Power supply for water electrolysis keeps blowing up. Please help.

SgtWookie

Joined Jul 17, 2007
22,230
You COULD use a number of 12v incandescent lamps in parallel as a limiting resistor.

That way you wouldn't be wasting the power - you'd be able to see what you're doing ;)

Automotive headlamps might be the ticket for you - you wouldn't need anything fancy, and you could easily tell when they burned out.

You could pick up a bunch of them from a junkyard - complete with convenient wiring harnesses attached; just clip 'em off.
 

beenthere

Joined Apr 20, 2004
15,819
A bit of caution on the headlamps, or any incandescent lamps. They take a while to warm up to full resistance, and can allow a big slug of current through at turn on. The fixed resistor idea is better.

Mineral oil will work fine to help dissipate heat. Remember that dropping the duty cycle does not limit peak current, just the average. Unless the drive to the FET is optimized, the device dissipation is increased while the FET goes through the ohmic region between cutoff and saturation. Guess the FET could go in the oil, too.
 

Thread Starter

H2OWiz

Joined Dec 12, 2007
8
I built a new board with out the LM317. You will notice on the schematic there are 2 lines to the #8 pins of the 555s. I plan on installing .1 caps here for filters but have none available at this time. I set the pulse width to 800 ohms which on an ohm meter read 5.32 volts which gave 12 amps out through the electrolyzer. However in a few seconds I noticed the current going down slightly the a few seconds latter going up and the the filter cap blew out. I have tried this twice. Is my filter cap to small ? I am including a snapshot of the gate voltage pulse at line 3 on the pulse width 555 taken after the 10 ohm resistor. I am not able to take this reading with the MOSFET charged at this time as my computer is in the house and the car with the power source is outside. Can someone help me understand how to read the frequency of the snap shot. I understand the time base but not how to read the screen. Is each square 50 µs ? I am also including a snap shot of the pulse width #8 line. The vcc line. I looked at the screen and tried different timebases but the wave form did not change. Again this is without the MOSFET being charged. Please help. I like the resistance idea to lmit the current because it is simple however a car alternator is limited in power and 20 amps for the electrolyzer plus the power wasted with resistance and this leaves very little left for the cars electrical system.
 

Attachments

gootee

Joined Apr 24, 2007
447
Hi H2OWiz,

Well, at least the MOSFET didn't blow, this time.

To read the frequency from a scope display, you count the number of major graticule divisions that are spanned by one complete cycle of a periodic waveform and mutiply that by the time/div setting of the timebase, to get the period (call it T), and then the frequency is f = 1/T.

Your waveform looks like it is 3.7 divisions long. Each division is 50 us. That gives a period of 3.7 div x 50 us/div = 185 us. So the frequency is 1/.000185 = 5400 Hz = 5.4 kHz.

I don't know what kind of filter capacitor you are using. It looks like it's 220 uF. So it's probably a polarized aluminum electrolytic type. BUT, what is its voltage rating? You will want to use a capacitor with a voltage rating that is higher than the battery voltage; quite a bit higher wouldn't hurt. A larger uF value might be good to try, as well.

Also, note that you _must_ connect an electrolytic capacitor with the proper polarity/orientation, or else it might... (you know). The radial through-hole types (with both leads from same end) usually have a white stripe or something, with minus signs marked in the stripe, along the side of the body, on the side where the negative lead is located. But its polarity should be indicated on it, somehow, in any case. You want the + side of the cap to go to the + side of the battery and the - side of the cap to go to the - side of the battery.

It would probably be helpful if we could see a snapshot of your actual circuit and setup.

- Tom Gootee

http://www.fullnet.com/~tomg/index.html
 

gootee

Joined Apr 24, 2007
447
Hi H2OWiz,

I atempted to simulate your setup, using LTspice (an excellent free download from http://www.linear.com ), which was more-or-less uneventful until I added the inductance and resistance of the connecting cables and wires. I just used "ballpark" estimates for the wire lengths and impedances, assuming about 4 feet of cabling between the battery and the mosfet/electrolyzer setup, and one foot of wiring between the battery and the circuit board, and a couple of feet of wire from the circuit board to the mosfet's gate. I tried doubling and halving most of the wire lengths and got similar results. And I adjusted the potentiometers to get a gate waveform similar to yours, in frequency, duty cycle, and amplitude. I have a couple of recommendations, based on that simulation. But, as usual, "your mileage may vary".

First, I would connect a 10 Ohm resistor in series with a 0.022 uF or 0.033 uF film capacitor (i.e. polyester or polypropylene), from the mosfet gate pin to the mosfet source pin, making the connections as short as possible. Without this small Zobel network, there might be overshoots and high-frequency ringing of the gate-to-source voltage at turn-off and turn-on, which might cause Vgs to exceed 20v and might also cause the voltage between the mosfet's source and drain to exceed 55V. (Without the Zobel network, both of those voltages went all the way to their absolute max ratings, during the simulation.) The high-frequency ringing would also contribute to the mosfet's heating. There might be a better way to suppress those transients. But that seems like it should work well-enough, and not slow down the gate voltage's rise and fall times, significantly.

I would also try increasing the value of 10 Ohm resistor that connects the NE555 output to the mosfet's gate. You could probably go as high as at least 47 Ohms. (After connecting the Zobel network decribed in the paragraph above,) I would check 10 Ohms, 22 Ohms, 33 Ohms, and 47 Ohms, and see which one makes the mosfet run the coolest.

Also, for bypassing on your circuit board: Very near each NE555 Vcc pin, I would have a 0.1uF or 0.22 uF ceramic or polyester capacitor to ground, in parallel with a 10 uF aluminum electrolytic capacitor, and also in parallel with a 2.2 uF polyester capacitor. AND, near where the positive supply wire connects to the circuit board, I would have a 1 Ohm resistor in series with a 2.2 uF polyester capacitor to ground, and also a 220 uF aluminum electrolytic capacitor from the positive supply to ground (in addition to the 220 uF that's already across the battery terminals).

You can use almost any type of polyester (or polyproplylene, if you want) capacitors you like (rated for 50v or more). But, if you want some very small pcb-mountable ones, mouser.com carries the AVX BQ-series polyester "box" caps, which have 0.2-inch lead-spacing.

Please keep us informed of your progress.

- Tom Gootee

http://www.fullnet.com/~tomg/index.html
 
Top