Testing a voltage ... using that voltage???

[carbonzit]

Another thread here, someone asking about a circuit to test the condition of a battery, got me wondering about this. What I'm getting at is this: how do you measure a voltage, when the voltage source of the measuring circuit is that same voltage?

What I'm guessing (could be way off here, I know) is that one would want to derive a reference voltage from the battery, then compare the reference against the battery voltage. Correct? Of course, the reference voltage would have to be a lot smaller than the battery voltage when fully charged, maybe somewhere below Vcc for whatever measuring circuit is used. And obviously, it'll only work down to the point where the battery is still capable of powering the measuring circuit.

So how is this done in real life? It seems somewhat of a circular (or recursive) problem.

How far down a battery's charge can such a circuit (more or less) accurately measure that charge?

 

[JimB]

In the good old bad old days, we just used things called moving coil meters.
No need for a separate supply to energise the measuring device.

JimB

 

[alec_t]

How far down a battery's charge can such a circuit (more or less) accurately measure that charge?

An analogue meter (as per JimB above) can measure all the way down to the tens of mV level with reasonable accuracy.
For another method, accurate 1.2V refs are available. Not sure what overhead they need. I'd guess you could use an op-amp, plus 1.2V ref self-powered solution down to ~3V.

 

[Reloadron]

There are more expensive panel meters that will measure the voltage accurately they are powered by, however, most low end panel meters will not. The meter creates a virtual ground for the common input so that point is not the same as battery negative.

Ron

 

[crutschow]

I have a small Chinese made digital panel meter that allows the power common and the input common to be the same point, which I use to measure battery voltage on my motorcycle It can measure down to about 7V or so before the LED display becomes too dim to read.

 

[Ziddik]

in my experience using a digital multimeter, no, you cant!! I had tried to measure the battery of a DMM using the same battery, but the meter went off itself!!! It is not possible least using my expensive multimeter

 

[carbonzit]

No, no, no, no, no.

Let me try to get this derailed conversation back on the tracks.

I'm not after displaying the measured voltage with a panel meter.

I'm asking about a device that measures its own voltage source and uses that measurement to perform some function (like the example I gave of a battery-condition indicator; in that other thread someone wanted to have several LEDs show the remaining battery life in %).

The question is how a circuit can measure its own supply voltage, and use this result to control some function in the device (turn a LED on/off, switch to reserve power, sound an alarm, etc.).

My guess is something like this (too lazy to draw a schematic just now):

• Voltage source--> reference voltage (low-voltage zener)
• Reference voltage and raw voltage source fed to +/- inputs of comparator
• Comparator output used to control function

This is just my guess at this point. How does this work in actually existing electronic devices?

 

[DemonicSandwich]

Such a circuit could be a simple as a comparator, a zener diode to supply the solid reference and a voltage divider across the supply to provide the moving voltage.

Possibly also a feedback resistor to add a little hysteresis to stop output oscillation.
As long as the comparator has a solid reference at one of it's inputs, it can be supplied by the voltage it's measuring.


This is how many low-end voltage monitor ICs are constructed.

Edit: This would be an example of what I was talking about.
**broken link removed**

 

[Mr RB]

If you don't need a really precise detection point you can just use an NPN transistor and 2 resistors as a voltage divider to drive its base. It can be used on any voltage over 0.6v (say from a 1.5v battery to a 12v battery).

If you can put a trimpot in there you can make it more precise by trimming for the trip point you need.

 

[Reloadron]

A series of comparators like LM311, LM393, and the LM339 or just a configured LM3914. The LM3914 is popular for applications like this.

Ron

 

[DemonicSandwich]

You could also use an LM431 shunt regulator for a decently accurate measurement and only a few resistors would be needed.

 

[ronv]

Yes, like the LM3914 you develope a lower reference voltage to measure the battery voltage against.

 

[carbonzit]

If you don't need a really precise detection point you can just use an NPN transistor and 2 resistors as a voltage divider to drive its base. It can be used on any voltage over 0.6v (say from a 1.5v battery to a 12v battery).

Not exactly the same thing, but would something like this work?

**broken link removed**

The idea being to use a series of different-valued zener diodes to establish several voltage levels to indicate battery life. (Problem: zener values aren't that closely spaced, so might require some tweaking, like inserting regular diodes to add 0.7V drops, etc.)

 

[crutschow]

One concern is that the circuit doesn't run down the battery from its own current. You want to have it use much less current than whatever the normal battery load is.

 

[carbonzit]

One concern is that the circuit doesn't run down the battery from its own current. You want to have it use much less current than whatever the normal battery load is.

Yes, good point. I would somewhat "starve" the LED, by sizing R2 to put it on a low-current diet.

Other than that, would it work?

 

[crutschow]

You could use a high brightness type LED which usually can give a reasonable bright display at a mA or so.

The circuit should work as you posted. But with R2 in the emitter the brightness will vary with the voltage as it goes above the zener voltage. For a sharper display turn-on point move R2 to the collector, in series with the LED.

 

[Mr RB]

Not exactly the same thing, but would something like this work?

**broken link removed**
...

Yep that's how it's done. You still need the second resistor in the voltage divider (above the base) to reduce current when the battery is higher than the setpoint (which will be most of the time). The zener is not needed but if you add the zener you improve the sensitivity where it turns on.

Really if you were going to make a few of these with all the related parts count why not just use the LM3914 to drive ALL the LEDs as Reloadron suggested?

Or a $1 PIC like a 12F675 that has ADC input and abilty to drive 5 LEDs.

 

[carbonzit]

Really if you were going to make a few of these with all the related parts count why not just use the LM3914 to drive ALL the LEDs as Reloadron suggested?

Or a $1 PIC like a 12F675 that has ADC input and abilty to drive 5 LEDs.

If 'twere me, I'd go the LM3914 route. After reading the datasheet, that would be the low-parts count and low-current way to do it. A PIC? Nah; too much programming for such a trivial device.