BroMikey's Capacitor Dump Circuit

[SeaMonkey]

BroMikey has built a CapDump Circuit which is in need of
some "fine tuning" to get it working efficiently.  He's
begun a thread discussion at the "other forum" explaining
his problems and his thoughts on resolving them.

If anyone here is able to enter discussion at the "other
forum" perhaps you'd consider inviting BroMikey to come
over here to get some real top notch assistance.

The problems he's experiencing are very common among
experimenters who've not yet got up to speed on the
"care and feeding of MOSFETs" and how to make them
switch capacitive discharge currents safely and efficiently.

Many thanks to any who provide technical assistance.

[MarkE]

BroMikey has built a CapDump Circuit which is in need of
some "fine tuning" to get it working efficiently.  He's
begun a thread discussion at the "other forum" explaining
his problems and his thoughts on resolving them.

If anyone here is able to enter discussion at the "other
forum" perhaps you'd consider inviting BroMikey to come
over here to get some real top notch assistance.

The problems he's experiencing are very common among
experimenters who've not yet got up to speed on the
"care and feeding of MOSFETs" and how to make them
switch capacitive discharge currents safely and efficiently.

Many thanks to any who provide technical assistance.

Any capacitor to capacitor charge shuttling circuit will suffer losses for several reasons.  The one that you cannot avoid without using a resonant topology is the N*(X/N)² problem.  That problem basically says that you don't want much change in voltage between the two voltages in the shuttle.

Efficiency versus low side voltage / high side voltage at start of transfer for equal value capacitors:

0%  50% efficient
50% 90% efficient
80% 99% efficient
90% 99.7% efficient

So, keep the voltage above 80% on the low voltage side.  The other trick is to limit the surge current.  You can keep the MOSFETs cool by switching them reasonably fast and using an external impedance.  If for example you use an inductor, then you can realize a resonant transfer (this was the basis of the original now expired Vicor patents) and reduce the loss.  If pick a resistor that sets a time constant about 1/7th of the switching interval but that is significantly larger than the MOSFET resistance then it will absorb most of the energy that must be dissipated.  The current load that you can impose on the circuit is going to be limited by the switching frequency and the size of the caps, and the loss you are willing to suffer.  Or you can get clever and pick a piece of magnetics that will limit the initial current surge before saturating.

[SeaMonkey]

Aye, the capacitor discharge surge current can have
explosive power and it may be necessary in some
applications to limit the surge to a safe value in order
to protect the switching MOSFETs.

The capacitive discharge surge rises near instantaneously
to a peak which can be hundreds to thousands of Amperes.
This is why Tesla loved capacitors and disruptive discharge;
it enabled him to produce very brief pulses of incredibly
great "horsepower."

It is essential when using semiconductor switches for this
purpose to turn them fully "ON" in the shortest possible time.
The proven most reliable way to do this is to use a MOSFET
Driver Chip capable of providing the necessary Gate Charge
Current located as close as is practicable to the MOSFET it is
driving.  Where a bank of parallel connected MOSFETs are used
which are each some distance from their adjacent MOSFET, it
is best to place a Driver Chip at each individual MOSFET.  The
logic signal input to the driver chips which controls their ON
and OFF timing may by applied from a single chip or pulse
source.

If a resistor is used between the output of the MOSFET Driver
Chip and the MOSFET Gate, it should not be too large.  In most
cases less than 10 Ohms.  This resistor is ordinarily used only
when switching the MOSFET at high frequencies in order to
relieve the Gate Driver Chip of excessive power dissipation
and excessive heating.  The Gate Drive Current at high frequencies
can be surprisingly high.

I've not yet seen the schematic diagrams BroMikey has made
up of his circuitry.  It would be appreciated if someone were
able to post them here for all to see and evaluate.

Commercial capacitive discharge welding circuits often use
a Silicon Controlled Rectifier (SCR) as the switching device.
The SCR is much easier to drive than a MOSFET and some are
able to switch thousands of Amperes in very short pulses.

[MarkE]

Yes, and no.  There are a couple of considerations:  An unintentional resonant circuit will cause current to pass back and forth between the capacitors multiple times, subjecting the MOSFETs to extra heating.  You want to avoid that.  This is where either turning the MOSFETs on in a controlled manner can actually reduce total heating by preventing the circuit from passing current between the capacitors over multiple oscillations.  The right amount of resistance either effected through the MOSFETs or applied externally can prevent a lot of loss.

An intentionally designed resonant circuit needs to have a mechanism to cut the current off when the capacitor voltages first match.  In that case you want a driver that can turn the MOSFETs off very fast.

[SeaMonkey]

You may have a point MarkE.  I'm not yet clear on how
BroMikey intends to make use of his Capacitor Dump
Circuit. Judging from the size of his capacitor bank and
the layout of his switching circuit from photos he's
posted at the "other forum" it appears that it is
probably for very low to low frequency pulsing.  It rather
reminds me of the 30 Volt banks which were/are part of
impulse spot welding devices made for joining small parts
with a single pulse.