inductance of the choke to filter differential mode noise...

[madires]

1. In the flyback design I am working on there is no ground connection (class 2), only Live and Neutral. Am I correct in saying that any noise flowing in the Live will flow back out Neutral, therefore there will be no common mode noise? - Hence removing the need for a common-mode choke?

2. I have seen many different methods of implementing filters to reduce differential mode noise in the hundreds of application notes I have been looking through. The most common arrangement appears to be a simple PI filter, as you may expect. However I have seen two variations of the filter (please see attached image). What is the reason for using two inductors? (It what I would describe as a "stacked" PI filter). Is the inductor required on the Neutral as we are dealing with an AC input? Can only the single inductor on the Live side be used?

For a really good filter you would use a common-mode choke and a LC combination (2 Ls, like the upper part of your diagram) to filter asymmetric and symmetric noise. Since you got two wires (Live & Neutral) you need two inductors to "protect" both wires.