5.4.5 Testing and troubleshooting previous
    index page

When a set of indicators is built carefully and without hurrying (even by an unexperienced person) and the quality of the parts is O.K., well over 50 % of them should work right away. So the testing and troubleshooting is merely to get the last few pieces to work as well.

For testing and calibrating, a stabilised voltage supply is needed. In annex E.1: A stabilised voltage supply, a design for such a device is given but similar stabilised voltage supplies can also be bought. The device from annex E has as features:

The voltage adjustment and current limitation cooperate in the following way:

Normally, there is a risc that devices that are being tested, are destroyed because of faulty connection, test leads that make short circuits and so on. With such a stabilised voltage supply, these riscs are virtually eliminated as long as the current limitation is set to 25 mA. All components of the indicator can stand a short circuit current of only 25 mA.

The current limitation value can be adjusted to 25 mA using the scale of the current limitation knob. Check whether the actual short circuit current comes close to this value:

Measure the current between positive and negative with the tester without anything else connected to the voltage supply: Should be 25 mA. Readjust the current limitation knob if necessary and note the proper setting on the scale.

Put the red tester leads back into the voltage/resistance measuring socket immediately afterwards (on some testers there is just one socket for the red tester lead, then switch selector switch back to voltage range). Trying to measure a voltage while the tester is set up for current measurements means creating a short circuit.

Complete indicators can be tested but then it is more difficult to measure and remove components. Therefor it is best not to fit the current shunt and front cover yet. Also switch S is not necessary yet if the positive wire from the stabilised voltage supply is soldered directly onto the copper strip where the switch is connected to.

The indicator should be connected to the stabilised voltage supply in the following way:

Then there are some general advises:

There are two main tests: Switch the stab. volt. supply on and let the voltage vary between say 11 and 13 V. See whether all LED's light up in turns. Adjust voltage to 12.08 V and turn trimmer R4 until LED 5 just goes off and LED 6 just lights up. If these tests were completed succesfully, skip the remainder of this paragraph and continue with par. 5.4.6: Calibrating indicators. If not, it is time for some troubleshooting.

General checks:

One very common error (at least I have made it many times...) is that the voltage supply is faulty or just not plugged in. One easily assumes that the error is in the complicated indicator circuit, but usually that has been made carefully. It is the temporary voltage supply that was made and connected hastily, that often causes the problem, so:

Check the indicator for possible errors, use the print lay-out in fig. 5.4 and the information in the parts list:

If this does not solve the problem, more detailed measurements are needed. Many involve measurements on the pins of the chip, so it is important to know how these pins are numbered. The place of pin 1 is marked with a `1' on the copper side of the PCB and pins are numbered in a round-going manner. So pin 18 is opposite pin 1 and pin 10 is opposite pin 9.

No LED's light up at all:

Check the voltage supply. This comes down to testing whether all components that need a positive or negative voltage, receive this voltage. In the electronic circuit, it means testing the top line (which should have a positive voltage) and the bottom line (negative voltage).

The sequence of measurements is from the voltage supply towards the `electricity consumers'. So if one measurement shows no voltage while the previous one was normal, it could very well be because there is a loose connection or interrupted copper strip between these measuring points. This possibility is not mentioned at each step, but is worth checking.

Have the negative tester lead permanently connected to negative of stab. volt. supply and measure with positive lead on the different points:

Maybe all LED's were fitted with polarity wrong or were destroyed during soldering, see below:

Maybe signal or reference voltages are wrong, see below.

LED 10 lights up all the time:

Probably the current shunt was not fitted and `L-' is not connected to `B-' by a temporary wire, repair this.

Signal/reference error, see below.

Some of the LED's do not light:

Switch off stabilised voltage supply and check:

If the polarity is O.K. and the LED was O.K. before fitting, the most likely cause is that the non-functioning LED was destroyed during soldering. Since LED's cannot be tested whether they conduct in the right direction as long as they are connected to the chip, either the chip or the LED has to be removed for a conclusive test.

If it is LED 10 that does not light when it should, measure the voltage at pin 9 (`mode'). Pin 9 should not be connected to anything and then its voltage should be 0.2 - 0.5 V lower than the input voltage of the chip at pin 3, so ca. 10.8 V (with a simple analog tester, connecting the tester itself might influence the reading, then measure the voltage between pin 9 and pin 3: Should be about 0.3 V).

When the voltage at pin 9 is too low, the chip will not light LED 10 irrespective of the input voltage. If pin 9 reads 0 V, it must be short-circuited to the copper strip that goes to pin 8 (`ref adj'). If pin 9 reads 1.2 to 1.34 V, it must be short-circuited to the copper strip that comes from pin 7 (`ref out').

All LED's light up:

Probably the chip itself receives a much too low voltage as its power supply on pin 3 (less than say 2.5 V). Maybe the stabilised voltage supply is adjusted way too low or there is something wrong with it (check supply voltage), or the wrong resistor is fitted as R1. If R1 has too high a value, the voltage drop over it will be too large, check whether it is 22 Ohm.

Another possible cause is that pin 9 (`mode') is wired up to the positive supply voltage (pin 3). Then the indicator will produce a `bar' display, meaning that all lower LED's will remain burning when a higher LED is switched on. This might look nice, but it is undesirable because the power consumption of the indicator will become too high.

Signal/reference voltage error:

Check for short circuits and loose connections between pin 1 to 9 on the chip.

Adjust stab. volt. supply to 12.0 V. Measure directly on the pins of the chip and work your way through the electrical circuit in the following way:

Reference voltages:

Signal in:
Measure `sig in' voltage at pin 5: Should be 1/11 of the supply voltage so ca. 1.098 V (with supply voltage adjusted to 12.08 V).

Measure again `low ref', `sig in' and `high ref' voltage. Make sure that the `sig in' voltage lies roughly in the middle between `low ref' and `high ref' voltage, vary the input voltage if it does not. Now the LM 3914 chip should light one of the middle LED's (say LED no. 3, 4, 5 or 6). If it does not, it is likely that the chip is faulty.

If it still doesn't work:

So far for the testing. If the indicator still does not work and you could not find out why with the testing procedure outlined above, refer to annex D.4: Electronic circuit. If you know how it should work, you can measure all over the circuit, find out where it deviates from how it should work and might find out how this can be solved.

Current compensation:

In the testing procedure, no test is included for the current compensation feature (see par. 5.5 and annex D.3.2: Current compensation). If indicators can be calibrated alright and a current shunt with the right resistance is fitted, it is highly unlikely that the current compensation would not work correctly. However, for getting to know the indicator design better, you could try it out once:

Un-soldering components:

If parts turn out to be faulty or a wrong part has been fitted somewhere, those parst have to be removed. This requires some skills:

Removing a soldered-on chip this way will be very difficult. Melting one connection at the time hardly helps because you can't pull this lead out even a little bit. You could try:

Of course there is an easier way of getting parts out. Professionals either use a vacuum de-soldering device that sucks up melted solder, or a special type of stranded wire that absorbs melted solder.


5.4.6 Calibrating indicators

Calibrating indicators is quite simple, provided that the necessary tools are available. Needed are:

After calibration, indicators should react to the input voltage as given in fig. 5.9.

Fig. 5.9: Relation between state of charge and open circuit voltage. Treshold voltages and reference voltages of the indicator are also drawn in.

Basically, the calibration procedure is as follows:

Then there are two precautions to take in order to reduce the error caused by temperature drift of the chip (see annex D.4.8: Temperature stability):

These precautions are especially important if no NTC resistor was fitted. It makes sense to check the temperature stability of a few indicators. This is especially worthwhile for the first indicators you have built, or if you are using chips that might be different from the ones you used before. Just heat up the chip some 10 or 20 C (by putting indicators in a heated box, or even by putting your finger on them if body temperature is substantially higher than ambient temperature) and check whether the treshold voltages change substantially.

Indicators that are calibrated wrong can have more damaging effects than indicators that do not work at all. Possible causes for calibration errors are:

index page