cable tester

I've just put 1% resistors on both ends to make a zig-zag series string. Check that with a good DVM. That will catch most errors.

Or a rotary switch/battery/resistor (or pushbuttons) on one end and N led's on the other.

The trick is finding shorted leads. The best tester I've had for cables (ribbon, etc.) created an algorithm of a good cable, then used it to verify other cables. The down side was you had to program it each time you turned it on - battery life isn't great I think. Old tech. But it checks up to 80 pins/leads for continuity, cross shorts, etc..

There are many others out there. The nice thing about the Assman is you can plug in home-made adapters to the 80 pin ports on the rear.

John :-#)#

Doesn't a terminated cable mean it has resistors built in? Wouldn't that cause another pin to be excited?

When checking pins for "no signal" you could add a resistor to ground with a matching value resistor and verify half the voltage. So each pin will see one of three values, zero signal, full signal or half signal.

Stimulate with a 3 Vpp signal to make sure the terminating resistor wasn't actually a diode by mistake. I had that happen once where the internal termination in a cable wasn't a resistor.

Terminated witha a connector, as opposed to bare wire.

I bought a cheap Chinese RJ45/RJ11 cable tester about 10 years ago and traced the circuit. I haven't used it much but it worked when I tested it. It uses a counter to scan each line in turn. It's like this one

It doesn't tell you when you have mixed-up pairs, like in an Ethernet cable where you have a blue/orange pair and an orange/blue pair.

You need a tester to show you the cross-talk for that.

CH

The aliexpress ads are ok, but when I checked for this on eBay I found mult iple people saying they crap out after just a few uses. I guess you can't make decent connectors that cheap. The eBay listing shipped in US and also included a pouch while the aliexpress listings seem to have much longer th an usual deliveries. One was 30 to 50 days!

If they were any good I would order a couple. But if it's going to crap ou t after a few uses, why bother shipping trash half way around the world?

There's always that factor of reliability with cheap Chinese products and that of long shipping times when you order from AliExpress unless you opt for an expensive courier delivery. But you can't beat the price and some of their products are very good. It's a matter of personal choice.

For example, I bought a true RMS 20,000-count DMM with all the bells and whistles and a claimed DC accuracy of 0.05% for US$28 in 2018, and a 6000-count model for $14 before that. I've used Chinese DMMs exclusively for the past 25 years and they last at least 5 years with heavy everyday use with the occasional drops and other abuse.

In any case, I offered the schematic to the OP in case the idea's useful for adaptation to his requirement.

I think modern Ethernet doesn't care.

The resistor thing usually does; the snipped LED thing certainly does.

It's "the 555 and CD4017 circuit" driving LEDs both ends of the cable. for this application the CD4017 data sheet explains to to extend the circuit to more than 10 outputs.

To automate it you could put 4017s at both ends of the cable clocked with the same clock and with a reset arrangement to synchronise them

(Q0 of the master 4017 to reset on the slave )

but only power one of them, Any significant current on the ground or vcc pins of the powered 4017 will indicate crossed or shorted wires

The unpowered 4017 will have voltage on the VDD pin (via the protection diodes on the clock input or the MOSFETs on the outputs) if that goes away there's an open circuit

A diode arrangement could be used to maintain a lower VDD if you're worried that the unpowered 4017 might lose its state. (I'm not)

On the contrary, it cares more than ever before.

The four pairs all have different twist rates to cancel cross-talk between the pairs. If you make two wrong pairs by splitting two pairs then neither pair will work, and both will interact with the other two pairs. You might get 10MBit out of your GB Ethernet, but you're unlikely to get 100, and you definitely won't get GB speeds.

CH

So buy good cables.

multiple people saying they crap out after just a few uses. I guess you ca n't make decent connectors that cheap. The eBay listing shipped in US and also included a pouch while the aliexpress listings seem to have much longe r than usual deliveries. One was 30 to 50 days!

p out after a few uses, why bother shipping trash half way around the world ?

Yep, I bought a programming adapter for $25 off ebay or aliexpress, don't r emember which, they are largely interchangeable. The one from the FPGA mak er is $200 now. The $20 one works great!

But good equipment is definitely the exception. So you pay your money and take your chances. You can always dispute a charge on the credit card.

I like eBay a lot better because their dispute policy works well. I bought several flash sticks on aliexpress and as each one arrived I tested it onl y to find it was fake. I got a full refund on the first. A partial refund on the second. The third they demanded I video the 12 hour test and send it to them, so I got zip. Had to use the credit card dispute to resolve th e last two.

From now on it has to be a pretty big bargain I can't get on eBay for me to use Aliexpress. That's rare.

I'm definitely not going to buy crap devices from aliexpress.

The issue is whether the wires are on the right connector pins. Pins 3 and 6 should be a twisted pair. You have to sense crosstalk to see if pins 3 and 6 are on wires of two different pairs.

What it doesn't care about is when a twisted pair goes from pins 1-2 to pins 3-6 at the other end, and 3-6 to 1-2.

Right. No simple tester is going to find pair crosstalk errors.

I did a *lot* of TDR/TDT testing on various cat5/6 cables, for a system that mis-applies them to, basically, SPI data at unreasonable distances and rates. They are all over the place.

Some people even vary the twists on cables that have individually shielded pairs! That adds prop delay skew for no reason.

It's a mess to use cat-cables for anything fast except ethernet itself, which usually does active equalization and doesn't care much about skew.

Don't do that, and don't expect any simple tester to report that. It would have to accurately measure 3-terminal mutual capacitances or something. UTP would be especially tricky. Are there affordable testers for that?

If a cable in your network doesn't work for whatever reason, throw it away.

I wonder when such a wiring error would really trash ethernet packets. For some lengths and rates, it might not.

What I described is a simple crossover cable. (Xmit pair to Rcv pair and Rcv to Xmit.) Any cheap tester can report that. I have a $5 one that does. But it can't detect split pairs that Cliff described.

You must be describing something else but what?

I saw a Cisco IP phone (the top brand after all) that would not connect with a single intermittent wire in the brown pair, which is an optional pair. It tried to connect at 1 Gbps with 4 pairs and when it failed it would not revert to 100 Mbps on 2 pairs. So I disconnected the blue and brown pairs and it worked fine with only the other two.

I was thinking Cat5/6 ethernet cable, with three pairs. Same problem of detecting switched wires that preserve DC continuity.