Towards an improved flash

by Gil Green

Reproduced from in focus 57 (Sept.. 1996)

The worst thing that can happen to an underwater flash unit, apart from gross mechanical damage, is an inleakage of water. If the inleakage is of fresh water there is some hope of recovery after careful drying. If salt water is involved the problem is much more serious. Salt water is an electrolyte, ie it can conduct electricity by virtue of chlorine (-) and sodium (+) ions and when dissimilar metals are present an electric current will flow resulting in an electroplating process which will wreck a printed circuit board and any components that are not watertight. The presence of batteries in the circuit will greatly accelerate the process of destruction. Above all electronic flash units must not leak.

Where and how do leaks occur? Almost invariably at demountable joints and rotating shaft seals, both of which rely upon '0' rings to exclude water. Demountable joints are regularly broken and re-made to allow batteries to be changed or removed for charging and it is after this operation, not always carried out under ideal conditions, that leaks are most likely to occur. Rotary shaft seals allow internal switches (ON/ OFF, TTL, etc) to be operated often under water and there is always the possibility of failure due to deterioration of '0' - rings or the lubricant necessary for smooth rotation.

The writer has modified an Oceanic 2000 flash unit (rebuilt after flooding but now using a Cobra CM 20 flash circuit) in an attempt to eliminate these causes of in-leakage. This has been done by sealing Ni/Cd rechargeable batteries permanently into the unit.

Charging the batteries is accomplished by connecting an external charger to two 1.5 mm diameter stainless steel electrodes sealed through the body of the unit and connected to the batteries via a diode which allows charging current to flow but prevents discharge through the water when submerged. In order to permit external monitoring of the batteries the diode can be bypassed by an internally mounted reed switch actuated by a small hand-held magnet (Note It). ON/ OFF switching is accomplished by a second internally mounted reed switch actuated by a sliding bar magnet mounted on the outer casing (see photograph).

It is important to know something about reed switches. In spite of their proven reliability when operated within the manufacturer's specification they do not have a very good reputation in underwater equipment and the writer believes this may be due to lack of understanding by designers, as to what takes place during the first few milliseconds following switching on flash units (inductive loads) or even torches (resistive loads, Note 2*).

A reed switch consists of two metal "reeds" of rhodium plated ferromagnetic material mounted in a cylindrical glass envelope containing a inert gas (argon) or a reducing gas (hydrogen) or a mixture of the two. In the presence of a magnetic field from, say, a small bar magnet the reeds are mutually attracted thus closing the switch. Compared to mechanical switches of comparable size, reed switches are poor performers when it comes to switching large currents. The initial current surge on an electronic flash is quite likely to exceed five amps, it needs an oscilloscope to observe it! Because the load is inductive breaking the circuit can give rise to a similar problem.

In the case of the Oceanic the problem was overcome using the circuit illustrated. The two NPN transistors constitute a "Darlington pair" and are available on a single chip (price about 80p). The emitter current of the first transistor is the base current of thesecond transistor, if the current gain of each transistor is, say, 100 then the current gain of the pair will be 100 x 100 = 10,000. The current through the reed switch is limited by the 2.2 k resistor.

*Note 1 During discharge the onload voltage of a fully charge Ni/Cd cell fails approximately linearly from 1.25 v to 1.20 v (discharge at the standard one hour rate) before decreasing rapidly. For a battery of four cells this represents a drop of 0.2 v in 5 v. It is easy to measure this fall accurately, with a sensitive voltmeter by "backing offirnost of the voltage with a stable supply, say, from an alkaline battery source stabilised by a zener diode. A better alternative is to incorporate a multiple LED display into the flash housing which would indicate the state of charge continuously when the unit is switched on. This would eliminate the need for the reed switch across the charging diode. Lack of space precluded this being done with the Oceanic.

*Note 2 The current taken by a torch bulb is dependent on the resistance of the filament. At the moment of switching on the resistance is low and the current higher than it is a few hundred milliseconds later when the torch is at its brightest. Measurements on a MINI Q40 diving torch showed that the "switch on" current is three times that of the running current.

Though the intention is to produce a non-leaking flash it seemed desirable to fit a sensitive leak detector capable of detecting traces of water too small to immediately affect operation of the circuit. Concentric rings of bare wire, mutually insulated and closely spaced were attached to the inner circumference of the housing. Initially a sensitive circuit was arranged to switch an ultrabright LED and a piezo sounder if a conducting path was established between the rings. Tests revealed the difficulty of establishing satisfactory acoustic coupling between the sounder and the body of the unit and the sounder was replaced with a small electromechanical buzzer. The buzzer had a much higher current demand which required a further transistor being added to the sensing circuit.

There is no reason why most flash units should not be modified to allow external charging without altering the other switching arrangements. The work took about half an hour in the case of a MORRIS AQUA Fill. Substitution of Ni/Cd cells for alkaline does, however, reduce the number of flashes from 130 to 40 (manufacturer's data). In the case of a simple non - TTL circuit the substitution would reduce the energy of the flash to about 60% of the original (less than one stop). In the case of the low power Oceanic this seemed too great a sacrifice but by using five smaller (2/3 Af size) Ni/Cd's in place of four alkaline cells the full power was retained.

Quite apart from the greater reliability of the permanently sealed unit the writer would like to claim another advantage - that it is a time-saver. Anyone who has taken part in survey expeditions knows the pressure on individual members, writing up results, filling cylinders, cooking, processing films, maintaining equipment, especially photographic equipment, where an error can be costly. A flash unit requiring no mechanical maintenance must surely be an advantage.