d123
Advanced Member level 5
Hi,
I know it's a really ludicrous question and asking the impossible, but I wonder if anyone can think of a (fantastically) clever way of generating a reasonably stable tiny voltage reference of about 40mV that:
a) wouldn't have a massive increase in Iq with increasing supply voltage.
b) could have a swing of 20mV across temperature range and supply voltage (that wouldn't affect circuit operation).
...Maybe using discrete components and low power ICs, and/or whatever building blocks that may fit the idea - but no microcontrollers, please. If it's impossible, which I suspect it is, well, not to worry. It doesn't have to be particularly precise or rock-steady at 40mV, just not fall below a voltage an op amp comparator could realistically sense, and ideally no lower than 30mV, nor go above a maximum voltage of about 55mV (59mV, in reality) which would stop the torch circuit turning off correctly when it should.
Voltage reference min. to max. range required: 40mV to 50mV, 30mV to 50mV would/could also work
Supply voltage: 3.9V to 6.6V
Temperature range: 0ºC to 85ºC
I want to - at least figure out a theoretical solution for the sake of the learning experience, if there is a solution to such a low voltage reference - shut down the torch circuit (the other thread I keep asking questions in) when it isn't being used because the LM4041 I'm using needs isupply + iload of 100uA at 3.9Vsupply - which would translate to ~300uA at 6.6Vsupply (as clearly, I don't have a supply current resistor that changes value with increasing supply voltage to keep Iq and Il at ~100uA). e.g. A quad LMC6464 uses about 80uA to 140uA across supply and temperature - a decent improvement on 100uA to 300uA.
I have come up with a logic and latch circuit that first in theory and subsequently in a simulation could work:
But outside of Alice in Simulation Land and miniature voltage references that don't exist, a 40mV reference seems utterly impossible. I considered using the minimum output voltage from an op amp follower connected to ground, but it seems to vary enormously across temperature in a simulation. I'd wondered if applying a variable gain block (attenuator) could make the voltage follower rising Vout fall with temperature at an equal-ish rate as it rises and so have something vaguely stable?
I tried simulating a few things with a reverse-biased diode and resistors to limit current and divide down the ~0.55V to something usable, but I see that that is not good either - if in a simulation the reverse-biased diode cathode voltage stays quite steady at 580mV until 60ºC then falls to 380mV, in the real world, I dread to think...
I'd wondered if some sort of Vos follower followed by an inverting amplifier and a resistive divider somewhere in there could achieve anything?
If anyone has any (serious) suggestions, even just theoretical, I'd be really interested in hearing them.
Thanks.
Hi again,
First, LM4041 Iq + iL is actually 76 uA + (22uA + 22uA), about 125uA at 3.9V, about 308uA at 6.6V. My mistake, didn't have my notes/calculations with me earlier.
Second, I imagine the cure is worse than the disease regarding Iq, all additional components considered, I'm aware of this, my interest is 'how-to/can-do attitude' more than 'you're ignoring obvious flaws in your circuit' s current draw, etc.' I would just like to see if a loose, very low voltage reference is a viable idea somehow.
Third, I just realized: I was looking at the problem from (one of) the wrong side(s) - rather than try to fit the reference to the signal, it must be easier to fit the signal to a reference with a non-inverting amplifier with e.g. a gain of 10...
That would leave how to generate a very low power 300mV min. to 400mV to ideal 500mV max. reference. I can't think of anything whatsoever that wouldn't have severe temperature fluctuations, besides supply voltage dependency - not a resistive divider..., not a diode connected BJT..., not a bjt with diode temperature compensation..., not an LM334..., every idea I think of is afflicted by supply dependency and excessive temperature swing, and presumably equally high Iq.
I know it's a really ludicrous question and asking the impossible, but I wonder if anyone can think of a (fantastically) clever way of generating a reasonably stable tiny voltage reference of about 40mV that:
a) wouldn't have a massive increase in Iq with increasing supply voltage.
b) could have a swing of 20mV across temperature range and supply voltage (that wouldn't affect circuit operation).
...Maybe using discrete components and low power ICs, and/or whatever building blocks that may fit the idea - but no microcontrollers, please. If it's impossible, which I suspect it is, well, not to worry. It doesn't have to be particularly precise or rock-steady at 40mV, just not fall below a voltage an op amp comparator could realistically sense, and ideally no lower than 30mV, nor go above a maximum voltage of about 55mV (59mV, in reality) which would stop the torch circuit turning off correctly when it should.
Voltage reference min. to max. range required: 40mV to 50mV, 30mV to 50mV would/could also work
Supply voltage: 3.9V to 6.6V
Temperature range: 0ºC to 85ºC
I want to - at least figure out a theoretical solution for the sake of the learning experience, if there is a solution to such a low voltage reference - shut down the torch circuit (the other thread I keep asking questions in) when it isn't being used because the LM4041 I'm using needs isupply + iload of 100uA at 3.9Vsupply - which would translate to ~300uA at 6.6Vsupply (as clearly, I don't have a supply current resistor that changes value with increasing supply voltage to keep Iq and Il at ~100uA). e.g. A quad LMC6464 uses about 80uA to 140uA across supply and temperature - a decent improvement on 100uA to 300uA.
I have come up with a logic and latch circuit that first in theory and subsequently in a simulation could work:
But outside of Alice in Simulation Land and miniature voltage references that don't exist, a 40mV reference seems utterly impossible. I considered using the minimum output voltage from an op amp follower connected to ground, but it seems to vary enormously across temperature in a simulation. I'd wondered if applying a variable gain block (attenuator) could make the voltage follower rising Vout fall with temperature at an equal-ish rate as it rises and so have something vaguely stable?
I tried simulating a few things with a reverse-biased diode and resistors to limit current and divide down the ~0.55V to something usable, but I see that that is not good either - if in a simulation the reverse-biased diode cathode voltage stays quite steady at 580mV until 60ºC then falls to 380mV, in the real world, I dread to think...
I'd wondered if some sort of Vos follower followed by an inverting amplifier and a resistive divider somewhere in there could achieve anything?
If anyone has any (serious) suggestions, even just theoretical, I'd be really interested in hearing them.
Thanks.
--- Updated ---
Hi again,
First, LM4041 Iq + iL is actually 76 uA + (22uA + 22uA), about 125uA at 3.9V, about 308uA at 6.6V. My mistake, didn't have my notes/calculations with me earlier.
Second, I imagine the cure is worse than the disease regarding Iq, all additional components considered, I'm aware of this, my interest is 'how-to/can-do attitude' more than 'you're ignoring obvious flaws in your circuit' s current draw, etc.' I would just like to see if a loose, very low voltage reference is a viable idea somehow.
Third, I just realized: I was looking at the problem from (one of) the wrong side(s) - rather than try to fit the reference to the signal, it must be easier to fit the signal to a reference with a non-inverting amplifier with e.g. a gain of 10...
That would leave how to generate a very low power 300mV min. to 400mV to ideal 500mV max. reference. I can't think of anything whatsoever that wouldn't have severe temperature fluctuations, besides supply voltage dependency - not a resistive divider..., not a diode connected BJT..., not a bjt with diode temperature compensation..., not an LM334..., every idea I think of is afflicted by supply dependency and excessive temperature swing, and presumably equally high Iq.
Last edited:
