48 Signal generators

Hi
I am currently working on a project where I will need to create 48 different numerical signals. This is to test out network points on old buildings.

So what I would need is 48 individual signals that can be read and traced back to each signal.

My most promising idea is to create 48 different signals in the 1,2,3,4....50mhz range say then attach a cheap signal reader at the other end. You would then see a signal of 30mhz and then be able to trace back to a specific network point in the building

I was wondering if this would be possible? And if anyone has an idea on a very cheap circuit that can be made 48 times at each different wavelength?

Cost is also a big priority, you can buy signal readers for about £8 then hopefully less than £1 for each circuit.

Hopefully there is a more efficient way of doing this though

Thanks for your help

I'm not sure I understand your objective.
Are you trying to fully test a network cable system, for example to check for transmission errors at high speed, or, are you simply trying to identify which outlet is which?

If the latter, how are the cables/sockets distributed? For example do they all go to one central point (a hub maybe) with the other ends in different rooms.
If it's a hub and outlet scenario, there are easier and cheaper ways to do it. A spectrum analyser to find dominant frequency peaks is probably the most expensive solution!

Brian.

Hi
Thanks for your reply

Just trying to work out which is which. There is one hub then an outlet in each room.

What I'm hoping to do it plug into all 48 sockets in the hub. Then go to each room read the signal, i.e 30mhz then I know which socket I have plugged the 30mhz signal into. So I would then be able to map out the building.

If this design works out hopefully it would be able to be used on lots of other buildings as well.

Thanks

The simplest solution would be a single oscillator and a divider chain. Start with say 4MHz, keep it fairly low unless you want problems with line impedance as it will not be a differential signal. Connect it to a pin on the plug at the hub end, any pin will do but make it simple by using the same pin on all cables. Then use simple divide by 2 circuits to produce 2MHz, 1MHz, 500KHz and so on until you have a signal for all 48 cables.

Now at each room, look for the signal and measure it's frequency or period. From that you can tell which stage of division it was produced from and hence which cable it connects to. I strongly advise you do this in stages, maybe driving 16 cables at a time then moving on to the next 16. The reason is that after 48 stages of division, even 4MHz would be reduced to 0.000000014Hz or about one cycle in 2.5 years! Splitting the problem into 4 lots of 16 cables will only reduce 4MHz down to 61Hz which is a bit easier to measure. :grin:

If you do it using your original plan, the signal generation end would be quite complicated and any cross-coupling, ringing or impedance mismatching could create a wide spectrum of signals at several, possibly all rooms and you would need signal analysis equipment to find the fundamental.

Having done something similar on a large hotel TV distribution network where all cables were inside walls but led to a central control box, the cheapest way was to enlist help from a friend and use a battery at source and a lamp for the room end. We used walkie-talkies to communicate although mobile phones would work just as well. You just need one socket connected to a small battery at one end and an LED wired to a socket at the other. Plug the LED in each room in turn. It only takes a moment to swap the battery from one cable to another until you get a shout from the other end that the LED has lit up. Label that cable with the room number and put it to one side so it doesn't get checked again. We did around 100 rooms in about 2 hours.

Brian.

Hi Brian
Thanks for your help.

Hmm your solution seems a fair bit easier. I was hoping that this could be used on future jobs as well so would be good to try and build a little machine.

Another idea was to buy a little Arduino and get that to send out 48 different binary signals of numbers. Would need to build a little LED display that could show numbers from binary code though. Does this seem possible?

Thanks

Its certainly possible, the biggest challenge is handling 48 connections at once.

There are many ways to do it, if you want full automation and a numeric indication I would suggest a simple serial protocol for signaling, something like a half second 'high' across all the connections to sync the receiver followed by a single 8-bit data byte carrying the cable number at say 300 Bauds. The low speed should eliminate the need for cable drivers and in theory you could address 256 different cables. The maximum time to discover 48 cables would therefore be around 30 seconds but most would be much quicker. You would plug a 'detector' box in at the remote end and within 30 seconds or so it would display the cable number on an LCD or LED display. If you use LCD with it's lower power consumption it might even be possible to power the detector off the cable.

The difficulty is in multiplexing the signal at the sending end. You need 48 outputs and therefore some kind of 'port expander' to get enough individual signals. I'm thinking the simplest circuit would be a 3 to 8 decoder (74LS138 for example) with the 6 of the outputs each selecting one of 8 octal drivers, that gives you 48 addressable outputs. All the bit inputs would be in parallel and driven by another 3 to 8 decoder. Doing it that way means you only need 7 data outputs to address and send to all 48 cables.

There is a faster way to detect the cables if you want to be ambitious, it's a variation of the method above but the bits of the cable number are sent in parallel to the drivers so they all carry the number simultaneously. It reduces the detection time to less than 1 second but is considerably more difficult electronically and in software.

Brian.

Hi,

digital:
Maybe use a UART signal.
Send out data "0" and MUX it to port "0"
Send out data "1" and mux it to port "1"
and so on.
Known data rate, 8N1.
Then connect a microcontroller (+ UART + display) and see directely the port number.

***
Analog:
Or generate 6 + 8 = 14 different frequencies.
Then build a 6 x 8 matrix. 6 rows and 8 columns. Each cable is connected to a unique crosspoint with exactely two frequencies on the line.
It is like an extended DTMF.
--> it´s more easy to generate and detect 14 different frequencies than 48 frequencies.

Klaus

Klaus's digital method is the same as mine but it's the 'mux' that causes the problem. Assuming Chesterthompson doesn't have access to programmable logic devices or SMD assembly equipment and will use DIP devices, I think it can be done with two 74LS138 and six 74LS244 devices. Some care would be needed to apply the UART signal to the line but given that the speed should be very slow, it can probably be done by tri-stating the 74LS244s if pull-up/down resistors are used to establish idle state.

Brian.