LLC resonant half-bridge isolated 200W Switched Mode Power Supply based on Fairchild's (now On Semiconductor's) FSFR2100XS power controller.
This repository include design files, shematic and single-layer PCB films for DIY high power high quality resonant AC/DC converter.
Forum discussion: http://forum.yu3ma.net/showthread.php?tid=10
The popularity of the LLC resonant converter in its half-bridge implementation is due to its high-efficiency, low level of EMI emissions and its ability to achieve high power density. These features fit the demands of many modern applications. LLC resonant half bridge converters are widely used in consumer electronics, like powering the display panel of LCD TV.
The FHA (First Harmonic Approximation) approach is based on the assumption that the power transfer from the source to the load through the resonant tank is almost completely associated to the fundamental harmonic of the Fourier expansion of the currents and voltages involved. This is consistent with the selective nature of resonant tank circuits. The harmonics of the switching frequency are then neglected and the tank waveforms are assumed to be purely sinusoidal at the fundamental frequency: This approach gives quite accurate results for operating points at and above the resonance frequency of the resonant tank (in the continuous conduction mode), while it is less accurate, but still valid, at frequencies below the resonance (in the discontinuous conduction mode). Good and properly constructed LLC works with basic harmonic of about 100KHz, has a very low content of other harmonics, sinusoidal current fed your transformer and output diode, has drastically lower dU and dI / dT of comparable hard switching topologies, minimal "ringing ", ie. generally speaking very "clean" waveform that is easy to filter.
An important aspect to analyze is the converter's behavior during output overload and/or short-circuit.
Unlike the hard switching there is immunity to impact loads being uninterruptedly repeated. This in turn means that the required rated output of LLC = amplifier power / efficiency amplifiers + 20% obtained. LLC converters can easily within a half-wave, for example - the bass tone on the amplifier, take load that is 50% higher than the declared and without any damage during the entire life cycle of the device.
Also, LLC of 180W has the same potential as the hard switching of 250W for use in an amplifier, and this is because of its principle of operation which allowed current limit that can be longer in time. Limiting the minimum operating frequency is effective in preventing capacitive mode operation only if the minimum (normalized) frequency value is greater than 1. This suggests that, in response to an overload/short-circuit condition at the output, the converter operating frequency must be pushed above the resonance frequency (it is better if well above it) in order to decrease power throughput. It is worth noticing that, if the converter is specified to deliver a peak output power (where output voltage regulation is to be maintained) greater than the maximum continuous output power for a limited time, the resonant tank must be designed for peak output power to make sure that it will not run in capacitive mode. Of course, its thermal design will consider only the maximum continuous power. In any case, whatever the converter specified, short-circuit conditions or, in general, overload conditions exceeding the maximum specified for the tank circuit, need to be handled with additional means, such as a current limitation circuit.
The converter specification data are the following:
- Nominal input voltage: 325VDC (230VAC)
- Maximal input voltage: 355VDC (250VAC)
- Minimal input voltage: 295VDC (210VAC)
- Brown-out: 260VDC (184VAC)
- Output voltage 1: +/-28V 3A
- Output voltage 2: +/-14V 1A
- Resonance frequency: 85 kHz
- Max operating frequency: 250 kHz
LLC converter belongs to the SRC (serial resonant converter) series, where the resonant network is established with two inductors and a capacity on the primary side of the converter (hence LLC).
This can be accomplished with a simple SMPS (Switched Mode Power Supply) transformer station with external inductance as Ls (where this is much easier to be done on the secondary side of the transformer), but integrated structure has some noticeable advantages regarding external inductance in terms of depending on the energy that is recirculated vs supply voltage. At this point we need to understand that the LLC with integrated structure is more flexible to change of the supply voltage than the SRC with external coil.
Basically, in terms of consumer, LLC is closer to the current than to voltage source supply (which applies to all SRCs), resulting therefore in exceptional immunity to overload and short-circuit. Also, all SRCs are dependent on relatively narrow span of voltage supply.
Unlike the SRC, there are PRC (parallel resonant converters), which can work within a much wider range of supply voltage (completely feasible with all the control circuits for SRC / LLC), but have some other drawbacks (again huge topic). For now, let us just say that they are more expensive and complex to carry out and that they can't perform with light-load and no-load.
LLC also can't work no-load, but it can work with very light loads (usually some parasitic load is sufficient for self-supply and possibly some light bleeder). Unfortunately, with light loads <1/3 to 1/5 (depending on model), LLC loses its great ZVS feature and efficiency abruptly falls at low loads.
However, this is not a significant drawback as it means that with low power, total energy is smaller, therefore the heat that is lost in general remains low and within acceptable limits. Therefore do not be surprised that the LLC is heated most with relatively small loads - this is normal.
As the load grows, its temperature decreases (we are here considering the primary side) to about 80% of power when a typical well-constructed LLC is coldest (except the output diode), then again his temperature rises towards 100% of load, when it reaches a similar value as with maybe 30%.
The hottest component under high loads on each LLC is the primary side of the transformer and it is quite normal. It is normal that the primary coil reaches about 60-80ºC, as well sa the core.
Otherwise, a ferrite core for SMPS has minimum losses within the temperature range of 60-120 ºC typically. Lowest peak is at about 80ºC in the majority and manufacturers use this feature. If you make a SMPS and the transformer is cold you should know that you have a smaller efficiency than feasible.
Our sample (as well as the most modern ones) was performed with so-called "Integrated magnetics" structure, where the first two L components are made on the same core as:
-
Ls (or Lr, serial leakage inductance) that is not coupled with the consumer, ie. the one that the primary chopper "see" as the sum of all the inductance of the magnetic structure, which are not coupled with the consumer. It basically almost does not change with the change of consumer resistance (it is changing the minimum size in practice, due to the characteristics of the ferrite, but it is not of a significant importance).
-
Lp (or Lm) parallel inductance, or inductance of the parallel connection of the primary magnetization inductance, ideal transformer and mapped Rac (of the consumer) to the primary of the transformer. This inductance is variable, ie. depends on the mapped consumer resistance and its value can range from Lp (magnetization inductance of the transformer when mapped consumer resistance is infinite, ie. when the inverter has no-load), all the way to short-circuited where Lp is equal to zero. This inductance is the aspect of the chopper (rectangular signal generator) connected in series with Ls and resonant capacitor Cr.
Both of these inductances in series with Cr comprise serial oscillatory circuit that has several interesting characteristics :
-
Q factor, or quality of the circuit, ie. its ability for one or both (depending on the points of observation) of the input values (voltage / current) to be multiplied to a greater value than entered. Size depends on the relationship between the total L and C, and the resistance of the load entered into the circuit. As this can be multiplied, it can also be reduced. It all depends on Q.
-
Phase shift between voltage and current, which depends on the frequency applied to the circuit and the resistance of the load entered into the circuit, as well as relation between L and C. A very important thing!
-
The characteristic impedance (reactance), and the impedance "as seen" by the generator, which also depends on all three of the above mentioned values. The character of the reactance-impedance is very important and below the resonant frequencies it is predominantly capacitive, and at frequencies above the resonance predominantly inductive.
-
Resonant frequency, which also depends on the L, C and the resistance of the load entered into the circuit. It is the lowest in an LLC network when there is no consumption and equals to Ls + Lp + Cr (you can use the Thomson's form, I have only mentioned components as information), and it is the lowest when the load is short-circuit, and there are left only Ls and C.
Serial Oscillatory Circuit finally has L and C connected in series, with reactants of each component having opposite phase status and are deducted from each other.
In the center of resonance the two reactance void and the current of the ideal oscillatory circuit becomes infinite (infinite Q). At this point the generator is seen as a short circuit, while the voltages of individual elements (L and C) also reach the infinite value, but with the opposite phase positions, in terms of the generator they are void, and their sum is equal to zero, which is seen by the generator as a short circuit.
In practice, it is difficult to achieve infinite Q but it is not difficult to achieve such a kind of enlargement that is resulting in multiple voltage and current at the center of resonance.
At the center of resonance the only "obstacle" to infinite sizes is inserted / mapped resistance-load.
If we allow frequency of our converter in idle mode to fall on resonance frequency Ls + Lp + Cr (Thomson's form should be used) it would have a few kilovolts at Cr and Ls + Lp (primary side of the transformer ) and electricity generators (chopper) several times greater than the normal. This would devastate the three "participants" if only the generator lasted long enough...
Magnetic structures like our LLC are the most frequently performed in several ways:
-
By moving away primary and secondary side of the transformer at the central pillar. In this way we are adjusting Ls, and by inserting the gap between the core halves we are controlling Lp.
-
With the use of long and thin magnetic paths and by moving away primary and secondary side of the transformer and the thickness of the middle post Lp (such transformers were implemented in older models of Sony TVs)
-
By coiling one or both coils away from the center pillar as way of controlling Lp and moving them away from each other as way of controlling Ls.
-
Combination of above mentioned methods.
Using all three, or just more than one pillar to accommodate part of the coil + gap (that is intensively used by Dr. Ćuk in his converters of older generation, and he started using them again recently).
By inserting magnetic shunt with an additional part of the core, where it is convenient to use (this method is again often used by Dr. Ćuk and it requires custom core).
In addition to the above mentioned, there is an unlimited number of possibilities, but sometimes we are sacrificing the quality of the inverter in order of performing an easy to produce and inexpensive transformer.
The most often performed magnetic structure for the LLC in practice is the standard core with moderate band-gap between primary and secondary side of the transformer. This model is quite suited and easy for serial production and works well enough.
Do not get deceived by the simple display of Ls and Lp in the application notes. Reality is different than that.
In app. notes the two sizes are shown as the sum of all individual Ls and Lp, which vary for individual turns in the transformer. However, It is far more complex than this simple view.
In reality, the transformer is made up of many small Lp and Ls and the whole crowd of parasitic capacity around the transformer, both at the primary and at the secondary side and control of these elementary parts is highly influencing the final operation of the inverter.
We mentioned that the performance is primarily based on a sufficiently good degree of efficiency and low cost of production. For example, one thin elongated core with four times smaller cross-section of the middle post convey the same power as the standard core which is more expensive.
But when we examine the long-term economic balance, the question is whether some companies earned more by investing huge amounts of money in the production line, to finally spare in the bulk of the core (which significantly changes the price), or other companies that use more expensive and massive iron-cores, but at the same time spared on investing in production?