Caving Headlamp – Switches
By Ian Mander 28-31 October 2010, updated 5 & 8 November 2010.
The choice of switching determines the number of modes available and what kind (and number) of drivers will be used. Unfortunately, affordable waterproof housings of suitable size and weight are sufficiently rare that the switching needs to be chosen to fit in the housing, rather than being able to choose a housing to fit the volume of the switching and cable seals. To get the desired current on turbo mode means using drivers that have a 1.5 A maximum output, but the switching that can be used limits the driver choice for the standard modes to the same drivers. The switching required to use different drivers is too complex, too expensive, and too bulky.
Mode switching can be most easily accomplished using 3-mode pushbutton switches, which are small, inexpensive, and reasonably robust and apparently of acceptable quality. To get the most from each 3-mode switch it will be most effective to use it just for switching the set resistance for a particular driver rather than to switch the current between multiple drivers, each with its own hardwired set resistor. This will give four modes using a single driver for each LED bank.
This is not a standard arrangement because contact resistance in the switches can create issues with the set resistance seen by the driver. It's more likely to be a problem with the higher current modes, for which a lower set resistance is needed.
I believe the 3-mode switches are able to cope with 0.5 A (and maybe more), so they should be fine used for all the standard modes, contact resistance notwithstanding. The switches should be kept close to the drivers.
Using the switch to switch out series resistors one by one (thereby increasing driver output with each click) would mean the LED bank would have to be off when the switch was off. Having every mode dependant on the lower value set resistors (for the higher current modes) would also be unnecessarily awkward for changing the mode configuration at a later date, as the whole lot would have to be changed.
Instead, parallel connected resistors is used. A "common" resistor for the lowest current mode, when the switch is off, is hardwired to the driver. That resistor has the highest value of any used. Each of the other three modes will switch in a different resistor in parallel with the common resistor. Using separate resistors (actually separate parallel combinations of resistors, to get the right values) for each of the three higher modes will allow the mode configuration to be most easily changed at a later date.
If a switch fails open circuit (off) the corresponding LED bank will default to idle mode. Hopefully both switches won't fail on the same trip, but it's nice to know there's at least some redundancy in the design, as caving is possible on either LED bank.
The turbo mode has higher current demands, so needs separate switching and drivers. Separate buttons are to be used for wide beam turbo and spot beam turbo.
The 3-mode pushbutton switches used for the standard modes cannot be used to switch in a very low value resistor for the turbo mode, as the full LED current passes through the resistor (and thus the switch) and would involve a whopping 3.0 A current for each bank.
For each bank, at least a DPDT or a couple of SPDT momentary pushbuttons will be needed to provide a crossover so the standard modes will be deactivated when the turbo mode is activated. This will require either a high current switch or a low current switch with a relay.
A relay would be particularly cool, as there'd be a clicking sound when the button was pressed. It would allow just about any momentary button to do the job, as the full 3.0 A wouldn't need to be handled by the button. Unfortunately there is almost certainly not enough room in the housing for at least a couple of low voltage relays ($3.00 at Surplustronics; SPST, 13 gram) along with 4 LEDs and at least 4 drivers.
Even with two DT switches the drivers themselves will stay on, as the switches will be needed on either side of the LEDs. With the drivers on all the time the capacitor in parallel with the LED will charge up to a higher than desired voltage, so each LED will need to be protected from the spike when the drivers are changed with its own capacitor. (Will it really protect it? Won't it already be charged from being in use?)
Using high current switches, ignoring parasitic resistance each switch will handle at least 2.65 A, so each switch will need to be rated for 3 A or more. If the switch's contact resistance is 30 mΩ there'll be almost 0.08 V drop across the switch. That means 0.21 W dissipated just by the switch. This seems like a good reason to use momentary switches for turbo mode.
IP rated momentary switches are much more common than latching versions, but one may not be needed. Small SPDT micro switches are available which can handle the current - the best option seems to be a 5 A micro switch ($2.50 from Surplustronics). I'm sure the roller could be removed if not required. The micro switch could be used with a good old glow in the dark switch cap to provide an IP seal, would provide a quiet but hopefully satisfyingly audible click, and would leave room for the switch cap to be lit from inside by a UV LED or stray light.
I need a DPDT switch because positive power needs to be switched between the drivers, but also the set resistors need to be switched. They are both connected directly to ground on the other side from the LED, which means they cannot be left directly connected or they will affect the set resistance perceived by the driver. I hope it will be possible to attach two micro switches right next to each other with their arms soldered together. The switches are very small so I'm sure they'll fit, and solid wire goes a long way for bridging.
This may not be enough because the driver layout is quite different for the EQB8L and AX2002 circuits. The sense resistors are on different sides of the LED, for example. It would be easier to use AX2002 for all, especially since 20 mA is mentioned in the datasheet (as requiring a 12.5 Ω resistor). Some testing is called for.
To switch the drivers on and off as well as switching the LEDs to a different driver, a 3PDT switch will be required, but it's tricky finding a suitable one ($11.95 from McPherson Stompboxes, or $14.50 at Jaycar; it's latching and only rated for 2 A).
The smallest switch option is probably a sub-miniature micro switch rated for 3 A ($4.90 at Jaycar).
Using a power transistor for the main power switch is probably ruled out by the voltage drop that would be across it. An MTP3055E MOSFET I have on hand claims a typical on resistance of 0.1 Ω. That's a fair bit higher than a good physical switch (at 30 mΩ), and becomes a problem when running at high current. Buck drivers powering four XP-Gs at full current will draw at least 5.3 A from the battery (ignoring parasitic resistance), so the FET will have 0.53 V across it. But that extra voltage drop will mean the driver is supplied a lower voltage so needs to draw more current, so the driver input current climbs to over 6.1 A (that's worse than a linear regulator!) with the voltage drop across the FET 0.61V. Over 3.7 W of heat is dissipated by the FET - not a problem for the FET (especially if mounted on the inside of an aluminium headlamp enclosure) but that's not good for the efficiency of the headlamp - it's a whole 14% lost in the FET. To put that in perspective, it's enough power to drive an XP-G at more than 1.1 A; 380+ lumens.
So a FET switch is out.
It will be better to use a physical switch that has a high current rating. A pushbutton switch with an IP56 rating might do ($7.50 from Jaycar). It's rated for 6 A, DPST (does that mean 6 A each pole?), but is quite large; mounting hole 20 mm. It could be mounted on the battery housing to save space in the lamp housing, but that presents its own problems. I don't think the aluminium enclosure would have room for it with the battery holder, and it couldn't be mounted on the side of the Klip It container because of the container shape. That leaves the lid of the Klip It container, but that might necessitate bash protection simply to stop the switch being turned off (or on) accidentally.
A non-waterproof master switch could possibly be installed inside the battery housing. Before entering the cave the switch would be turned on and the battery housing sealed. With the master power switch fully inside the battery housing there is no need to have an off setting with external switches. This may be a bad call, but I'd rather have four modes.
It may be possible to just switch power to the chip enable pin of the drivers, although that might leave a small parasitic drain, depending on the driver. Probably too fiddly. Instead of a switch inside the battery housing, a simple 9V battery clip onto the 4xAA battery holder could be used. It would introduce a bit of contact resistance but any type of master power switch would do that anyway. It would mean accessing the interior of the battery pack, but an internal non-waterproof switch would require that anyway. It would allow for spare battery packs to be pre-packed with cells. A dummy protective clip on each unused battery pack would be needed when the real clip isn't clipped on.