Tips needed for dissecting 4 layer pcb

[Kjelt]

Anyone have a good solution to seperate the layers from a four layer pcb ?
I want to reverse engineer the schematic for myself , private hobby use that is.
Normally with a bright lamp you can see a lot but this pcb has a ground plane inside, stopping the light 
It is no problem if the pcb "dies" just want to track the traces.
Thanks for any tips.

[Rerouter]

drop off all the components, scan the outer layers clearly,

then get a sanding block and get to work, you will eventually reach the first inner layer and can finish off with sand paper to expose it all,

You scan this layer, then continue sanding down to the last inner layer, when clear, scan and combine for tracing.

[jbb]

Sanding/grinding is the usual method.
X-ray might work with rather more effort (all layers will be stacked, lots of post-processing required).

Be careful of the fibreglass dust. It’s really nasty if inhaled. You need a good dust filter mask, safety goggles, gloves (basic is fine) and to thoroughly clean the work space after you’re done sanding. I think (but could be wrong) that a standard domestic vacuum cleaner won’t catch all of the fine dust, and will exhaust some out the back in a horrible dispersed form ideal for inhalation. I suggest the gentle use of dustpan and brush followed by a wipe down with a disposable damp cloth instead. Ideally, use a glove box.

Before getting messy, be advised that you can just beep it out one circuit at a time with a multimeter. Slow but tidy.

[Kjelt]

Thanks all, sanding had crossed my mind but hoped there were cleaner methods like some chemical to dissolve the glue between layers or something.

[Zero999]

A continuity test will do and could be performed in circuit, if the Ohm meter has a test voltage much lower than a diode drop.

What does it do? It might be easier to redesign it, than reverse engineer it.

[Kjelt]

A continuity test will do and could be performed in circuit, if the Ohm meter has a test voltage much lower than a diode drop.

All components have already been stripped, the pcb is not that large 

What does it do? It might be easier to redesign it, than reverse engineer it.

That might well be how it is going to end. It is not exactly clear what it does since it is out of its functioning habitat (machine), it is a microcontroller and ttl circuitry few opamps, logical inputs are a rotary encoder and uC in and outputs are unknown, probably some protocol for driving the motor a few steps and reading the encoder to correct any mistakes.
I hoped to find out what the protocol and all was by reverse engineering one pcb and perhaps re-use the other ones. If not than I'll design my own but I have little time for the hobby.
Since the uC is a pic16f84 which is very limited I was just curious how they solved the rotary encoder decoding in hardware to help the micro.

And I love puzzles so it was my Xmass puzzle to reverse engineer the schematic 
Pictures: https://www.eevblog.com/forum/projects/tear-down-of-an-assembleonphilips-pp-head/msg1288011/#msg1288011
Sanding pcb's is less fun 

[flolic]

Thanks all, sanding had crossed my mind but hoped there were cleaner methods like some chemical to dissolve the glue between layers or something.

It's not a much cleaner method, but using a wet sanding paper and working under water will get rid of a nasty dust and make sanding quite faster.
Another quick method is by using (CNC) mill/router to mill through the layers using carbide endmill.

Sent from my SM-G920F using Tapatalk

[lgbeno]

This seems like the type of thing you could pay a back alley shop in Shenzhen $20 to do for you


Sent from my iPhone using Tapatalk Pro

[Nusa]

Are you sure it's even necessary to sand that board? The board is translucent enough I can see the buried layer on the first photo fairly well, and that's with components still on. The flip side of the board should show the other buried layer, if there's anything more than a ground plane. Any vias and holes NOT connected to the ground plane should have circles around them that will pass light. So visual inspection combined with some confirming beep tests may be all that you need.

[Kjelt]

Well I started the schematic and I'll see how far I will get, but the gnd plane is pretty big and does obfuscate lighting through however as you noticed the second layer can be seen here and there so perhaps I'll get 80% so I know what it is.

[taydin]

The epoxy that holds the layers together might loosen itself up at a certain temperature. Might want to try with another PCB first.

[rstofer]

The epoxy that holds the layers together might loosen itself up at a certain temperature. Might want to try with another PCB first.

Wouldn't the plated-through vias hold the board together?
As a matter of fact, it might be difficult to figure out on which layers a given via is connected.  Or a via might connect on multiple layers.

[phil from seattle]

Unless it's a really complex board, continuity check is going to be a lot easier than sanding it down. 

