Hydrogen burns with an invisible (ultraviolet only) flame.  You discover it by walking into it.  And any leak is virtually certain to be lit after a while.  NASA searches for hydrogen leaks by having a guy hold a piece of cardboard and walk slowly forward until it catches fire.

I was interested in the possibility of generating H2 to use like natural gas, as a remote location substitute for propane. I haven't built one, but here's the info I collected:

H2 is from negative cathode, oxygen is from positive Anode,
twice the H2 volume is produced compared to volume of O2

http://oupower.com/phpBB2/viewtopic.php?t=1398  
One Amp for 4 sq inches of plate, any more and there is an exponential increase in electroplating/erosion, 1 amp for 6-8 sq in is better for long plate life, but if quality stainless steel is available (instead of low-quality Chinese stainless), the higher amps will produce more HHO.

1.8-2.0 Volts per cell, any more and there is excessive heat and water vapor production... From my experiment, even 1.7V per cell, can hardly see bubbles flow out from cell. 1.8V and above the cell produces fair a mount of bubbles.

For electrolyte, use KOH 28% mix with distilled water 72%. Use variac and slowly increase the voltage from 0V to 14V while monitoring voltage & amp. Once the amp reaches 9A, check voltage per cell it will be less than 2V. You may check best gas output by playing with voltage. (this is for using a car alternator flowing through 14 stacked plates to make it compact, 7pos 7neg, in plastic tub, plastic plate separators)

http://www.oupower.com/phpBB2/viewtopic.php?t=1541
Most automobile electrical systems actually run on 13.3 V. 7 cells in series will use 1.9 V per cell. As far as the gaps or plate spacing, 1/8" is recommended. I have found that with my test using different spacings, that the smaller the spacing the weaker the electrolyte has to be in order to control the cells amperage draw on the alternator. the larger the gap the stronger the electrolyte can be. The gas is formed on the plates surface, and I found that the larger gap between cells has little to no noticeable difference in output to the same cell with smaller spacing. But the larger spacing has advantages of easier refill.

In this type of cell, the O2 and H2 are allowed to mix making hydroxy gas (HHO). If you use a membrane taken from the inside of an old lead acid battery (between the plates), you can keep the cell plate gas outputs separate, and divert the two gasses to separate columns. The membrane allows electrical flow, but doesnt allow gas to mix.

It is cheap to buy carbon arc welding rods. These work very well to produce hydrogen and they do not melt away on you. I bought a pair of 12" x 3/8" rods for $1.50 at a welding supply store (this is one option instead of plates for a simple two column electrolyzer)

I thought I could also use the oxygen with propane in a MAPP gas torch. It will cut steel, obviously not anywhere near as fast as oxy-acetylene, but if the O2 is free and the propane is cheap...

http://www.hydropowercar.com

Luckeydog

My plan was to compress the H2 into a propane tank at perhaps 100 PSI. Stove burners that are set-up with the proper air-mixing ratio for natural gas will work just fine with H2 (I'm told).

I think on-demand generation as an engine fuel supplement would work best with a constant RPM diesel generator with a governor. As gas supply varies, the amount of diesel fuel that is injected will be automatically throttled. An alternator from some 18-wheelers will have two pulley grooves. By adapting it into the stock location, the original engines alternator can be bracketed to accept a second belt (run by the first alt). This way the stock alternator and voltage regulator will function normally and the other alternator is free to supply the hydroxy cell.

The plastic tub (squarish?) should have 6 plastic dividers glued in to make seven thin cells with a couple small weep-holes near the bottom so the water levels stay the same.

Each cell should have two stainless steel plates (square inches depend on how many amps you wish to pump, bigger=more gas) If the plates are connected in series and fed 14 volts, then each plate-set will see about 2 volts. A plate-set will try to flow the voltage from the edges, so the edges should be covered with plastic, forcing the voltage to pass across the plate faces.

The plate sides that face each other should be rough-sanded (it helps, I don't know why). Since the plates should be kept as parallel as possible, and kept 1/8" apart, and edged in plastic, you might take finger-thick strips of plastic and cut two parallel slots on one side (1/8" apart) perhaps with a hacksaw blade, then 4 of these would make a "frame" for one plate-set.

The top of the tub will be the gas outlet, it should feed a "bubbler" similar to a radiator overflow catch-can. It is a container of convenient size with two openings on top. One outlet has a tube connecting the top of the tub to a tube stub that extends down to the bottom of the can (half-filled with water). The other outlet has a tube connected to the intake manifold.

This bubbler will act as a check valve and will also stabilize the moisture in the gas. When the engine is running, there will be a partial vacuum on top of the can water. Then, when the tub produces any gas, it will flow to the bottom of the water can, and "bubble up".

As the electrolyte water gasses away, you have to add distilled water (tap water has an inconsistent mineral content). One way is to have yet another container alongside the tub (the two water levels will stay equal). Make a float valve similar to what's in the back of a toilet tank. Connect the two bottoms with a tube and a check valve so the water can only flow one way.

An elevated distilled water store feeds the float valve, which keeps the "leveling container" at a water level above the plate tops in the tub.