It seems pretty clear from those posts that BroMikey already "knows" all he is ever going to know about mosfets and cap dump circuits.

Well I decided I had some time so I set up a dump circuit using one mosfet from my LF power control board and dumped 26 volts into a 12 volt battery from a 45,000 uf cap as shown. The smoothing bank should have been at least three times larger.

I got about 13 amps peak current and the current sense resistor at 0.8 ohms got real hot but my mosfet stayed cool  to touch.

I'll put some shots here and the circuit for those who cannot log in over there.

Circuit
then applied voltage in blue and current in yellow across the 0.8 Ohms.

then The voltage on the two cap banks with the inductor between them

Second shot shows gentle waves but the first shot shows sudden violence. 

Oh and a rise time shot. Slow mosfet.

The 14 Amp hour motorcycle battery I'm charging could do with the experience of some decent current it hasn't been used to start the bike for a long time.

Took out the 0.8 Ohm resistor and all is going well, mosfet cool. Battery voltage bouncing and increasing. 

Mm-hmmm.

The peak current shown is a bit over 13 amps, I estimated 13.7 or so. But the average current during the pulse seems to be about 3 or 4 amps, and taking the 33 percent duty cycle into account that means that the average power dissipation  I²R in the mosfet itself is ((4 amps)² x 0.01 ohms) x 0.33 = a bit over 50 milliWatts. And since it's being switched cleanly by a good squarish pulse of 12 volts from the gate driver, it's not adding much switching loss to that. So no wonder it stays cool!

If you now make the capacitors bigger (more capacitance) and charge to the same voltage and use the same duty cycle, you will get higher average current (the exponential decay time constant will be longer) and you will get more energy transferred into the battery per pulse. The _peak_ current will still be the same.

Well all I need to do is to program the picaxe chip to pulse with narrower pulses and faster and then I get a squarish current wave form if I go narrow enough. Then the capacitor is only partially discharged so current flows the whole time if the pulses are narrow enough.

Anyway I think bromikey was under the impression his caps were fully discharging, and he may have a slow turn off. The residual current flow I think is through the inductor anyway. If I was doing it my way I would use a smaller capacitor and pulse at some kHz. I think with his resistor there is always residual current flowing no matter how long his discharge time is, and if his mosfet is not turned off fast he will heat it up.

If I used smaller caps and faster narrower pulses then I would not need to worry about using an inductor even, I only did that to get the dump cap voltage down at the end of the dump. A MOT primary would work better with 250 mH inductance and thick wire.
A low loss current limiter for a moment.

SeaMonkey I have only 100 nF caps across my driver chips on that board, I need to add the 10 uF caps as well yet. The driver supply is regulated to 11.95 volts so the chips need more reserve energy.

The coil I used had way too much resistance and not enough inductance and the caps were too big to get a voltage rise from the resonant charging circuit. After an hour or so the coil was so hot I could not pick it up, I waited 30 minutes so I could pick it up and used it for a pocket warmer, it's a bit cold here now.

I charge my batteries with a 6 amp dual output solar system, I only pulse batteries to keep them in good condition or rejuvenate them. And never excessively. I treat them usually with only 1000 uF discharges from about 22 volts and at random frequency the cap discharges when the desired voltage is reached and pulses to keep 17-18 volts on the panels, max power voltage.

All my batteries are in good condition and are kept charged and used, with the exception of the motorcycle battery, it doesn't get much use, but I keep it charged.  Motorcycle batteries fail at the drop of a hat, they are too compact to be long lasting.

..

Here's a couple more shots one at 400 Hz 10% duty and one at lower frequency, at 400 Hz it pulled the dump cap voltage down due to the MOT primary's 260 mH inductance. I should have tried 200 Hz, at 400 Hz it was pulling 90 Watts out of the wall without the CSR and driving the battery voltage up like mad but also warming up the mosfet just a bit, not hot just warm. At the lowest setting it pulls a fluctuating 20 to 35 Watts and makes the voltage bounce.

Haha at 400 Hz it punching rectangles of current into the battery at a rate of over 10 amps a shot for 260 uS.

