Many low voltage indicator circuits oscillate on and off when the battery voltage is low because the voltage is low when loaded and rebounds up when unloaded.

That would be fine.
As long as the LED comes ON when the LOADED voltage dips to approximately 2v (for the two AA's in series) give or take

 

Ok,
Last night I bread boarded the circuit and tested it.
Instead of a static resistor for R3 I used a 10k POT so I could adjust the resistance.

What I found was that the TL431A could not give me the resolution down to 2v as was explained by another member above per the Specifications for this device and my tests confirmed that. Obviously the TL431A has a 2.5v threshold. I was only getting a response from the device down to 2.8v. Below that it was unresponsive.
However, for any application where I needed a visual alert (LED alert) from 3v to 36v, this circuit would work fine.

So I'll be trying again with the TLV431 when that order arrives. The TLV431 can be set anywhere between VREF (1.24 V) and 6 V so it should be perfect for this application.
http://www.ti.com/product/TLV431

I still need to figure out the voltage cut off point I will use for the two batteries UNDER LOAD because loaded, with the vacuum running, when it reaches a minimum loaded voltage is when I need the LED to come on. This breadboard will be good for those experiments. I imagine it will be different for different battery types (NiMh vs NiZn vs LiIO etc). It's all in electronics hobby fun for me. Nothing critical. I guess I could even make up one for my scooters to tell me when the battery drops below a certain voltage (important on scooters). I may add one (if I can make the PCB small enough) to my battery powered, 9.6v Harbor Freight Rotary Tool....battery powered drills etc....haha.... Lot's of fun possible applications. I'm assuming that the current drain for this circuit will be negligible.

If it all works as needed with the TLV431, I guess I home brew a few PCB's for it and put it to use. There's actually a good number of low voltage devices I have that this would be handy on to help save the batteries. Especially if I use NiZn's which I happen to like.

 

I'm stuck.
I replaced the TL431 with a TLV431 and now the circuit doesn't work.

Do I need to recalculate the values needed for the resistors?
I tried various resistors of different (generally lower) resistance but the LED no longer comes on.

 

I'm stuck.
I replaced the TL431 with a TLV431 and now the circuit doesn't work.

Do I need to recalculate the values needed for the resistors?
I tried various resistors of different (generally lower) resistance but the LED no longer comes on.

Believe it or not they have different pinouts.

 

Believe it or not they have different pinouts.

Richard,
I checked the pinouts and they seem the same....123 Left to right across the flat face.

I did actually get it working by experimenting with differnt value resistors.

The BIG issue now seems to be that LED's I have dont want to operate at voltages below about 2.5v

To do this with a TLV431, it seems I will need Very Low Power LEDs
Something in the 1.5 - 2v range

I wonder if anyone will scream if I start a new thread on Very Low Power LED's ?

 

that LED's dont want to operate at voltages below about 2.5v

They won't if the forward voltage drop of the LED is 2.5V or greater.
What is it specified at?

 

The BIG issue now seems to be that LED's dont want to operate at voltages below about 2.5v

That's true for the blue LED you show. A red LED needs only about 2V.

 

That's true for the blue LED you show. A red LED needs only about 2V.

Wow!
You are correct. I just tried a RED LED and now I can go down as low as 1.7v before it dies. I may be able to work with the RED LED.
Thanks! What about white, yellow and green?

Time to try some experiments now with the actual device (vacuum) under loaded conditions.

 

They won't if the forward voltage drop of the LED is 2.5V or greater.
What is it specified at?

I didn't get ANY specs with these LEDs. Just a lot of 100 I bought off ebay.

They could be 12v, 5v or who knows?

 

I checked the pinouts and they seem the same....123 Left to right across the flat face.

Oh. I was thinking about the surface mount parts. Sorry about the confusion.

 

So I got this working with considerable help from a few most excellent members.

I still have to tune it in for the specific device (battery powered vacuum) I'm going to apply it to (under load) but the basic breadboard that works used....

TLV431 (in place of the TL431)
R2 - 3K ohm (in place of a 4.7 ohm)
R1 - 220 ohm (in place of a 2.2k ohm)
R3 - To be determined. Right now I still have a 5K POT on the breadboard.
RED LED - Thanks Alect_t !!

I have no clue how much of the battery power will be lost powering this circuit, as long as it's not much, it won't make much difference. The important thing is to let me (or whoever the operator is) know when it's time to recharge the batts.

 

BTW

Vishay has some Low Power LED's
https://www.vishay.com/leds/low-current/

 

BTW

Vishay has some Low Power LED's
https://www.vishay.com/leds/low-current/

I think you're confusing voltage and current limitations. The forward voltage (Vf) of the blue LED was your problem before. The amount of current it takes to achieve a certain brightness isn't really an issue. There's a certain minimum voltage drop across the LED, below which it simply won't work. This voltage drop varies a bit from one LED to the next, but primarily varies as a function of color (which I believe is due to the varying chemistry required to achieve different LED colors.)

If you look at the Vf specs on the LEDs you linked to, you'll see that they're no better than the 1.7V limit you found with your red LED. In fact, most of them have higher Vf specs than that.

 

This voltage drop varies a bit from one LED to the next, but primarily varies as a function of color (which I believe is due to the varying chemistry required to achieve different LED colors.)

It's related to the fact that shorter wavelength light (towards the blue, away from the red) has a higher photon energy and that requires a higher bandgap in the LED material to generate the light, and this higher bandgap means a higher forward voltage drop.

 

It's related to the fact that shorter wavelength light (towards the blue, away from the red) has a higher photon energy and that requires a higher bandgap in the LED material to generate the light, and this higher bandgap means a higher forward voltage drop.

Thanks! I think I've read that before, but had forgotten. Short attention span!

 

Question.....
Where Do I tap into the power stream? I need the actual voltage of the two batteries combined (series) while the motor is running.

The motor is interfering with the voltage readings and they vary a lot while the motor is running.
I need the actual battery voltage while loaded (running). Do I need a Diode? Capacitor? The wiring is simple. Series wired with a switch in between.
There appears to be no circuitry at all. Just the straight wires and the on/off switch.
Here is a picture of the device.....

 

Simple RC low pass filter seems like a solution. I might be wrong, but I *think* adding a capacitor in parallel with the LED will form a low pass filter in combination with the existing resistor.

 

I'd try an RC low pass filter. Check here

You might also want to build in hysteresis. I'd probably use a comparator and experiment to see the best hi and low points.

 

Simple RC low pass filter seems like a solution. I might be wrong, but I *think* adding a capacitor in parallel with the LED will form a low pass filter in combination with the existing resistor.

Argh! I don't know why I have such a mental block on this TL431 as a comparator concept, but I do. Anyway, I suspect a simple low pass RC is the solution, but I'm pretty sure it isn't a cap in parallel with the LED like I suggested earlier.

After looking more closely, I'm pretty sure the filtering needs to go between the junction of R2-R3 and TL431 input pin 1. However, I'm at a loss on how to calculate RC time constants, 3dB down points, etc. because I strongly suspect that the resistors in the voltage divider interact in a complex way with any added filtering.

I'm guessing if you put a cap across TL431 pins 1&2 you'll get some low pass effect, but I'm not sure how much, or whether there would be a better way to implement a low pass for this circuit.

I'm afraid my misguided ramblings may be doing more harm than good. I'm clearly out of my depth. I'll shut up now!

 

Try using the LT6700 part http://cds.linear.com/docs/en/datasheet/6700123fh.pdf You can go down to 400 mV and it's low power. The first page has exactly a low battery indicator.