Intended primarily to monitor a 4-cell R/C receiver battery, it soon became apparent during design that the same unit could also be tweaked to monitor 12V gel-cells. The component values given means it can be user-adjusted for standard 4-cell and 5-cell packs or any battery up to a maximum 25 volts.
For the sake of simplicity this article deals with the discussion and setting-up of the 4-cell version only, but by the end of the piece you'll have gleaned enough info' for setting up other voltages.

So what's it all about, Alfie?

For my part at least, knowing exactly when to recharge the Rx battery has always been a somewhat hit-and-miss affair. Should I risk the model one last sail / flight before a recharge?... do I risk incurring the dreaded memory-effect on the Ni-Cads if they're charged before necessary?... and so on. Something was needed - something that would totally illiminate all the guesswork.
Trial-and-error testing on the bench brought this circuit about. One nice feature is that it's self-powered and requires no batteries of its own. It's also easy to set up because all you need is a freshly-charged pack and some time to kill. Another nice feature is that instead of the usual zero-to-maximum scale, this scale is what is otherwise known as an Expanded Scale Voltmeter.
Nice, nice.

So what does an ESV do?

In easy English, this particular expanded scale only shows voltages in the range from 4.4 volts to that of the fully-charged battery (usually around 5.6V).

Why?

Because 4.4 volts to 5.6 volts is the usual working range of most 4-cell rechargable battery packs. It is generally accepted that 1.1 volts is the critical point at which a 1.2V cell runs out of steam. Their natural tendency is that the voltage drops off rapidly soon after the cell falls below this critical level.

Meaning what?

Meaning, there is little point in monitoring the battery once it falls below 4.4V, because although the battery will still register readings below this level, it doesn't necessarily mean that the needed oomph will be there, and by then your receiver and servos will probably have died anyway. Whereas an ESV monitors the level of the battery across its working range only, and the visual display gives fair warning before the liklihood of the battery going bad at the wrong time.

So without further ado...

Preparing the board.

Start by cutting a piece of veroboard, 18 holes x 12 tracks, then take a file and tidy up the four edges. Flip the board track-side up and make 30 breaks at the locations shown.

Six wire links are needed on this board. Single-strand telephone wire (or a leg snipped from a resistor) is ideal. But forming the correct length of each wire link with needle-nose pliers is time consuming and tedious, so I usually stuff them in first. In this case they have to go in first because two of them are under the chip and another under VR1.

Note that the link at location Q9 coincides with a break on the underside. You need to bend this link towards P9 and solder it there (see below).

Stuffing the components.

There is one final wire link to solder. Its job is to connect all the anodes of the display to the pos' supply rail. Probably the neatest way is to flip the board over and solder it from beneath. Refering to fig.4 on the right, you need to solder this link from location D2, all the way through to D12.

The link poking through the break at Q9 should be bent towards P9. Don't solder it just yet in case P9 should get clogged with solder -- insert VR1 first.

(In order for VR1 to sit flush against the board, you might want to file a small groove where it sits over the link.)

The DC power connector.


While the board is still flip-side up, now is a good time to solder the 2.1mm power connector. The solder lug is first snipped off, then the threaded end soldered between tracks A6 and A7. A small length of wire is needed to make the connection to the negative rail.
Refering to fig. 5 on the left, note the blue wire from location J4 to the center conductor on the power plug.

Parts-placement overlay.


Perhaps a neater way of doing things would be to paste a parts-placement overlay to the board first. It's then just a matter of stuffing all the components where they are shown. And although aligning the overlay to the board can be a bit tricky, there are perhaps a two advantages:

1. Reduces the chance of stuffing a component in a wrong location.
2. Gives the board a neat PCB-type appearance.

Throw the switch, Egor. Let's give it life!

• Is pin 1 of the chip where it should be?... Check.
• Is the display chamfer where it should be?... Check.
• Is your battery freshly charged?... Check.
• Have you made up your mind whether it's to be Bar or Dot mode?

If no spare channels are available at the receiver, an easy alternative is to route the ESV through a Y-lead between the battery and receiver. When everything is connected, the first adjustment is at VR1. This is the 'Set High' trimpot and it sets the display to show all ten LEDs^(*) on a fully-charged battery.

1. First, set both trimpots fully counter-clockwise.
2. Next, plug the unit into a freshly-charged battery.
3. Now turn VR1 clockwise until the tenth LED just turns on.

Just to get you in the right area, you'll find VR1 needs 4 to 5 full turns before the display comes to life. Nothing bad will happen if you overtweak, it just means there isn't much point in turning the brass screw any further once the dispay is showing full scale. Tweak the screw higher/lower until you're satisfied that your freshly charged battery corresponds with your full scale setting.


(*) This assumes that the 2-pin plug is inserted in the 2-pin header (bargraph mode). Without the plug the display will be in dot mode and only one LED at a time will glow.

1. A digital multimeter.
2. Time to kill.

The best way to measure any battery in order to get the most accurate reading of its V is to make the measurement while the battery is under load. A no-load battery will otherwise usually show a higher reading once it's disconnected from the circuit. For our purpose it makes sense to have the first LED just and just turn on at 4.4 volts. It's best to do this while the battery is connected to the load, else the ESV could end up showing a higher reading than is actually true.
Connect your multimeter probes to where the servo wires meet the solder tags on the 2.1mm socket. When the multimeter reads 4.4 volts, adjust VR2 gradually clockwise until only the first LED remains on. Try to tweak it that the LED is neither fully on nor fully off, but at a point where you consider it as the critical "Charge Me Now!" warning.

