LLC resonant converters for cheap SMPS in small volumes?

[grizedale]

Hi,

am i right to assume that LLC resonant converters are not suitable for cheap power supplies in small production runs.?

after all, the LLC demands high toleranced resonant components, since it is essential to be very certain of the pretty exact resonant frequency, and high tolerance LLC components are expensive.

 

[Orson Cart]

certainly suitable for cheap power supplies in large production runs - Regards Orson Cart.

 

[grizedale]

though do you agree that the resonant capacitor has to be a well toleranced high voltage component, and such components are too expensive for small production runs......and a plain half-bridge would be a better option?

Also, the resonant inductor ("leakage") has to be very tightly toleranced, and whether its a separate inductor, or part of a "divided bobbin" transformer, its very expensive....too expensive for small volumes for cheap power supplies.

As stated, surely the resonant frequency has to be pretty well defined?

...so i would conclude that LLC resonant converters are either for exotic applications in small runs (eg radar PSU's) -or for high volume production runs where the tightly toleranced components become cheaper due to the high volumes....is this right?

 

[mtwieg]

Generally for LLC converters you don't necessarily operate right at the resonant frequency. You can under full load, and it helps, but you'll usually operate above fr. If your fr is uncertain, then you'll just have to limit the minimum frequency to the highest possible fr. This isn't optimal, your maximum power output will be lower, but that may be acceptable.

the biggest issue for low volume would be the transformer. I doubt integrated magnetics would be economical at small volume, but maybe discrete components could work out, if you find them off the shelf.

 

[Orson Cart]

mtwieg is correct, as you should always operate above resonance - the exact freq of the resonant parts is not absolutely critical, using discrete seperate inductor to lower the Tx magnetising L is a good idea as it makes the transformer more repeatable and with lower losses (no fringing flux to heat wires), sizing the L's and C to give a bit more than full power will ensure you always get rated power just above resonance even if the parts are +/-5%. Regards, Orson Cart.

 

[grizedale]

Hi,

I am sorry, but i beg to differ, ...cheap high voltage capacitors, (for the LLC Cap) will be +/-20% at least...........so too will be the resonant inductor, and even more tolerance for the leakage of the trafo.

I am afriad this is just unacceptable.

One would have to operate miles above resonance, and the topology is no longer worthwhile.

 

[mtwieg]

How "cheap" are you talking? MKP or similar caps are +/-5%, and are usually well under a dollar. The inductance will be worse, but as Orson Cart said if you allocate most of the "leakage" inductance into a separate discrete inductor, then that inductor's tolerance will dominate and will probably be better than the transformer. So for the inductor you'll probably be looking at +/-10%. So you might end up with +/-8% error in fr, which isn't that bad (consider that over your range of loads you'll probably go up to 300% of fr).

 

[grizedale]

Hi,

-I'll have to check this.......the High voltage caps i've seen have awful tolerances.

I would not be happy going to 3 x fr.

-this is because fr would likely be around 75KHz.........and mains connected switchers (even if following PFC stage) should always switch at less than 150KHz, otherwise conducted emissions get bad.

So one has to limit the frequency, which means the converter, at some light-ish load , goes into a kind of strange burst mode, and unwanted ripple appears at the converter output.

 

[mtwieg]

Hi,

-I'll have to check this.......the High voltage caps i've seen have awful tolerances.

Could you give the specifications for the primary circuit?

I would not be happy going to 3 x fr.

-this is because fr would likely be around 75KHz.........and mains connected switchers (even if following PFC stage) should always switch at less than 150KHz, otherwise conducted emissions get bad.

So one has to limit the frequency, which means the converter, at some light-ish load , goes into a kind of strange burst mode, and unwanted ripple appears at the converter output.

Well this is an entirely different issue from the component tolerances. Why do you think exceeding 150KHz would cause excessive radiation? So long as ZVS is nicely achieved, I don't see why EMI should get too bad.

 

[grizedale]

-the problem above 150KHz switching frequency is the conducted emissions