The process proved to be very long and painstaking, but also very rewarding. The steps presented here are not exact, and you will no doubt find your own way of working which is more effective in the case of any particular board. However, I believe the basic procedure is essentially correct for the majority of boards which are too complex to reliably hand trace. Without the need to invent the process, things should progress more quickly.
For simple boards, it is probably better to use pencil and paper to draw the board onto squared paper, replacing components with their symbols, and changing track crossing places to avoid components.
For complex boards, it is too easy to lose your place using this method. The next "level" of complexity from a simple, drawable board, would probably be a good candidate for the tracing methods described in step 16 of this instructable, missing out all the image editing procedures, and modified accordingly.
At a higher level of complexity, as on the board shown in this project, I decided there were too many tracks for hand tracing to be reliable, and used the graphical techniques described. These methods arise partly from the photographs I took being too unevenly lit and inadequately sharp. Better photos means less work.
All the editing was done on a laptop running Linux. Windows versions are available of the main tools, Gimp, Inkscape, AutoTrace and Dia, but I don't know if a Windows version is available of the utility pstoedit, needed for an unfortunate extra workaround step. I'm sure another workaround can be found, if it isn't.
I also discovered as a result of this project that completely hidden traces can exist - in my diagram there is a transistor with it's base apparently unconnected, but I can't find a connection or through hole for it. I assume it's completely hidden in some way.
I'm pleased to say that after creating the diagram, I was able to use it to confirm the suspected faulty part.
Addendum: since writing this instructable I have discovered there are such things as "blind vias" - where the hole only goes part-way through the pcb and is connected to an inner layer only. It would appear that the board I worked on here utilises these, as a number of components appear to be missing connections. I believe, unfortunately, that these are hidden in the SMD pads themselves, so impossible to find without de-soldering. So if you end up with an incomplete looking schematic, that could well be why. Google for "blind pcb vias" to learn more.
Remove these ads by
Signing UpStep 1: Tools and equipment
You need:
* A good camera, preferably with a tripod
* Even lighting - a badly lit picture will cause a lot of extra work. A bounced flash is only just good enough. A ring flash may be your best option if you have access to one.
* A computer
* Internet access, to look up components (but see note)
* A multimeter - not essential, but helpful (depending on how precisely you need to know how the circuit works, if you are dealing with SMD's, the ability to measure capacitance could prove useful)
* A strong magnifying glass - you may need to check details which you can't make out from the photos.
* A photo editing program. I use the Gimp, which is free, and these instructions are Gimp-specific. I also used the vector program, Inkscape, to clean up the lines.
* A program to turn the photo into a vector line-drawing. I use AutoTrace, which is free. I chose AutoTrace specifically because it has a "centre-line" option, originally designed to help pick out text in an image, but ideal for this job. I found AutoTrace picks up too much detail to be useful for general cleaning up though.
* A program to manipulate vectors. I use Inkscape, which is free, and these instructions are Inkscape specific.
* A program to draw manipulate diagrams. I use Dia, which is free, and comes with an excellent set of component symbols (but see caveat, in step 18)
* As the current version of Dia has a problem importing svg files, and an intermediate program, pstoedit, was needed.
* A graphics tablet is a big help, I discovered, especially for hand-tracing some of the tracks.
* Lots and lots of time and patience...
A note about finding datasheets on the internet.
It's an absolute blessing that data for components is widely published on the internet. However, component markings are not necessarily complete or clear, and you should study the context of the device to make sure you have identified the device correctly. For example, on the board I used in this instructable, there is an 8 pin chip identified "7101". I found datasheets for an opto-isolator, an op-amp, a voltage regulator, and other devices before discovering that is almost certainly a DC/DC converter, confirmed by an inductor being connected as shown in the example circuit. SMD components have markings that may bear no resemblance to the part number. There are several guides published on the internet, this is one of them: http://elektronik.googlecode.com/files/SMD_Catalog222.pdf





















































Visit Our Store »
Go Pro Today »




How can I get a photo of internal layer of PCB? How can I break PCB to layers
On the other hand I have used GIMP in conjunction with a flat-bed scanner quite successfully for measurement purposes. Scanning at 1200 dpi gives an accuracy slightly better than 0.001" and using the measurement function of GIMP dimensions can be determined quite accurately.
All in all a good instruct-able given a few caveats.
The components and PCB have been potted with some compound. How this compound can be desolved.?
http://www.instructables.com/answers/Dissolving-Epoxy/
http://kevtris.org/Projects/votraxpss/unpot.html
And there are a couple of other pages on the internet about various attempts.
Black = 0, Brown = 1, Red = 2, Orange = 3, Yellow = 4, Green = 5, Blue = 6, Purple = 7, Grey = 8, White = 9.
First two bands are the base value, 3rd band is the multiplier (ie, number of noughts) so brown, grey, red woudl be 1800 ohms, or 1.8K, brown, grey, black would be 18 ohms, brown, grey, blue would be 1800,000 ohms, or 1.8M.
