Hi everyone.

Here is my simple project where I'm trying to get a 15 led battery charge strip using a couple of LM3914 circuits.

As a power source, two 18650 li-ion batteries in direct connection are concerned to be used.

Here's the scheme I tried to simulate with Proteus.

The working principle is next: the voltage rating I'm trying to detect is 8.4V --6V (fully charged and almost fully discharged li-ion battery pack) witch gives a 2.4V of difference.

Since the lowes possible value of the voltage difference between Rhi and Rlo is 1.2V, and the Vdif is 2.4 then logically that the voltage for Rlo1 for the first LM-circuit must be:

8.4-1.2=7.2V (Rhi1 = 8.4V).

The Rlo2 voltage for the second LM-circuit is:

7.2-1.2=6V (Rhi2 = 7.2V)

So virtually it's working pretty fine.

But some problems appear in the real world.

The Rlo voltages which I set with R2 and R4 are less than I expected. So I thought that it must be trouble with the reference voltage of the IC.

I tried to measure that 1.25V between VRO (pin 7) and ADJ (pin 8) but for me, it's equal to 200mV. I tried 4 different IC but the values are the same for each.

Moreover the REf. voltage source in the assembled scheme is in depending of the power voltage, provided to the IC while it should be stable for the Rlo reference.

Could you guys tell me what I'm doing wrong here and why there isn't a so much desired stable 1,25 at the "naked" LM3914 IC.

Thanks!

upd.

Forgot to mention where the value of 440 R appeared from.

The LED current I want to have is 30mA. For that by the formula from the datasheet, I get approx. 416R for R1. Then I take into account leakage current from the ADJ pin and a bunch of resistors inside of the IC's comparators divider, so I end up with 430R (pretty close to 440R I have).