Got a variable Power Supply Unit?

Setting it to 4.4 volts saves a whole load of time, and easily takes care of the VR2 Low adjustment. The PSU can also be used for setting VR1 High adjustment to, say, 5.6V since most fully-charged packs will be somewhere close to that voltage anyhow.
However, accuracy is not so important at the high end -- it's the 4.4V low end of the scale that we're most interested in. You want it adjusted so that when the last LED is the only one on, then is the time to haul the model in for a recharge.
Once you're happy with the setting of both trimpots they should never need adjusting again. Stick a blob of thread-lok or hot-melt glue on each of the brass screws.

The 2.1mm socket shown in figures 7a and 7b represents the type normally used for pcb work. An alternative method is to use the threaded chassis-mount type. But whichever version you use, make sure the negative wire is connected at the center conductor.
A 6.0mm hole is required for the pcb-type socket. Drill the hole then epoxy the socket precisely in place. But bear in mind to drill the hole where it allows easy access for plugging in the ESV, and bear in mind also the issues with water ingress. A typical location on model boats might be to have the socket disguised behind, say, a porthole, or a deck hatch.
On model airframes you don't want the hassle of removing the wing each time you want to take a V reading. If it's a high-wing plane, and/or with an IC engine, it's best to fix the socket inside the fuselage, at the opposite side of the silencer. You want to be able to see the LED display easily, even with the wing in place.

Finally, re-solder the connections at the 2.1mm socket and sleeve both joins with heatshrink.
Just for completeness, stick a piece of foam tape on the underside of the circuit board. The heatshrink prevents against possible short circuits, and the foam tape protects the negative wire and help prevent the copper tracks from tarnishing.

The screw on the trimpot has no end stops. How do I know when it's fully CCW?
Three ways: The first is to listen for a tell-tale audible 'click' when the wiper reaches the end of its travel. The second way is to measure for minimum resistance across the middle pin and the pin at the opposite end of the brass screw.

What is the third way?
A multi-turn trimpot means just that. These particular pots require fifteen complete turns from end to end. If you've installed both trimpots before setting their minimum resistance, then a sure-fire way is to keep turning 'til your wrist aches.

When in bargraph mode, how much current is pulled at full-scale deflection?
The 1K resistor on pins 6 and 7 was chosen to drive each LED at around 15mA. I measured slightly over 150mA with all ten LEDs on.

Is that bad?
Pulling 150mA on a fresh battery for just a moment or two will hardly have any diverse effect. As the voltage drops so does the display, which means just a measly 15mA is drawn at the last LED.

Is there a way of reducing that 150mA draw?
Dot mode reduces the draw to around 15mA. Alternatively, try increasing the value of R2. For instance, increasing R2 to 1K5 reduces the 10-LED current to around 100mA, whereas decreasing to 680-ohms will pull around 210mA. A higher value means a lower draw, but the pay-off is a dimmer display.

Is the display easy to read in daylight?
Yes and no. Like most LEDs, if you point the display directly towards the sun it becomes almost impossible to read. Setting R2 at 1K seems the best compromise between brightness versus current draw.

Why does the display sometimes jump one LED higher when selecting between modes?
Because Dot mode pulls less current. Less current means less of a drain on the battery, so the voltage goes up.

How do I set it to monitor batteries other than 4-cell packs?
Let's assume you're using a fully-charged 7.2V pack. Adjust VR1 first for full-scale deflection, then, to set the Low adjustment, you simply multiply 1.1 by the number of cells in the pack. In this case, multiply 1.1 by 6, then adjust VR2 so that the first LED just turns on when the battery drops to 6.6 volts.

So can this gadget monitor my 12v gel-cell?
Yes. Tweak VR1 in the usual way for a freshly-charged battery, then tweak VR2 so that the first LED illuminates at the point where the cell is just above its critical shut-down voltage -- usually somewhere around 11.6 volts.

I'm running two 12v gel-cells in series. Can I monitor 24 volts?
Probably not. The LM3914 data sheet specifies 25 volts as the absolute maximum. A fully-charged gel-cell can be as high as 14 volts, so 14V + 14V would be pushing the chip three volts above its limit.

What is the least voltage I can monitor?
About 3.8V - a bit less if you link out resistor R1. But 3.0 volts is as far as you can go anyway, because then you reach the lower operating limit of the chip.

Doing it the easy way:

Alternatively, make a printout of this component-overlay template. With one of these, all you have to do is stuff the board where the components are already shown.
The image shown on the left is obviously too large. So once you've copied it to your drive, and in order to have the printout match the actual board dimensions, you'll have to set your printer reduction to 54%.
When you get your first printout, hold it, and the board, against the light, then try aligning the holes nearest the corners. If you're lucky, they should be pretty much spot on. If not, try adjusting your printer setting just one per-cent plus/minus either side. It's a case of trial-and-error, but starting with 54% will put you in the right numbers.
Once you're armed with the correct setting, make a final printout on gloss photo paper. Roughly trim it to size, leaving approx 5mm around the perimeter. Paste the reverse side (I use Pritt Stick), then align / stick it on the board - careful not to get glue in all the wrong places.
Depending on the exact size of how you cut / pruned / filed your board, you might have to trim any excess overhang from the overlay. Overlays are drawn with a slightly larger perimeter purely for this reason.

Pushing those skinny legs of resistors, etc, through plain-paper overlay is easy. Pushing them through gloss paper isn't. Jab a needle through there first.