As to the quad encoder, those are very easy to do in software. They might use interrupts or polling but either way, it's not much code at all. Lots of example code out there.

[Kjelt]

Unless it's a really complex board, continuity check is going to be a lot easier than sanding it down.

Started with that today see how far i'll get 
The first critical discovery was that the encoder pins do not make contact with any component at all, it runs straight to the header exit pins that go to an unknown board   
So for my encoder hardware solution that seems to be out of the window, but still i am intrigued what this pcb is for then.

As to the quad encoder, those are very easy to do in software. They might use interrupts or polling but either way, it's not much code at all. Lots of example code out there.

Thanks but I know, I have to see how much time i get left if i do an all sw solution,  but with 1000 pulses per rotation and 3000rpm i think that i might run out of processing power 

[Prime]

You can strip soldermask with heat and caustic soda. The uncured stuff is cleaned off with baking soda during the developing stage (potassium carbonate or sodium carbonate) but I know from experience making PCBs that you can strip the cured stuff using oven cleaner and heat.

Get the boards nice and hot in the oven, put on a sacrificial surface outside, and spray with ZEB or a similar aerosol oven cleaner. Scrub, rinse repeat.

Sent from my HTC One M9 using Tapatalk

[phil from seattle]

Thanks but I know, I have to see how much time i get left if i do an all sw solution,  but with 1000 pulses per rotation and 3000rpm i think that i might run out of processing power 

Yeah, could be a problem. Maybe some glue logic...

I was looking at the the Atmel ATTiny 1617 datasheet and it allows some custom logic that operates outside of the processor itself.  It's called custom configurable logic and has two channels that support up to 3 inputs with a software definable lookup table. Might be enough to do a quad enc to get step and direction.

[BurningTantalum]

Many years ago, my colleague and I wanted to de-pot a circuit cast in slate-loaded epoxy. A passing sales rep gave us some samples of liquids that "would dissolve anything". The one intended for epoxy did a real job- we left it immersed over the weekend and were confronted by a container of gloop sediment. The pcb was just a wavy mat of glass fibre with copper track floating on it. (Unfortunately the chips and transistors had also reduced to just silicon on legs!)
I have no recollection of what this chemical was, but I assume that it would be of use if still available.
Happy New Year, BT

[Gregg]

Many years ago, my colleague and I wanted to de-pot a circuit cast in slate-loaded epoxy. A passing sales rep gave us some samples of liquids that "would dissolve anything". The one intended for epoxy did a real job- we left it immersed over the weekend and were confronted by a container of gloop sediment. The pcb was just a wavy mat of glass fibre with copper track floating on it. (Unfortunately the chips and transistors had also reduced to just silicon on legs!)
I have no recollection of what this chemical was, but I assume that it would be of use if still available.
Happy New Year, BT

To dissolve most plastics, Dichloromethane (AKA methylene chloride) is probably the fastest solvent.  It is the active ingredient in many paint removers.  Most of the paint removers are a gel to keep the Dichloromethane from evaporating too quickly and to make it difficult to use the solvent to make illegal drugs and their derivatives.  It may be hard to find Dichloromethane pure liquid as it may be banned is some locations due to the chlorine constituent like many refrigerants. It evaporates more quickly than acetone.  You may be able to peel off layers with the paint remover gel variety by putting a thick layer on the board and sealing it in an airtight glass / metal container overnight or longer.  Dichloromethane won't dissolve the fiberglass or the copper traces, but it is hard on skin.  Nitrile gloves will last longer than most other synthetics, and natural bristle brushes will last better than synthetic.  Wear glasses and probably a active charcoal respirator rated for paint fumes. 

[Prime]

Many years ago, my colleague and I wanted to de-pot a circuit cast in slate-loaded epoxy. A passing sales rep gave us some samples of liquids that "would dissolve anything". The one intended for epoxy did a real job- we left it immersed over the weekend and were confronted by a container of gloop sediment. The pcb was just a wavy mat of glass fibre with copper track floating on it. (Unfortunately the chips and transistors had also reduced to just silicon on legs!)
I have no recollection of what this chemical was, but I assume that it would be of use if still available.
Happy New Year, BT

To dissolve most plastics, Dichloromethane (AKA methylene chloride) is probably the fastest solvent.  It is the active ingredient in many paint removers.  Most of the paint removers are a gel to keep the Dichloromethane from evaporating too quickly and to make it difficult to use the solvent to make illegal drugs and their derivatives.  It may be hard to find Dichloromethane pure liquid as it may be banned is some locations due to the chlorine constituent like many refrigerants. It evaporates more quickly than acetone.  You may be able to peel off layers with the paint remover gel variety by putting a thick layer on the board and sealing it in an airtight glass / metal container overnight or longer.  Dichloromethane won't dissolve the fiberglass or the copper traces, but it is hard on skin.  Nitrile gloves will last longer than most other synthetics, and natural bristle brushes will last better than synthetic.  Wear glasses and probably a active charcoal respirator rated for paint fumes.

Won't acetone do the same thing?


You might be able to use an aerosol bug spray with Deet. It dissolves certain polymers.

Sent from my HTC One M9 using Tapatalk

[bson]

Is there anything in the inner planes other than ground and power?  If not, removing any tenting and sounding it out with a continuity tester sounds much quicker and easier, and less destructive and potentially hazardous , than trying to peel it...  (In fact, in the time since you posted I'm sure if you had started then it would have been done already. ))

[Kjelt]

Is there anything in the inner planes other than ground and power?  If not, removing any tenting and sounding it out with a continuity tester sounds much quicker and easier, and less destructive and potentially hazardous , than trying to peel it...  (In fact, in the time since you posted I'm sure if you had started then it would have been done already. ))

It has been done 
https://www.eevblog.com/forum/projects/tear-down-of-an-assembleonphilips-pp-head/msg1388295/#msg1388295

Still some vias were a PITA luckily it was a small pcb.

[amyk]

Many years ago, my colleague and I wanted to de-pot a circuit cast in slate-loaded epoxy. A passing sales rep gave us some samples of liquids that "would dissolve anything". The one intended for epoxy did a real job- we left it immersed over the weekend and were confronted by a container of gloop sediment. The pcb was just a wavy mat of glass fibre with copper track floating on it. (Unfortunately the chips and transistors had also reduced to just silicon on legs!)
I have no recollection of what this chemical was, but I assume that it would be of use if still available.
Happy New Year, BT

To dissolve most plastics, Dichloromethane (AKA methylene chloride) is probably the fastest solvent.  It is the active ingredient in many paint removers.  Most of the paint removers are a gel to keep the Dichloromethane from evaporating too quickly and to make it difficult to use the solvent to make illegal drugs and their derivatives.  It may be hard to find Dichloromethane pure liquid as it may be banned is some locations due to the chlorine constituent like many refrigerants. It evaporates more quickly than acetone.  You may be able to peel off layers with the paint remover gel variety by putting a thick layer on the board and sealing it in an airtight glass / metal container overnight or longer.  Dichloromethane won't dissolve the fiberglass or the copper traces, but it is hard on skin.  Nitrile gloves will last longer than most other synthetics, and natural bristle brushes will last better than synthetic.  Wear glasses and probably a active charcoal respirator rated for paint fumes.

Some of the other liquid chlorocarbons will do the same thing, although whether they're any easier to acquire is an open question.

[Gregg]

Won't acetone do the same thing?
You might be able to use an aerosol bug spray with Deet. It dissolves certain polymers.
[/quote]
Acetone and Dichloromethane are entirely different solvents.  Acetone may work but would take weeks of soaking to dissolve the same amount of epoxy that the Dichloromethane would dissolve in hours.  Bug spray would only work if the propellant which is very volatile attacks the plastic component of the board.  As Amyk stated there are other solvents that would work, but obtaining them may be very difficult.  Back in the 1970s they used to use vapor degreasers  to clean silicon wafers that used chlorinated hydrocarbons, but they've been banned for many years. 

[Omgitskillah]

I think the best way to go about this is to use a multimeter to check continuity between various points. But for you to quickly figure out the points to test, you will need the datasheets for all the components on the board. That way you can see example circuits for each component and easily deduce how the components are connected together and draw a schematic from it.
Create a schematic with a pencil on plain paper and iterate it every time you learn of a new connection. You may also use a proper schematic capture tool but that would most likely force you to make your own footprints/parts to match.
SOme good news is that most designers will avoid having important connections in the middle layers especially if it's only 4 layers.

All the best!

[Md Mubdiul Hasan]

Anyone have a good solution to seperate the layers from a four layer pcb ?

]

PADS layout can give you nice way to separate your layer plane wise.