Electronics > Beginners
LLC vs LCC Converters for High Output Voltage
T3sl4co1l:
IGBTs come in different flavors, roughly speaking differentiated by the hFE of the BJT element. Fast ones are capable beyond 200kHz; much of the current flow is through the MOS element, it's more like a MOSFET with a little bit of boost. At turn-off, most of the current drops sharply, then a small tail (a few percent of I_on?) "drools" out as the BJT element continues to turn off (recombination), taking perhaps hundreds of ns to reach full off state (leakage current).
Slower ones are limited to lower frequencies, usually the 10s of kHz, even the 1s of kHz for the big modules (100s to k's of amperes, low kVs). These have a smaller fraction due to MOS current flow and more due to BJT amplification. The voltage drop is lower (that's your tradeoff), but the turn-off is not nearly as sharp (maybe it drops sharply by 10-50% at turn-off), and the "drool" is much longer (~µs).
Smaller IGBTs always run faster; if you need operation at 100-200kHz say, at high power (industrial application), you may be better off using a stack of TO-247 sized parts in parallel, instead of a module proper. It may be advantageous to operate at lower voltages (600V parts are faster than 1200V+ parts), though that's not always practical for industrial application (where 400/480VAC 3ph input is most convenient).
But down at the power levels we're talking here, I don't think it matters. You're mostly paying for the package. What's inside it, doesn't really matter. Might as well go with the MOSFET, and not have to worry about "drool".
Tim
state_of_flux:
Thanks Tim.
Could you help clarify this point, as I find it an interesting method:
--- Quote --- For that, I would suggest, inverter (fixed duty and freq), driving the transformer, which has some variable capacitors on the secondary to trim its resonance just below F_sw.
--- End quote ---
Does this mean that the primary side would only have a series combination of L and C, then the additional parallel C value to form the LCC resonant tank will be realised in the form of placing variable capacitors in the secondary side between each of the secondary windings. And as I understand it, this is also acting as a way to help tune the transformer, rather than design the transformer according to the switching freq? And tuning its value tunes the transformer by manipulating the resonant frequency of the tank, thus changing the amount of current supplied to the transformer etc? Sorry if I'm sounding a bit dumb, just trying to get my head around it. :-/O
T3sl4co1l:
The equivalent circuit would be, inverter (constant voltage output), series capacitor, series inductance (including leakage), shunt magnetizing inductance in parallel with secondary self-capacitance.
I guess the point of LCC is the magnetizing inductance is large enough to be negligible, and the capacitors and leakage dominate the response. That would suggest an ungapped, high-mu core, which is fine.
Primary shunt capacitance wouldn't be needed because the primary will have very little capacitance in comparison to the secondary. From an RF perspective, it's just a magnetic loop to couple energy to the secondary, which is resonant.
Also, because secondary capacitance is inevitable, tuning the primary would probably be folly -- either you have so much leakage that you end up with a double-tuned network anyway, or the leakage must be extremely small so that Cpri acts in parallel with Csec. But that's going to be impossible with the number of turns and thickness of insulation required on the secondary.
So, the capacitor just goes on the secondary instead, which is fine, and it can be tuned there similarly.
Uh, the one catch, of course... finding a 20kV trimmer. ;D Probably better to do it with "gimmicks", e.g., pours on the PCB that can be cut apart or jumpered together. (FR-4 is a terrible capacitor, not so much because of losses at this frequency, but because of tempco -- it might be good enough, or it might exacerbate whatever tempco the transformer has. Or maybe it cancels out and is actually best. No idea, you'd have to test.)
Also in your favor, the Q factor is supposed to be quite low for resonant supplies, so tuning shouldn't have to be that tight. Probably tempco doesn't matter, and swapping out a few leaded ceramic caps will do.
Incidentally, if stacked secondaries are used, you only need to tune one or a few of them -- again, they act in parallel at AC, in series at DC. As long as the leakage between sections isn't huge (again, same argument as leakage to the primary), they'll behave. And that means you only need, say, a 5kV cap, instead of the full 20kV or whatever.
Tim
state_of_flux:
Seems quite complicated. :scared:
I'm quite a newbie so honestly its quite overwhelming but I appreciate your help regardless. I suppose if one was to use voltage multipliers at the output also, this would further reduce the value of variable capacitance voltage that would be needed to be placed on the secondary, since for example voltage doublers share the output voltage value across them both (Vo/2)? If it did turn out only the first secondary winding needed to be tuned, would it then be that the other secondaries would just have standard HV capacitors and wouldn't need to be trimmers?
So given what we have discussed in this thread, do you suppose the buck converter fed LCC is the best way forward topologically for the design I have outlined? You've definitely answered my question in regards to whether LCC or LLC is better for high output voltage, but do any other possible approaches spring to mind at all?
Cheers
MagicSmoker:
--- Quote from: state_of_flux on May 14, 2019, 12:03:01 pm ---
--- Quote ---re: flyback
--- End quote ---
I suppose then a design trade-off would be selecting the turns ratio as to reduce the parasitic capacitance while also ensuring the voltage blocking capabilities of the MOSFET switches in the primary and the diodes in the secondary don't become unreasonably large?
--- End quote ---
Yes, though it is a seesaw type of tradeoff: a higher turns ratio (from primary to secondary; ie - step-up) reduces the blocking voltage requirement of the switch (though see caveat below on constraints with the two-switch variant) but increases it for the diode, and vice versa.
