Looking into the abyss

[dfidler]

Hi all,

Further to the thread (What is this stuff?), I'm taking apart an old amp to see how it ticks and I've hit another problem.

I've removed the output stage from the heatsink and have started tracing the board to find out that the PCB is three layers (DOH). Both of the outer layers cover the majority of the board so I can only see the odd connection in the middle layer when holding it up to the light.

Has anyone got any suggestions on how I can look through the outer layers and see that middle one (without an xray machine; I left mine in my other jacket )?

I thought of maybe building a lightbox with a diffuser and popping my two high-powered flashes in it, turning off the lights and doing a remote trigger. However I did a couple of test flashes and, to my eyes, I couldn't see any real detail through the board.

I'm also scared of hitting this thing with too much radiation as I don't know what it will do to the caps, transisors, smd resistors, etc.

Note: I realize that I could do continuity tests on all of the pins, just like I realize that I could stab myself in the eye with a live whose connected meter is set to measure capacitance. However, I'm not really keen on doing either if I can avoid it.

[ gross comparisons would be, for the approx 100 pins; n*(n+1)/2 = 100*101/2 = 5050 comparisons.... *stab* ]

Cheers,
Dave.

[dfidler]

Well, I can confirm that the flash thing doesn't work. It makes for some pretty pictures though

[dfidler]

Aaaaaah! My eye!

[marce]

Basicly no,
you've mentioned the time consuming and non destructive method of continuity checking, a pain and very time consuming.
The destructive way is to de-populate the PCB and top and bottom, then sand down to the next layers. Trouble is your left with no PCB, so best done when you have a couple to play with. The scanned images can be converted to gerbers, or used as templates to re-create the routing, this can then be back annotated to a schematic. A quite involved process, but one that works quite well, I am told.
Even with a x-ray set up it can be quite hard, you still have to manipulate the images and rebuild the design.
When I've reversed old designs, the first thing we do is recreate a library of the parts and footprints, then rebuild the layout copying the component placement.
I would suspect that the board is actually a 4 layer, as most boards tend to always have positive numbers of copper layers, this is due to both the etching process and to avoid uneven lamination stack up, when the final build is laminated. Un even copper layers have been done, but I cant recall ever seeing one. This isn't two bad usually as the inner pairs are often power and ground, so the majority of tracks are on the outer layers.
If you draw out the component layout, you can mark any chip supply pins, this cuts down the number of iteritive continuity checks. If I am doing it this i will also add connections to the layout as I go along, and keep a hard copy where I use different coloured pins to indicate connectivity (I colour the pins on the drawing).
However you do it its a pain.

[c2cthomas]

Hello Dave - look as if marce has put things out clearly - but I do have a silly question to ask. Uhhhh - do you know the make and model number of this thing? Is there a part or model number on the PWB? You might be able to find a schematic on-line with that information - or someone here might be familiar with it.

Where I worked we had a lab where we reversed engineered things of this sort and it is a time consuming process. I hope that the time you are going to invest is worth the effort!!!

[dfidler]

Originally Posted by marce

Basicly no,
you've mentioned the time consuming and non destructive method of continuity checking, a pain and very time consuming.

Aye.

Originally Posted by marce

The destructive way is to de-populate the PCB and top and bottom, then sand down to the next layers.... snip

Cool idea but too destructive for my tastes. This was intended to be a learning and confidence building exercise. Ultimately, I still want to listen to the amp when I'm done with it and maybe even re-bay it one day when I get tired of it.

Even with a x-ray set up it can be quite hard, you still have to manipulate the images and rebuild the design.

Yeah, I was wondering about that; the xray would be affected by the different densities of the components.

I would suspect that the board is actually a 4 layer

You're right, I found this out, also, while tracing.

This isn't two bad usually as the inner pairs are often power and ground, so the majority of tracks are on the outer layers.

Just the opposite with this one; the supply and gnd are the outside poors; signal is inside, but I know what you mean.

I use different coloured pins to indicate connectivity (I colour the pins on the drawing).

That's an interesting idea; I'm finding that when I get half way through tracing the input stage, I start to lose my focus and start migrating around the board to different pins and traces. Almost to the point where I've considered creating a matrix for the pins and recording the resistance to each...

Okay, so I only entertained it briefly.

[dfidler]

Originally Posted by c2cthomas

Uhhhh - do you know the make and model number of this thing? Is there a part or model number on the PWB? You might be able to find a schematic on-line with that information - or someone here might be familiar with it.

Yeah, I had a look for the schematic online, no love.

Where I worked we had a lab where we reversed engineered things of this sort and it is a time consuming process. I hope that the time you are going to invest is worth the effort!!!

I've already learned a ton playing with the circuit so it's already been worth it.

Cheers,
Dave.

[wakibaki]

It may be time-consuming, but writing a netlist is the proven way of discovering the information you want.

w

[marce]

I played about with a basic x-ray machine, for BGA inspection, but wasn't impressed. The images were grey scale, ghostly images, limited field of view, to much image manipulation. It was a basic machine though, and with a better one and some image manipulation software, it may be possible.
These days reverse engineering isn't worth the hassle, unless there is some gain. I've seen some interesting copies of products I've worked on, some actually improved! But with most comercial/industrial products having programmable devices, and the increase in software based signal processing and control over hardware, the funtionality of the product gets copied. Quite often with less complex technology, again I've seen products that look the same on the outside, with a double sided SMD PCB replaced by 3 PTH designs and ribbon cable! We actually EMCed it, it didn't pass.
The enclosures get 3D scanned and replicated. 3D cloud scanning and the right 3D cad package, a machining add on and your away.
Most reverse engineering done these days within firms these days is the updating of legacy CAD designs from Gerbers. This is quite automated these days as most good DFM packages, such as CAM350, can recreate the connectivity from the gerbers. The packages can be defined by selecting their pins, a libray of footprints is created. This can be then transferred back to a layout package. The layout can then be recreated and part information added. Then a schematic is created from the layout via back annotation. This is then tidied up, its like doing a PCB in reverse, all the symbols are jumbled up, with rubber banding connections, these have to be laid out logically! Then a EE has to check the circuit...
As you can gather it is a very time consuming task, and even a quite simple design can take up to a week. It tends to only be done where cost is almost no object, or where the cost of not doing it far outwieghs the cost of doing it; usually aerospace/military where 15 year plus lifetimes are the norm.
Have Fun