The fourth band is tolerance, used to be gold or silver for 5% or 10%, but now red or brown for 2% or 1% is common. There may be also a band to indicate temperature coeffecient. Gold and silver are sometimes also used for the 3rd band to indicate divide by 10 or 100.
At first I thought you had one of these and wanted to clone it, but it looks like you just want to build your own version from the photo of the one they are selling! Cheeky :-) Unless you can program your own microcontroller, your best bet is to find a free circuit diagram to build one yourself, or better yet, design your own, rather than try to copy a proprietary design from a photo.
I'm sorry to say there just isn't enough information in a photo like this to be able to work out any more than the basic power control - it has the triacs connected to the lamp outputs, these are fed by what look to be high value (and possibly special purpose) resistors from the microcontroller. It looks like there is a very basic live power supply to give the low voltage DC. There may of course be SMD's underneath the board, which of course you can't see, though it looks unlikely.
I hope this is helpful to you - wish I could give you more useful information.
http://www.instructables.com/id/Find-schematics-wiring-diagrams-etc-for-everyda/
FYI, I just added a reference to your Instructable from my video page as well.
- John
Guys, to take a photo of PCB, any regular printer scanner seems to do the job.
I use polarising filter to contrast the text on IC.
Also good to have microscope, backlight for details of PCB etc.
cheers
You can use the clone tool by finding a spot that looks the same as the place you want to correct, ctrl-click (I think, you will have to check) that spot, then paint over the place you want to correct. It will copy the first spot over the second. Use blur for if you have a small detail you want to remove. Just paint over the spot until you have blurred it enough.
Looking at your picture, I would seriously consider hand tracing it and missing out most of the graphical techniques, since it's a fairly simple board. Also, as mentioned in a previous comment, if you have a scanner you will probably get a better image by scanning the board.
Good luck!
+10,000 points for doing it all with open source software on Linux. There's so much amazing stuff that people don't know about. I've been using Inkscape and Gimp for a while, but I haven't seen AutoTrace or Dia before. TIme to hit the repositories!
*Revealing the PCB holes with masks and channels seemed magical, even if it's just math.
I accidentally reversed the polarity when I replaced the car power supply and let the blue smoke out!
As you all might know, electronics works with blue smoke ? If you let the blue smoke out, the gadget does not work anymore.
Be careful to do this reverse-engineering as some company might want to erase your memory, lol!
I've also been surprised that there's no solid standard for making vector graphics portable and platform-independent. I had done work in AppleWorks and had absolutely no way to get that data into OpenOffice's drawing package. I got a weak copy via some convoluted Encapsulated PostScript; SVG gave me nothing but false starts and dead ends.
That reminds me ... I have to get back on that project!
http://gimp-win.sourceforge.net/
If you have a sparse 2-sided board or a single sided board it can make the traces much easier to see.
Also high voltage and low voltage sections of a circuit will usually be separated on the PCB and it is very easy to see this if you light it from the back. I did this and found I could simply break the CFL high voltage part off of the board in the second picture (I replaced it with LEDs)
I am very curious to know how many hours you would estimate (honestly) that you put into this board.
The reason I ask is because I am the same way - when I want to achieve something I won't be stopped by difficulty or time expense to get something like this done. (Lately, it has been 3D technical/mechanical drawings in Sketchup that I put a frighteningly excessive amount of visual detail into, spending 10-12hrs/day for 10-14 days to complete, and think nothing of it) Perfectionist much??
It's refreshing to see that there are others that will expend extreme effort, ESPECIALLY when it means the difference of whether or not an electronic item gets discarded.
An alternative method that I thought I'd mention might be good for those who are doing simpler boards. I have used it with great results on many occasions, but I do realise that it's not for everyone:- -
Using a variety of differently-coloured markers, paint the traces as you follow them along the board. Use the same colour along a trace on one side over to the other when encountering a via. Drawing the parts on paper or computer, you can roughly wire point to point as you determine trace routes. When you have drawn a given trace, simply paint over it with black to indicate it is completed. The famous Sharpie markers come in more than two dozen colours, although there will be some that won't show up well on the board. These are "permanent" and hold up well while handling, but a bit of acetone, alcohol, lacquer thinner, WD-40 or other solvent will easily remove the ink. Of course you can use certain colours exclusively for Ground, Power, Data etc. and a million other ways of optimising the process to your taste. It's actually kind of fun too.
Thanks for the 'ible, there are many valuable methods and techniques in here that can be used for other purposes too. Also, the SMD Catalogue reference is a gem and exceedingly useful.
Cheers from Canada!!
Hopefully version 0.98 of Dia will import svg files correctly and there will be no need for the extra postscript step. On the other hand, using postscript does open up a wide range of alternative formats to use. You takes your choice...
For those who are interested in how long it took to do this, it's hard to say. I did a number of multi-hour sessions, but had several false attempts along the way where I had to try a different approach, or re-start a step. I also wrote down these steps and produced illustrations as I went, which took up more time as well.
So I'm making a very rough estimate here, maybe 30 to 40 hours including time to work out the methods. Now the method is known and written down, it should be quicker.
SIMPLY AMAZING ! ! !
You taught me alot!