2

I'm working on an isolated 2-sw forward converter.

The output section has a toroidal inductor that I've a problem with. Here's how the output section looks like:

schematic

simulate this circuit – Schematic created using CircuitLab

The switching frequency is 100kHz and the required value of the output inductor is calculated as about 100uH.

I put a toroidal 100uH inductor (I think it has iron powder core) taken from a broken power supply and left the converter operating under 6A load for half an hour. Then I measured the output inductor's temperature as 50°C which is quite acceptable. So I calculated its \$A_L\$ (from \$L = A_L \cdot N^2\$ of course) as about 95nH.

Then I bought an iron powder core (Part Nr: T130-26K - http://www.pe-coils.com/IPC.htm) having the same dimensions as the other one and calculated its \$A_L\$. Surprisingly, it has the same \$A_L\$ value! I thought that I "accidentally" found the same core, so I wound a toroidal inductor with the same number of turns and using the same dia wire. As expected, I got a 100uH toroidal inductor. No problem until here.

The problem is this: If I use my hand-wound inductor, and under full load, it gets as hot as 95°C! How can it be? They both have the same wire thickness, the same inductance and cores with the same \$A_L\$ values. But where does that temperature performance comes from?

More funniness: if I use my hand-wound inductor, I get %87 efficiency; but if I use the other one, I get %92!

I measured Q values of both with a calibrated LCR meter: My hand-wound has 2.4 but the other one has 1.8. So, what can be the thing behind this issue? What should I do?

[SOLVED] EDIT:

I've experienced that it really comes from core losses. The one with model number of T130-26K is a quite lossy one. So I purchased some toroidal inductors from Mouser and tested each under the same conditions. Finally, one beat them all in terms of thermal and electrical performance. For the ones who are interested, it is 2300LL-101-V-RC from Bourns.

PS: Is there a way to close or flag this question as "answered" or something like that?

Rohat Kılıç
  • 26,954
  • 3
  • 25
  • 67
  • 1
    Well, the reduced efficiency shouldn't come as much surprise as you are wasting some heat. \$A_L\$ is not everything to characterize a core, different magnetic properties make things go into saturation regions in different ways. The cores datasheet should have some illustratnig graphs. – PlasmaHH Nov 10 '16 at 13:28
  • @PlasmaHH Thanks. Switching frequency is 100kHz, btw. So, for example, can I say "My core cannot be used at 100kHz" ? – Rohat Kılıç Nov 10 '16 at 13:32
  • strength of the magnetic field i.e. current plays a big role too. Since you seem to have the setup, to learn something about your cores, you could graph current vs. efficiency for both cores, as well as (for a fixed current) frequency vs. efficiency. – PlasmaHH Nov 10 '16 at 13:37
  • What core did you buy and what's your switching frequency? – Andy aka Nov 10 '16 at 13:42
  • @Andyaka Switching frequency is 100kHz and the core's part # is T130-26K. – Rohat Kılıç Nov 10 '16 at 14:33
  • Does that core have a data sheet? – Andy aka Nov 10 '16 at 14:42
  • @Andyaka Unfortunately, it doesn't. I found [this page](http://www.pe-coils.com/IPC.htm) but it doesn't provide technical details enough. – Rohat Kılıç Nov 10 '16 at 14:52
  • 1
    Ditch it then and find a replacement - most likely it cannot handle the frequencies - core losses are the most likely cause. – Andy aka Nov 10 '16 at 14:55
  • What material are you using? Iron powder is a broad range of materials. What flux swing density did you design for? What frequency? – winny Nov 10 '16 at 15:03
  • @Andyaka . . [Andy's Answer](http://electronics.stackexchange.com/questions/70604/ferrite-or-iron-powder-how-do-i-tell-what-an-unknown-core-is-made-of) here might help in diagnosis (did you forget about your old answer Andy ? ) :) – Marla Nov 10 '16 at 15:06
  • @RohatKılıç Put that T130-26K, vendor and 100 kHz in your original question. People don't like to plow though comments to get such vital information. – winny Nov 10 '16 at 15:13

1 Answers1

1

The core is clearly no good .It would be fine for DC but not for big flux swings .Your 2 switch foward convertor has good efficiency with a good coil.You could do better with maybe a Cool Mu type or better still a EE type if your budget will allow it .If you must make the crappy coil work you will have to reduce the flux swing which will come with penalties .The really bad thing about powdered iron is that it ages with temp and time .The core actually gets more lossey and gets hotter and gets even more lossey .This can and does lead to thermal runaway .Even if you do your temp tests and things are acceptable they may not be in say a years time .If you ran your hot coil for a few months you will find that it will be hotter and the efficiency will be less than 87% .In fact 48V PSUs are being recalled by a credible manufacturer due to output choke burnout while I am writing this .

Autistic
  • 14,235
  • 2
  • 27
  • 65
  • Wow! I forgot that question :) Thanks for your detailed and informative answer. I purchased some 100-125uH toroidal inductors from Mouser and tried each under the same operating conditions. One -which is a low-loss one and is designed for especially SMPS applications- beat them all in terms of thermal and electrical performance, so it took place in BOM. ... – Rohat Kılıç Mar 09 '17 at 14:54
  • ... Unfortunately, I cannot make any changes because the converter is tested and approved, and now it is in production. Also, size is a limiting factor for us so I "had to" use a vertical toroidal inductor. The converter will be used in railway vehicles and required minimum lifetime is 10 years. If the inductor causes something in that lifetime then it will be quite problematic for both the customer and us. – Rohat Kılıç Mar 09 '17 at 14:56