This should get the creative minds going. Lines prefixed ~ are ~= part of requirement - rest is comment / fill / shopping list / dreaming ... . I can do this well enough already (probably) but am interested in comments / alternatives / stunning insights. Required: Either Hall switch, < 50 uA mean drain, Vdd min <= 0.9V, sample rate >= 10 Hz. Sensible cost ($US0.05 preferred, - $US0.20 max max) in volume NO Hall sensor that I have found meets these specs. OR Reed switch in same price range that is bulletproof / soldierproof* in real world portable apps * Soldierproof - Equipment can be thrown 10 metres onto back of truck or off moving truck 50 kph 10m sideways, 2m fall onto concrete -30C ambient. [[[ :-) ]]]. OR - *********** The probable real need ***************** Available Hall with specs as below, and a voltage booster Following relates to v booster - Hall switch is known available. ~ I require a voltage of 1.6V <= V <= 3V3. ~ Vin is 1 NimH cell so say down to 0.9V. ~Output load is a small capacitor charged to whatever voltage is achieved. ~Mean load current is 10 uA. ~Desired current from circuit proper is "as low as possible. ~A few tens of uA would be OK. 10 uA x 24 hours ~= 240 mAh / 0.25 mAh. For a say 800 mAh cell that's 1/3200 of cell capacity. OR 1/32 of 1% of cell capacity. ie when the cell has 1% capacity left this circuit would take ~ 3% of it per day per 10 uA. So if all up load was 50 uA it would need over 1% cell capacity per day. ~ Circuit will ideally stop working at under about 0.9V and draw about zero uA thereafter. This can be enforced by other circuitry if need be. Load current may peak briefly to "somewhat higher" (see below) but this can be accommodated by the capacitor. Application is to power a time multiplexed Hall switch which needs Vdd >= 1.6V. The actual detector sleeps most of the time and is woken occasionally to check for a magnetic field and then sleeps again. Versions vary but an eg 3 mA hall detector with a 300:1 mux ration will draw 10 uA.Outputs are latched. Main limitations are that the greater you make the mux ratio the longer the delay between sampling points. At 300:1 the sample rate is typically 10 Hz with a ~ 30 uS wakeup sample and resleep period. Good for eg person switching (keyboard, pushbutton, ...) applications but not suitable for high speed applications. Allegro make some nice examples but there are many others including quite a few Asian sourced ones. No prizes for guessing my application :-). The alternative is to use a reed switch, but they are more fragile with a number of failure modes to consider. ~ Objects: As above plus as cheap as possible to implement. Price includes component cost & assembly effort. PCB area less crucial as liable to be small enough to not affect available space or board cost greatly. A few cents would be good $US0.10 may be upper limit. At 0.9V in a voltage doubler with ideal diodes will yield 1.8V. One Schottky will eat about 0.3V yielding perhaps 2 x 0.9 - 0.3 = 1.5 V = not enough. A tripler may help - but diminishing returns apply. 2 x Schottky losses add 0.3V steps at 0.9V in. Inductor 2 to 3 cents. Transistors ~= 1 cent. Resistors ~ 0.1c. Tiddly caps < 1c. Big ecap 1 - 5 cents (prob 1) AN6601D 5 or 6c A circuit which needs Vin say > 1.2V and then runs continually is conceivable but not ideal. Starters for 10 points: 1. Discrete 2 transistor astable multivibrator. Classic or version of NPN/PNP pulser from long ago. Driving voltage multiplier. 2. One transistor self oscillating inductive load. Inductor dearer but OK. Iq tends to be high. Voltage cuttoff needed when Vout OK to keep Iqmean down. 3. 1. driving an inductive loaded.Tends to be more controlled. Low Iq easier. 4. IC able to oscillate from 0.9V startup. eg AN6601d Low Iq hard Cost pushing limits. Thoughts?. Insights? Flashes of genius? * Russell * Reed switch recommendations need to come with signed in blood affidavits. -- http://www.piclist.com PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist