This is Power question but not high power but low power one. To avoid XY problem I will start from the beginning:
I want to do Roomba virtual wall. The original idea was to power ATTiny with two AA batteries since the ATTiny can run on whole voltage range they will provide. The problem is with the IrLed. It has voltage drop around 1.5V and I want around 10 mA across it. If it should work at nearly depleted batteries @2V I need current limiting resistor of about 50 Ohm. But when the batteries are fresh the voltage is around 2.8V and the current across LED would be more than 20mA. It is waste of power and it is close to the maximum ratings of pin current.
I decided it would be much more power effective to have the circuit powered from regulated 2V. But I don't know how to achieve it. I would like to have extra low voltage drop - like 0.1V while most LDOs I have seen have voltage drops 1-2V. Moreover they are rated for "huge currents" - hundreds of mAmps or so. I will have peek current around 10 mA and (hopefully) average current under 1mA. I looked at LDOs from discrete parts but they all look quite complicated. I suspect that since my circuit will have known current draw, input voltage will be 2-3V and desired output fixed 2V it should be quite simple to make such "LDO". But I have no clue how to construct it/which IC I should use. Can you please give me some advice? Thanks
The IrLED will be simulating Roomba virtual wall. It means it will send IR signal with 38kHZ frequency (changing off and on state of the LED) in a fixed pattern (just like other IR transmitters). When the LED is on I expect current about 10mA will be needed to get brightness surely recognized by the robot even for larger distance. The LED is on about 1/10 of time. I will do some testing to get the right value but so far it seems 1mA is not enough while 20mA is more than sufficient.
I thought about using large capacitor and charge it by the MCU itself to get "fixed" voltage but it looks like a bit hazardous to use MCU to control it's own power source in such way not mentioning power consumption of the ADC.
An LDO regulator won't provide any benefit. If you're dropping from 3V to 2V that's still 1V wasted, LDO or super-LDO or non-LDO or whatever linear regulator you choose. In this respect it's a very common misconception that LDO regulators are somehow more efficient than non-LDO regulators; the only advantage that the LDO provides is that it works when the input voltage is closer to the required output voltage, and again that's typically a very small advantage.
The only way to improve efficiency is to use a switching regulator with the requisite inductor, cap, diode, etc. But even then the power consumption of a well-designed switching regulator is poor at very low currents and unlikely to provide as much value as you might expect.
To answer your question about finding an LDO regulator with a very low dropout, they certainly do exist and are pretty common. Using the search tools of any of the big electronics distributors you should be able to filter based on the dropout voltage to find available parts. Here's a listing of LDO regulators under 100mv dropout on Newark.
Chagrin:
An LDO regulator won't provide any benefit...the only advantage that the LDO provides is that it works when the input voltage is closer to the required output voltage, and again that's typically a very small advantage...
Well I imagine LDO as a resistor with such resistance that the output voltage is the required one. So I know it eats some of the power BUT there is still room for great savings. As I wrote before I want 10mA across the diode when the batteries are low (around 2V). If I use current limiting resistor it needs to be around 50 Ohm because voltage drop on LED is around 1.5V. But if the supply voltage is 2.7V the current over LED will be more than 20mA - dangerous for MCU and waste of battery capacity (it is clearly visible if you think in terms of mAH, not Watts). Moreover the MCU itself consumes more current when powered on higher voltage even with the same clock speed. So I expect an ideal LDO would save me some power (10-20%???) while the batteries are full and also removing timing issues caused by different frequency at different voltage.
I have found some LDOs that looks like what I need. But most of them are SMD - I have problems to solder THT parts, there is no chance to solder SMD properly. Most importantly our local shop never has the LDO I find on internet. So I had hoped someone will send me a link to something that will work this way but consist of only a few discrete parts. It would be also educative to see how such thing may be created.
Robin2:
I did not realize that. In the circumstances my suggestion in Reply #1 is irrelevant.
Can you save energy by increasing the interval between IR messages?
...R
I am trying to get all means to reduce the energy consumed more from interest than need because currently the batteries would last for months. In fact it would be little difference if I recharge them each 3 or 5 months. But when I am have everything ready I will definitely try to find optimum between intensity of the LED and interval of transmitting the message.
(As a sidenote: I am also using LED with narrow beam (10°) and using sleep modes as much as possible - but ATTiny13A has only one timer + watchdog. Which is a problem, another timer would allow me much more time in Idle and less in Active mode while PWM transmitting the signal.)
MarkT:
Well you didn't avoid the xyproblem at all, but you did give enough information to identify that you
want a constant-current source to drive your LED.
Though an LDO is one way to provide that constant current...
Do you have some better solution? I know about current stabilizing ICs but from my point of view it doesn't matter if I have current or voltage stabilizing IC/discrete circuit. But using stable voltage would be much better for the MCU. Or is current stabilizing circuit so much easier to make?
Smajdalf:
Do you have some better solution? I know about current stabilizing ICs but from my point of view it doesn't matter if I have current or voltage stabilizing IC/discrete circuit. But using stable voltage would be much better for the MCU. Or is current stabilizing circuit so much easier to make?
It does matter, but for small LEDs not too much. The Vf of an LED is a bit dubious based on manufacturing tolerances, age, or temperature, and even small changes in Vf will of course result in larger changes of current.
You mentioned that your concern about using resistors to limit current is due to how the current will change as the battery loses power. In another thread I made the recommendation to someone to connect the LED's cathode through various size resistors to digital pins on the microcontroller which can be set as digitalWrite(LOW) to sink current or pinMode(INPUT) to turn off. If you have the spare pins you could do this as well; adjust your resistance based on the voltage of the battery as read by an analog pin on the micro.
Chagrin:
It does matter, but for small LEDs not too much. The Vf of an LED is a bit dubious based on manufacturing tolerances, age, or temperature, and even small changes in Vf will of course result in larger changes of current.
You mentioned that your concern about using resistors to limit current is due to how the current will change as the battery loses power. In another thread I made the recommendation to someone to connect the LED's cathode through various size resistors to digital pins on the microcontroller which can be set as digitalWrite(LOW) to sink current or pinMode(INPUT) to turn off. If you have the spare pins you could do this as well; adjust your resistance based on the voltage of the battery as read by an analog pin on the micro.
I expect to measure the LED that will be used and then use the "right" resistor. Didn't think about possible changes due to aging/temp (ambient light?) changes. I will look at is, thanks for warning.
Your method consumes many pins but it is clever - I will keep it in mind (possibly for another project).
As it looks like noone here can point me to a "clever circuit" working as I need I will try to ask in our shop if they have something similar to what I have found on internet. If not I will look at shiping charges for one $1 item.