Designing circuits is the fun part of engineering. And, as has been often noted here on Planet Analog, finding solutions to problems is what keeps an engineer employed. As an engineer, it is necessary to work with the parts available to meet design criteria. This sometimes necessitates using a part for a purpose for which it was not intended.
I just completed a design project that was powered by only a 1.5 Volt AA cell. Due to space constraints, a small switch was used to apply power from the cell to a load (shown in the schematic as resistor RL) and to turn on a blue LED showing the load was powered. The initial design problem was to turn on the blue LED from a 1.5V single cell when the blue LED required 4V. This was not an issue, since there are several ICs that can easily perform the task, such as the Diodes Inc. ZXLD381. This is a nice SOT23, 3-pin device that operates from 0.9V to 10V. Combined with a small inductor, it is able to boost the input voltage up to 20V to drive a small string of LEDs up to 20mA — so it can work easily in this application:

Everything worked well until I found out that the load being turned on exceeded the current rating of the switch — switch rating 100mA, load current of 2+ amperes. The switch did not last very long, though it worked well for short durations. I measured the cold resistance of the load to be about 70mΩ. We decided that a transistor should be used to carry the current, and we could just use the switch to turn on the transistor. The issue then came down to what type of transistor is best for the application.
With the low supply voltage from the single cell, the first thought was using a bipolar switch. However, at low cell voltage of 0.9V, a typical 0.3V drop across the transistor is 1/3 of the available voltage being robbed from the load. This was not an option.
The next option was to use a low-resistance MOSFET. MOSFETs used in applications where the Vds (voltage, drain-to-source) is low can exhibit an Rds-on (drain-to-source on-resistance) of as low as 1mΩ, and they are available in an SO8 package. The problem arises with the required gate voltage to obtain the low Vds, typically at least 4.5V. After some time thinking about the problem, I realized the solution was very simple and at hand.
U2, in the above schematic is a boost circuit that in normal operation provides a mostly constant current to the blue LED. The IC will boost the voltage to whatever is necessary (within its voltage compliance limits) to pass the required current through the LED. In this case, that's around 4V. With minor modifications to the circuit, we get the following:

I use a small Zener diode in series with the blue LED to ensure the ZXLD381 will generate more than 4.5V. I used two FETs in parallel to get very low on resistance. Adding the gate resistors is not an absolute necessity, but the 0603 resistors are small, and they reduce the likelihood of unwanted oscillations in the MOSFETs. The 100k resistor is used to ensure that the MOSFETs are turned off when there is an absence of gate voltage. The current robbed by the MOSFET is virtually nonexistent. The current draw of the 100k resistor is miniscule compared to the draw of the LED and RL. In this final schematic, I have solved the main problem by allowing no high current to flow through the switch, plus, with the 0.9mΩ Rds MOSFETs, the power loss is kept to a minimum.
This was a good example where the datasheet for the ZXLD381 talked all about driving LEDs, yet nothing about turning on a MOSFET. What design problems have you encountered that required a part used in a function that was not envisioned?
Did I miss something, or does the circuit, as shown, simply short the battery through the load resistor, RL?
Clever use of bootstrapping for the necessary gate drive voltage, though.
Yes it is.
Now, the load current doesnt go through the switch, thus
not destroying the switch, only the led current does.
The fets switch the load instead of the switch.
Schematics are correct since they were taken from the functional schematics that were used for board design at the time. Second one is moving to production.
I am sure they did not see this use of the LED driver at the time. This is why I used it as an example. I have another one a bit more complex that I will share in a future date. It involves a Linear Tech part in a power supply for power-factor correction, or in proper terms, an ac-dc supply that has a constant resistance input impedance.
Ah ha! I misunderstood the premise of the design. Thanks for clarification, and again, that's clever use of the boost switcher.
I saw an application, where the internal charge pump circuitry of a 5 V powered MAX234 transceiver, were used for driving 4 mosfets (10 V gate voltage), instead of standard RS-232C serial comm.
I encountered such a problem few years back while replacing an incandescent lamp with a LED lamp. The customer has monitoring circuits for the lamp status that are not supposed to be modified. We solved the issue with a digital approach by using a small micro-controller. Would like to know how you addressed the issue.
The basic circuit is exactly a Joule Thief . I have worked on this device with same basic circuit for driving 1W white HBLED using single AA cell.
Similarly I LM317 can also be used out of its basic application as a constant current driver for LED driving.
Here is a practical implementation of the jewel thief. Quite innovative.
Ya that is where I landed, when I first time encounterd Joule Thief. I need to try it our, but haven't got enough time to try it out.
Yep – some simple boost circuits there for sucking the last bit of energy out of a small cell. Not an especially sophisticated boost switcher. Poor input regulation, poor output regulation, poor regulation with respect to ambient temperature – but so what – good enough for the application.