L7805 + Capacitor as an as-is buck converter?

I’m trying to design a tiny-as-possible voltage regulator that is kinda-efficient as it will run on ~1200 - 2200 mAh LiPo batteries.
The problem is that the input voltage is not fixed, it might run on 2 or 3 cell LiPo or Lion, or NiMH between 7.2 - 9.6 volt.

To provide a stable 5v input, I’m thinking on using an L7805 as a regulator, but to save battery life, I also plan to use capacitors and a P-FET to switch on/off the regulators.

The controlling would run on an Attiny84 that would monitor the input voltage and if it falls under the limit, it switches on the P-Fet that would let the L7805 to charge the capacitors, and if the voltage raised above a limit, the P-Fet is turned off.
Note: The P-FET’s number is not the actual I’m planning to use, it would be a logic level FET.

Beside the control circuit that the battery will also supply a high-load DC Motor (~20-25 amper) which will be controlled with an N-FET.

To monitor the battery voltage, I also added a really high Ohm voltage divider. The load between them would be basically zero, as all I need them to do is monitor the voltage.

Would this “theory” work in practice?

Mmm... If your batteries are supposed to operate a 20-25A load, why do you even care about saving power on a part that takes 3-4 orders of magnitude less?

Can that tiny battery even supply that much current without exploding or bursting into fire? If it can those motors would drain the batteries in 2-5 minutes or so...

Then the schematic:

  • no input cap for your regulator.
  • no decoupling cap for the ATtiny.
  • add a small cap (1-10nF) in parallel with R9 for stable readings.
  • replace that optocoupler with a gate driver. That MOSFET no doubt has a pretty high gate capacitance, and you really really don't want it to linger in the partly on region with this kind of loads, which it will especially when switching off (that gate is discharged rather slowly through R2).
  • why is RESET of the ATtiny connected to A6?
  • I don't see how you could ever power up that ATtiny. How that PMOS is wired is definitely wrong, as in completely wrong. Get a regulator with enable input, much easier.

You can’t make a linear regulator more efficient by switching it on and off, its efficiency depends on the ratio of input to output voltage.

Hi,
Have you looked for LDO 5V regulator, LDO == Low Drop Out.
As you are only using it to supply the ATtiny84 a TO-92 case version would probably work, with a few bypass capacitors.

google linear 5v regulator low drop out

They are even available in SMD packages.

Tom.. :slight_smile:

TomGeorge:
They are even available in SMD packages.

Reading that I couldn't help thinking: yes, of course they're in SMD packages, I was more like: are they even available in TO-92?! I didn't know that package is used much if at all for modern designs like LDO regulators. Used to see them as SMD only, often in tiny sizes (such as SOT23-5).

wvmarle:
Mmm... If your batteries are supposed to operate a 20-25A load, why do you even care about saving power on a part that takes 3-4 orders of magnitude less?

Can that tiny battery even supply that much current without exploding or bursting into fire? If it can those motors would drain the batteries in 2-5 minutes or so...

Then the schematic:

  • no input cap for your regulator.
  • no decoupling cap for the ATtiny.
  • add a small cap (1-10nF) in parallel with R9 for stable readings.
  • replace that optocoupler with a gate driver. That MOSFET no doubt has a pretty high gate capacitance, and you really really don't want it to linger in the partly on region with this kind of loads, which it will especially when switching off (that gate is discharged rather slowly through R2).
  • why is RESET of the ATtiny connected to A6?
  • I don't see how you could ever power up that ATtiny. How that PMOS is wired is definitely wrong, as in completely wrong. Get a regulator with enable input, much easier.

Yes, the batteries can supply the DC motors, they aren't draining them dry in minutes because the ON/OFF time ratio is low, but the regulator would run all the time constantly draining the battery.

  • Noted
  • Isn't C1 and C2 should be this?
  • Noted
  • Noted, but we use these setup for years with zero problem so far. The difference that we directly connect them to the battery like: the Battery + => Microswitch => 100o resistor => FET Gate => 30ko resistor => Battery -, they handle the load pretty okay-ish most of the time (95+%). This time I want it to be controlled by an MCU.
  • That was a mistake when I partly re-wired the blueprint.
  • Noted

MarkT:
You can't make a linear regulator more efficient by switching it on and off, its efficiency depends on the ratio of input to output voltage.

Indeed, the regulator itself won't be more efficient, but (if I think right) the overall consumption should be better.

rabirland:
Indeed, the regulator itself won't be more efficient, but (if I think right) the overall consumption should be better.

No, the current will be proportionally higher during the "on" periods so the net effect is that the average current and voltage across the regulator is about the same as without the switched power.

rabirland:

  • Noted, but we use these setup for years with zero problem so far. The difference that we directly connect them to the battery like: the Battery + => Microswitch => 100o resistor => FET Gate => 30ko resistor => Battery -, they handle the load pretty okay-ish most of the time (95+%). This time I want it to be controlled by an MCU.

"pretty OKish 95% the time" doesn's sound good to me, at all.

"Working well all the time" would be more like it. So that would give you that 95% of the time from "working OKish" to "working well" and that 5% of the time that it apparently doesn't work to also "working well".

C1 & C2 offer bulk capacitance. That's useful, but not enough. Every single digital IC needs a 100 nF ceramic capacitor placed physically as close as possible to the Vcc pin. If more Vcc pins on the IC, that's one capacitor per Vcc pin.

wvmarle:
"pretty OKish 95% the time" doesn's sound good to me, at all.

"Working well all the time" would be more like it. So that would give you that 95% of the time from "working OKish" to "working well" and that 5% of the time that it apparently doesn't work to also "working well".

C1 & C2 offer bulk capacitance. That's useful, but not enough. Every single digital IC needs a 100 nF ceramic capacitor placed physically as close as possible to the Vcc pin. If more Vcc pins on the IC, that's one capacitor per Vcc pin.

That 5% is mostly the edge cases when someone makes a horribly beefy build and puts extreme load on them. In regular up to "kinda-harsh" loads I've never seen any problem with properly installed N-FETs.
But after all it looks like this solution will not work, so I will need to somehow squeeze in a tiny buck converter on the PCB.