LAN Cable Tester Repair With FFC and a 4017

My employer supplied me with this LAN cable tester, which I don't use very often, but is extremely useful on the occasions that I do.

Unfortunately it had an encounter with POE, which killed some of the channels. So I decided to fix it.

Yes, I could have asked my employer to provide a replacement, but where's the fun in that?

This particular tester has a custom chip, with a soft on/off and mode button.

Investigation showed that some of the chip's outputs were completely dead. With no possibility of replacing the chip, what's a person to do?

Well, it turns out that LAN cable testers are often made using a 4017 1 to 10 decoder variant, and some kind of multivibrator.

So I decided the original chip can do duty as multivibrator (and provide auto/manual modes too) and I would try to piggyback a 4017 onto it.

• Broken LAN cable tester with custom chip
• MC14017 or other 4017 variant
• A piece of FFC. You get lots of this if you keep dismantling stuff.
• Solder
• Sharp craft knife
• Scissors
• Wire cutters
• Soldering iron (low temperature if possible)

My cable tester works in the same way as the majority of others, the exception being the custom chip that was used. The magic happens with the LEDs and diodes, however. Refer to schematic.

Ignore D19, R1 and C1 for now, they are part of the repair.

The tester works as a "1 to any" multiplexer. The logic 1 presented on each pin of the 4017 in turn drives current through the local LED, down the conductor being tested, through the remote LED, and back via the diodes in the path of any of the other conductors. This works because all the other output pins of the 4017 at at logic 0, and can sink current.

With the custom chip, the number of outputs was fixed at 9 (for 8 wires inside the cable, plus a shield, though it's rare to actually encounter one). The 4017 however has 10 outputs. Rather than get a dead beat, the tenth output is connected to the reset pin, which causes it to start again.

The clock signal (which I'm calling the "heartbeat" for the purposes of this Instructable) is normally produced using a multivibrator, which may be 2 transistors, a 555 timer, or something else. The custom chip used in my tester has a high frequency clock using a ceramic resonator, and produces the "heartbeat" as an extra signal to flash an LED (which I was able to use to drive the 4017's clock pin, with a slight modification).

In my case, this meant simply de-soldering and cutting off all the pins which were connected to outputs, since it's a custom chip there isn't really an option of replacing it. You might find that you have a faulty cable tester which actually uses a 4017 - in which case you can simply replace it, and skip the rest of this Instructable!

Make a note of where the pins are actually connected to so the correct holes are easy to locate afterwards.

The holes were originally 0.8mm. I had to enlarge them a bit. I used 1.0mm but things might have gone a bit easier if I'd used 1.1 or 1.2mm.

First get yourself a piece of FFC. I cut two pieces but with hindsight could have cut sections linked by the connection between pins 13 and 8 with the sections offset.

People will suggest various ways of cleaning the insulation from FFC, but I think the best way is to scrape across it with the tip of a sharp craft knife. It's quick, gives reasonably clean edges to the insulator, and does minimal damage to the conductors.

Scrape the insulation first, then trim the cable to shape with long bits to connect under the chip. Trim the insulation off the ends of the long bits by peeling it back with the tip of the knife, and clipping it off.

Using a soldering iron on a low temperature, tin the cleaned areas. By doing it on a metal surface you can minimise melting the insulation on the back of the cable. Position the chip with it's pins on the solder bumps you've created, and solder the pins down. Push them down with the tip of the iron as you solder. They will bend a bit, which is fine.

Trim the excess FFC away.

Starting with the connection from pin 11 to 15, bend the "extra" bits under the chip so they can be connected together. Use the tip of the knife to fold over the ends so they hold together, and solder them. Slip a bit of polyamide tape (Kapton is well known, Koptan is much cheaper) under the joint, then do the connection from pin 13 to 8 (if you didn't try the offset suggestion, above).

Put a small piece of polyamide tape around the whole thing. You can use some other kind of tape but it's less stable and won't be heat resistant.

Finally, using scissors, separate the conductors down to a little way from the chip, maybe 3mm or 1/8". Try not to drift too close to either conductor as you cut because the insulation can come off if you do.

The next step is to simply thread the FFC leads through the holes you enlarged in the board.

I didn't plan how I did mine at all and ended up with a couple of awkward ones at the end, so I recommend planning what order you will do the leads in!

Just thread them through at first, and make sure the power wires (Vdd and Vss) can reach to either the original chip's power pins, or some other points you choose, and the clock wire will reach the heartbeat LED (or again, some other point like the original chip's output pin. It just depends how your particular tester is set up.)

Flip the board over, peel back the insulation as close to the board as possible and trim or break it. When you solder the leads it will melt back a bit so it doesn't interfere with the joint. Trim each lead with wire cutters, as you go,

Connect the power and clock leads. There's no special reason for doing them last, it's just what worked for me.

I put a battery in my tester and turned it on, and it went crazy, racing through the connections instead of nicely following the heartbeat LED.

A bit of investigation with my oscilloscope revealed an extra signal in the heartbeat pulse. Since the switch is also connected to this signal it is presumably associated with that.

I made a filter using a diode, resistor and capacitor. (D19, R1 and C1 in the schematic). A 1nF capacitor in parallel with a 20k resistor nearly cured the problem. With 10nF it completely disappeared but also the switch became completely non-functional. I used 4.7nF but really could have done with a lower value.

1nF x 20k gives a time constant of 20µS, and 4.7nF x 20k gives 94µS, so the best time constant is somewhere in that range. (1 time constant is simply the value of a capacitor multiplied by the resistance through which it charges or discharges. It's the time taken to halve the remaining difference between the capacitor's voltage and the supply. You can easily see the value can never reach supply voltage, or zero. For practical purposes, 5 time constants is considered to be full charge or discharge)

It seems the circuit is loading the signal too much, so dropping the cap to 100pF and using a resistor in the 100's of kilohms (perhaps 330k) would be the way to go.

The capacitor is charged quickly via the diode, which blocks it from being discharged into the driving circuit, so it discharges slowly via the resistor which is in parallel with it. The clock pin is connected to the diode/capacitor/resistor connection. The other side of the resistor and capacitor are connected to 0v. The capacitor need to hold just enough charge to stop the clock pin being pulled down by that extra control signal.

After getting the circuit to behave itself, another problem became apparent. 2 of the LEDs had failed. The original LEDs are rectangular, but a couple of round 3mm ones work just fine.