I think rather than paralleling mosfets a better way might be to use ie. four mosfets switched in turn so that each one has one 1/4 of the total work to do. Time to cool.

..

Well I guess at 10 Amps @ 10% duty it is giving a 1 Amp charge to the battery.

OK with about 250 mH inductance in the coil and if I reduce the output caps to 15000 uF then I should get resonance at about 2.5 Hz.

So then I think I can switch at double that frequency to still almost double the voltage in the dump cap after switching.

And with the diode the voltage stays and no reversal so anything between 2 Hz and 5 or 6 Hz should work well with the charging circuit.

If so the 15,000 uF cap will end up with almost 50 volts on it. I'll try it out while I'm going, mosfets are only rated to 55 volts so I might lose it.

Ahh the joy. 
..

I hear youse all, MileHigh, I already had one such setup dealing with solar input to a system that is often under what most charge controllers deal well with, it sensed the input voltage, the output cap dump voltage and the battery voltage when the mosfet was conducting with the same voltage divider that sensed the output dump cap voltage. It used a boost converter to boost the low input from the panels to 22 volts so as to apply some rejuvenating to the battery, and when the battery was charged it would go into float mode then kick back in when the batt voltage dropped. It is no use if the battery is badly sulfated though because it automatically goes into float mode if the voltage rises above 14.4 volts. I made it so it would float at 13.6 volts.

I'm only doing this for fun and to help BroMikey cast out the fallacies he hes been fed by "the crew" over there. One of the big problems is that the battery is drawing from the supply through his resistor, using a MOT primary I can dump 15 K uF in 40 mS and 350 mS between pulses to recharge the cap and get down to almost zero current flow, longer and the MOT primary starts to pass current directly from the supply to the battery, this is to be avoided.

The other way is to disconnect the  + rail of the supply from the dump section before dumping and I've already done that too.

Using a big low resistance coil and the right on time is simpler and less parts.

My coding is very basic self taught, I can only do so much without wasting too much time.

..

Now if it was me I would be inclined to go down to a 4000 uF dump capacitor and increase the frequency.

SeaMonkey, Talking of driver capacitors, this board I made mainly for controlling relays and small motors ect. and for LF experiments for ease of power control. So I ran out of space for another capacitor on each driver chip so I might just use a 100 uF cap across all of them and I can fit a 10 uF cap across the two with PWM outputs. They all have a 100 nF across them. I've seen several combinations used but usually they all contain at least one ceramic cap. Why is that ?

The small ceramic cap right at the chip is "AC bypass" that keeps the chip from responding to noise on the supply feed and helps to prevent false triggering. This is just about universal, especially if supply leads are long. Put the bypass cap as close as possible to the chip itself.

I've got four mosfet switch outputs on the "B" output side of the 14M2 picaxe chip (two have pwm ability), on B.5 pin I use the analogue to digital converter to sense the variable voltage provided by a voltage divider which is a big 5 K pot, I use the pot to set the program into different "stages" manually which I can make to do whatever. On the input "C" side of the chip I have them set up to take inputs or be outputs to other boards.

That is the same technique I use to control which subroutine is running in my NeoPixelRing demonstrator. I have eleven different independent program segments, each selectable by the 50 k 10-turn pot with turn-counting dial on the front panel. Select something between 100 and 200 on the knob for example and you get program segment 2, and while "2" is running I can even control a function by varying the pot within that range (one full turn of the 10-turn pot). Since the picaxe (or in my case Arduino Pro Mini) ADC input is sensing voltage and is a high impedance input, you can make the voltage divider pot just about any value, even 1 megohm. Making this voltage divider pot large will cut down on the overall current consumption. I use 50K because that's what I've got on hand.

I think I can have the board play a music tune when the battery is charged.  I wonder what funky tunes are available.


 

There are billions of bits of 8bit music out there. It takes a bit of fiddling to get accurate note values but sure, you can do that if you have the memory available once the "meat" of the program is in the chip. I don't know what is available for picaxe but for Arduino there are many many aftermarket "shields" that simply stack onto the main board, no soldering or wiring required, that will do all kinds of things. Music shields that incorporate microSD card slots are available that will play high-resolution music or any audio file you can put on an SD card.