Kaizer half bridge driver

[donnersm]

I though i already posted this last night but i dont see it now.
So sorry is it's posted double.
Anyway.. I was looking at the kaizer half bridge and i wonder if anyone can tell me how to calculate c4 'n c7.

[Mads Barnkob]

You need something in the range of 1-10 uF, low ESR and low dissipation factor, so that pretty much says MKP (polypropylene) capacitors, derate the voltage 50% as the mid-point voltage can vary according to load changes. MKP capacitors also have the required frequency and current ratings you need here.

The higher sudden load changes and the higher overall power consumption of your inverter, you need more capacitance to avoid sagging.

If you look at different MOSFET/IGBT manufacturers application notes for inverters/converters using half-bridges, they usually use something in the range of 1-20 uF.

The voltage splitter capacitors is however not necessary, they are perfect if you want to add a voltage-doubler, but you can also just remove them both and tie midpoint end of the primary coil to ground instead.

I can not give you any math to calculate it, maybe someone else can chime in on that

[profdc9]

I will venture an answer...

I think the answer to this is perhaps a little complicated, because it depends on the ESR of the bus capacitors and the inductance of the connection of those capacitors to the half bridge.    Ideally, you would want the impedance of the snubber (also called dc-link) capacitors to be much less than the total impedance of the inductance of the connection and the ESR of the bus capacitors at the frequency of use.  Of course, the snubbers themselves have their own ESR and inductance of the connector to the MOSFET/IGBT sources/emitters and drains/collectors, which adds to the impedance of the snubber, which is why bridge layout is important.  The inductance of the connection from the capacitors (both the bus capacitors and snubbers) could cause voltage spikes under transient changes in current that cause the drain-source/collector-emitter voltage to exceed safe operating area and failure of the device.  This is why short and fat connections between these capacitors and the transistors is necessary, and for example why some snubbers screw right onto the terminals of IGBT bricks.

Another criterion that could be used is to reduce the voltage transients on the MOSFET/IGBTs under a certain amount, which would basically be then the average current supplied to the transistors over a cycle multiplied by the impedance of the capacitor.   The tolerable amount of voltage change is partially going to be determined by how closely you operate your device bus voltage compared to the maximum safe operating area voltage.   If you are operating, for example, with a bus voltage from a doubled single-phase 220VAC supply, you will be operating perilously close to the limits of a 650 V transistor.  Operating with a lower bus voltage, however, can also stress devices because of the increased current for a given power, so you have to consider that when choosing the MOSFET/IGBTs.

So while that is a complicated answer to your question, the bottom line is that if you err on the side of making the capacitors larger and the connections short and fat, you are likely to have fewer problems.  I have an example of a half-bridge PCB I designed (you can get the plans from https://github.com/profdc9/DRSSTC-PCB-Pack/tree/master/half-bridge-transistor if you want to make the PCB).  Like Mads said, usually 2 to 4 uF total snubber capacitance is good enough for most applications, and probably the best estimator on the total snubber capacitance is given by the circulating power (voltage X current) in the primary circuit, as in general the capacitance needed will tend to scale with that. But a combination of good bridge layout, adequate snubber capacitance, and not operating too close to the SOA voltage of the devices is likely to work.

Dan

[profdc9]

There's one other thing I realized too about C4 and C7...

C4 and C7 provide both snubber behavior (stabilizing the voltage at the MOSFET/IGBTs).  But because they also split the supply for driving one end of the half-bridge, there is another consideration when driving a tesla coil.  When driving an inductive load like a tesla coil primary,  you have a series resonance with L=tesla coil primary inductance and C=C4+C7.   With the added series capacitance C5, the resonance frequency given by L and Ctotal = 1/(1/(C4+C7)+1/C5).   So that the voltage drop of this LC series circuit is mostly across C5, C4 and C7 must be much greater than C5 so that the LC resonance is dominated by C5.  Otherwise, the voltage drop across C4/C7 approaches the supply voltage and may exceed it, causing catastrophic failure either due to excessive primary current or overvoltage on the transistors.  Of course the voltage on your Tesla primary coil and C5 depends on the Q of the primary, and so the exact capacitance for C4+C7 needed to keep the voltage swing on those capacitors under a certain amount is going to depend on the Q also.   Again because C5 is generally going to scale with coil power, as the frequency of the primary tends to drop with increasing power, so to will C4 and C7 increase as well.

Not to ramble on, but I made a half bridge to drive a hefty ferrite transformer and used about C4+C7=16 uF, but I was driving the transformer at 25 kHz, so a larger capacitance was needed.  In addition, I needed snubbers across the emitter/collectors and load itself as well as the power supply rails, as the inductive kickback of the transformer would blow out the IGBTs as well, because there is a lot of energy stored in the magnetic field of a ferrite with 6 square centimeters of cross-section and a permeability of 1500 as well as the gap in the transformer magnetic circuit.  The kickback occurs when you have a spark gap and the current is cut suddenly by quenching of the spark.  But maybe this given you a rough idea of what you need, because C4+C7 is going to scale inversely with the square root of frequency.

Dan

[donnersm]

Thanks Mads and Dan

[Mads Barnkob]

This question about the voltage splitter / DC blocking capacitors, led me to find an unfinished article I started long ago, so I finished it up and hopefully that answers a few questions and I would like to get feedback on it: http://kaizerpowerelectronics.dk/tesla-coils/sstc-design-guide/