This guide is aimed at any individual with some basic technical skills. It enables the reader to manufacture this device at low cost from common, non-specific parts readily available at common retail shops and using common household tools and several non-critical cheap electronic hobby tools that are easily available also in common retail stores. The device (Reasonable Optical Near Joint Access) allows fast, legal, unlimited, uncontrollable, and mostly confidential data communication free of any charge between two houses or flats that have direct optical visibility between them. The data communication is performed between two PC's and may include routing IP packets, maybe extending the reach of Internet into one of the two households.
Aiming of the device is very easy and intuitive and in no phase between going shopping and using the device it is necessary to borrow any specialized equipment like oscilloscopes, cameras, telescopes, walkie-talkies or similar. And no special operations like Printed Circuit Board manufacture are required - the device contains no PCB's. The device has been designed from the beginning to be as easily built as possible.
This guide discusses first some general characteristics of Ronja Metropolis which should help you to decide if it is really suitable for your intended application. Then there is a section helping in decision of some important technical details, like lens diameter and holder type. A shopping list follows, building guide, testing guide, installation guide, aiming guide, and at last a maintenance and troubleshooting is discussed.
Proceed in building on the following sequence, omitting the unnecessary holders:
If the device has been correctly soldered together, it should work on the first try. But it is possible that you have made a bug somewhere and the testing chapeter is here to help you find it.
I know it is annoying, but now check the topology of all electronics according to the schematics. Be sure all parts are on their place, none is missing, and all joints are reliably soldered together. If in doubt, reheat them to melting point and let them settle down again. If you omit this phase, you have little chance to debug the device without an oscilloscope. If you though have problems, you might ask at the mailing list. Link to the mailing list is accessible from top level of Ronja homepage.
There are several places in the schematics, where the voltages are described. If something goes wrong, try to check them with a multimeter and if they are out of range, either correct them changing the value of part indicated, or find the bug which causes them.
Insert the NIC into your computer and switch it on. Check the voltage on the AUI. If it is bigger than 13V, then insert some 3-Ampere rectifier diodes into the place specified on receiver schematic to keep the voltage down. Insert them in forward direction on the power wire from the AUI. AUIs on PC's have all 12V I think, although the specification permits anything between 12V and 15V, so this is not a real problem. Switch the computer on.
Plug the sole interface into AUI and the yellow light should come up, green light be undefined and stable, and red light not shine. If yellow light is not shining, there is probably some short circuit, so disconnect, measure the power consumption (if 500mA, then there is a short circuit), and find the bug. If red is shining, there is a bug in the logic.
Now plug it off AUI, screw the receiver and transmitter on their short wires into the interface wire nut, and plug it into AUI again. The transmitter should start shining. If not, find the bug. Then align the receiver and transmitter and set up your NIC to full duplex, use arp, ifconfig, route and tcpdump to setup a sniffing loppback test, and put some packets into the device. They should be seen back, with negligible packet loss, and the red and green lights should shine according to the packet density. If this doesn't work, watch the lights and try to debug the device (cold joint, bad topology, missing part).
Now "tune" the transmitter to 50% duty cycle wave. It is optional, but enhances the performance. Place a photocell over the transmitter and measure the DC current when no packets are going and when many long packets are going (the red light on interface is shining vigorously). The values will somewhat differ (probably) and you have to manipulate the value of the resistor that is between pins 14 and 1 of the triple chip in the transmitter. Increasing the value will give less light during the packets.
Now take a mirror and try to reflect the beam on as big distance as possible. It should be able to make something around one meter of total path length (say 70cm to 140cm). If it is suspiciously short, then the receiver is botched. Find the bug and if there is none, then there is a cold joint somewhere - reheat the joints again. If it doesn't help, do the same with the transmitter. Last chance is there is something bad in the interface near the wires that go to receiver and transmitter. If you can not find the bug, try to measure various values with a multimeter, and last instance is the mailing list of course or borrowing an oscilloscope.
Screw on the holders as necessary. Then put the pipes (optical heads) onto them and tighten. Connect the wires together on the wire nut inside the pipe cap, both on receiver and transmitter. Ask the other side of the link to install a retroreflector near their receiver (or on it) and to switch off the transmitter.
Connect the heating pair in your home to the suitable 12VDC wall cube and leave it on unless there is low humidity out there.
Now put one bag of sillica gel into each pipe (beside the box with electronics, behind the thermal shield) and close the cap and screw it down with the screws.
The picture shows what the transmitter beam looks like from 260m
distance (130mm diameter transmitter lens) and the big image is the
90mm transmitter in operation from close, out-of-axis view.
Now wait until dark (or at least twilight) and aim the transmitter onto some distant house. Loosen the focus (with #7 wrench) and focus until you see a bright spot. Then tighten. Move the transmitter onto the retroreflector and play with both M8 bolts (two #13 wrenches) and the focus until you get maximum brightness on the retroreflector. Then carefully tighten and watch the spot to be sure you have not drifted during tightening. Measure the received signal strength shortly before you leave the roof and write it down and put it into a safe place for future reference to be able to say if something went wrong.
Ask the other party to aim their transmitter. Then connect the DC voltmeter on 200mV range to the measuring port of the receiver. Then put the focus of the receiver into middle position and fuss around with the receiver until you see some numbers on the meter. Now play with the focus and aiming until you get maximum count. Now freeze down the setting by tightening the nuts down, again watch the meter not to decrease the count (which would mean corrupting the optimum aiming).
Now set up your IP network and do some pings, FTP's, ssh's, and so on, to be sure that it works flawlessly. Then climb to the roof again and seal all seams on the tube with the sillicone sealant to be sure no humidity or water will leak inside.
It is possible that the lenses will become dirty from the outer side. You may either spray them with a water with a drop of detergent (Ronja will be defunct for some short time after it because it will not see anything) or wipe them with some soft rag and be careful not to scratch the surface. Nevertheless, they must be really dirty for the signal loss to be observable as excessive packet loss.
Maybe the sillica gel will eventually become saturated from the trace humidity that leaks inside no matter how it is sealed. Then tear off the sealant around the cap, unmount the cap, remove the bags, put the into oven for half an hour until they are absolutely dry, and replace them, replace the cap and seal it again with the sealant.
All electrical contacts are made from wire nuts. If the device suddenly stops working after a long period of troublefree operation, this is probably the case. The oxidization together with temperature changes brought eventually two layers of copper oxide against them and prevented the contact. Loosen all wire nuts in the device and retighten them again. This will cut brand new touch surfaces and ensure reliable contact again.
If it stops working, check visually if there is a signal from the other side, and if there is, check the alignment with a meter. If the receiver is dead, then there was probably a cold joint that became a no-joint over the period of time undr the influence of temperature changes. Resolder the box thoroughly and neatly. Do the same if the transmitter stops to transmit light (first check the diode if it is not burnt out).
The LED has lifetime up to 100,000 hours, which is 12 years of 24/7 operation. The light output very slowly decreases over time. If you think it is shining too dimly, then replace the diode with another one. The diode is ran exactly at recommended maximum current. The LED lifetime is being exhausted when the device is powered - no matter if data go through or do not. Also the LED shines 50% more at -20 degC than at 40degC. This is somehow a compensation for the bad weather during winter and good visibilities during summer.
I congratulate you to the decision of building Ronja 10M Metropolis. I wish you a clean success and many years of reliable, uninterrupted operation.
Enjoy the Megabits.
(c)2001 Karel 'Clock' Kulhavy, e-mail address: clock at atrey dot karlin dot mff dot cuni dot cz