With the two-switch variant, however, the switches never see a voltage higher than the incoming supply (plus two diode drops) and leakage spikes are non-existent (assuming the clamp diodes are sufficiently fast). This sounds great - and it is - but that means the turns ratio must be set such that the clamp diodes do not conduct during the period when the output diodes are conducting, only at the very beginning of switch turn-off, to return leakage energy back to the supply. In other words, the voltage reflected back to the primary when the secondary is conducting must be less than supply voltage, which means a higher minimum turns ratio. For example, with a 335V nominal supply and a 10kV maximum output the turns ratio must be at least 30. That sets the minimum turns ratio, but what sets the maximum? Well, the higher the turns ratio the higher the voltage the output diodes have to withstand during the switch on time - that is a clear disadvantage of this configuration with HV outputs as now the output diodes have to withstand at least double the output voltage. Also, a higher turns ratio results in a shorter flyback period, which means a higher peak current in the secondary diodes, so just make sure this peak current is acceptable. Again, this is something which is unlikely to be a deal-breaker with a low voltage output, but high speed diodes rated for >10kV and more than a few hundred mA are few and far between. In fact, the only one I can think of off the top of my head is the somewhat antiquated (likely obsolete) 2CL2FM which (IIRC) is rated for 100mA to 200mA average and 20kV reverse voltage with a 100ns reverse recovery time. Finally, the higher the high turns ratio (in either direction) the higher the leakage inductance, usually, (and the higher the losses from proximity effect, which, fortunately, does not apply to the flyback). While the clamp diodes in the two-switch variant do return all the energy stored in the leakage inductance back to the supply, while they are doing so the output diodes can't deliver energy to the load... So, a lower turns ratio secondary feeding a half-wave doubler/tripler/quadrupler could be highly advantageous here for that reason, as well.
And if you put all of these conflicting issues together you start to see why I keep recommending a flyback with a voltage multiplier, despite it not appearing to be the best choice at first glance. Yes, a properly designed (and protected) LCC or other resonant converter would achieve a higher efficiency and likely take up less space, but forget this being the deep end of the pool; at that point you are swimming with the sharks out in open blue water. If you have sufficient dedication and budget (in both time and money) to dick around with this, then by all means attempt a resonant converter design - it will be better for the job - but if the real world has placed practical restrictions on you then, no, none of the exotic topologies are really appropriate. Note that I am not saying you can't or won't be able to get a more exotic topology working - who am I to judge what you are capable of? - I'm just saying that it will be a much more difficult task, relatively speaking.
--- Quote from: state_of_flux on May 14, 2019, 12:03:01 pm ---But aren't IGBT's only capable of low switching voltages, say, 30kHz? Would that not mean a significantly larger transformer size, or is there a point in which an increase in frequency no longer has a significant impact on size?
The primary design criteria for me is to ensure a small, lightweight solution.
--- End quote ---
HV and "small, lightweight" don't really go together. Minimizing the size requires vacuum impregnation and potting to prevent failure from corona or outright arcing, while minimizing the weight requires avoiding potting and relying on generous spacing between components to achieve the withstand voltage.
And you won't be running nearly as high as a switching frequency as you appear to think is possible - after all, a mere 10pF of stray capacitance has an impedance of 160k at 100kHz, so a 10kV square wave applied across this will result in 63mA of current. Whether that current is flowing across the junction of a diode that is supposed to be off, or across a transformer secondary, it amounts to a pure loss...
--- Quote from: state_of_flux on May 14, 2019, 12:03:01 pm ---So, essentially, incorporating some kind of lossless snubber to the buck-converter for turn-on will allow one to reach a high efficiency solution despite using a hard-switched full-bridge as the main regulator to the HV transformer?
--- End quote ---
The bridge in a buck current-fed converter doesn't really operate under hard-switched conditions - the overlapping on time ensures that turn-on is at 0V, and the buck inductor limits current rise at that time, too. The only real downside is that the last transistor to turn off in each leg experiences an overvoltage spike, but that can be tamed with a snubber or a clamp connected across the buck switch(es) and inductor.
--- Quote from: state_of_flux on May 14, 2019, 12:03:01 pm ---Just so that I understand - what you are saying is that uni-polar topologies can reach similar power levels to bi-polar counterparts just by incorporating a second switch - while reducing the iron losses due to reduced flux swing (since they only occupy one quadrant of the hysteresis loop)?
--- End quote ---
Sorta not really. I was mainly saying that there is not such a huge advantage in power output between a bridge and a forward converter once the switching frequency is high enough and the forward has a switch with twice the average current rating as any single switch in a bridge. What is "high enough" for the switching frequency? Well, that very much depends on a number of factors, obviously, but in my experience, starting around 50kHz and 400W of power the total flux swing ends up being limited to the same value regardless of whether operation is confined to one quadrant of the B-H curve or two. A bridge converter will still deliver more power from a given core than a forward (and much more than a flyback), it's just that the real advantage of the bridge is splitting the power up among more devices (and being able to use full wave rectification on the secondary). Consequently, I would be more inclined to go with a two-switch forward even as high as 600W, but as mentioned earlier, this applies mainly for low output voltages, as for HV outputs the flyback starts to look better, despite the higher peak currents, owing to more flexibility in transformer turns ratio and no need to worry about proximity losses (which to minimize requires interleaving the primary and secondary windings).
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version