I grew up taking things apart and I see no reason to change now. So, I made fixtures to make it fairly easy to remove tops from TO-5 and TO-8 metal IC packages using a Dremel motor tool and diamond cutting/grinding tools. (I can also cut the tops off of TO-3 cases without needing any fixture).

I then built an adapter to mount my digital camera to one side of a stereo microscope. This allows me to take pictures of of the opened chips.

So far I have taken pictures of these:
DH0008 Solenoid driver (in TO-5 case)
HA2-2630-2 High speed analog buffer (in TO-8 case)
LM121 Precision amplifier (in TO-5 case)
PA10 Apex power op-amp (in TO-3 case).
RFPIC12C509AF 18-pin DIP PIC microcontroller with RF transmitter circuit
RFPIC12C509AG 20-pin DIP PIC microcontroller with RF transmitter circuit
C1702 2K EPROM (256x8)
D2732A 32K EPROM (4Kx8)

My original intent was to upload these photos to this site but they are too big. The size is needed to see the details such as the "bits" in the EPROM's. Since I don't want to downsize the pictures, I put them here:
http://s242.photobucket.com/user/ottosenpb/library/?view=recent&page=1

 

Warning! This link prompted me to install an unidentified piece of software..

 

Warning! This link prompted me to install an unidentified piece of software..

Oh, that is very bad. What software do you think it was

Note: I assumed that anyone motivated enough to look at the pictures would be willing to sign up to PhotoBucket. And, I understand if others don't want to sign up... I too hate signing up like this without significant motivation.

 

I'm using Firefox and I'm a photo bucket member.

I spent several decades working with hybrid circuits so it's nice to see some again, though they leave me itching to get a probe and pull up some of the downed wirebonds.

 

Very neat,
Adding few more for you, feel free to use them, it's from my account.
http://i.imgur.com/9cGjO1a.jpg
http://i.imgur.com/RWnS30e.jpg
http://i.imgur.com/QaRdkLs.jpg
http://i.imgur.com/kGTb65d.jpg

 

Warning! This link prompted me to install an unidentified piece of software..

Interesting. There was no prompt for me to install any software, and I was able to view the pictures on photobucket. Perhaps they were protected earlier?
Very nice pictures, by the way.

 

Interesting pictures, Richard. I had access to them without problems.

Steady hands required to solder those wires inside...

BTW, I recall seeing the bonding wires, IIRC in a 2N3055 I opened (by carefully filing the top of the cover) and they seemed much smaller in section than the terminals. Thus, what is all the fuzz about currents if they use so small ones inside? Yes, I know there is a smart answer to my stupid question.

 

Inside 6502 http://visual6502.org/welcome.html

 

You can get by with skinny wires inside the transistors because they are very short and their connections are welded (as opposed to the TO-3 2N3055 which needs a large contact area to make contact with a socket), thus, the skinny wires present only a very small resistance.

 

Steady hands required to solder those wires inside...

No solder required. The wires are forced down with pressure and scrubbed in by exciting the tip with ultrasonic energy. Each end of the wire looks different:

Typically the first side is a ball (the ball bond side) made by melting back some wire to the tip: a small flame or an electric spark melts the end of the wire and the ball naturally forms back on the tip.

The tip is then moved to the destination paying out wire, and this end is crushed down forming the “wedge” end. As the tip is withdrawn the wire breaks off at the connection and is melted back to form the ball for the next wire.

Obviously the machine is well calibrated to do these tasks. For manual machines the operator looks thru a microscope and adjusts the tip position with a micro manipulator: a simple hocky puck like disk that can me moved around, and the linkages reduce the large puck movement to small increments of the actual tip.

Or a completely or semi-automated process is used where the operator moves targets to the actual chip corners and header (package) indicators. Then the computer takes over and shoots down the wires extremely quickly.

BTW, I recall seeing the bonding wires, IIRC in a 2N3055 I opened (by carefully filing the top of the cover) and they seemed much smaller in section than the terminals. Thus, what is all the fuzz about currents if they use so small ones inside?

The wires internal to the package are very short. The smaller gold colored wires are actually 99.9% gold and very good conductors: 0.001” wire is good for 1 amp (fuse current). Larger wire is typically aluminum in .005 or .010” size: larger wires are available too.

Since they only travel a very short distance a smaller diameter is acceptable. A full inch of .005” al wire only has about .05 ohms of resistance, and these wires are much shorter than that.

http://en.wikipedia.org/wiki/Wire_bonding

 

Thanks for the details Ernie. I was kidding about soldering, (even more done by hand!!)

Just an additional question: what kind of motors do they use to move things at such a minute level of detail? VCMs?

Along the time and many posts, Bill Marsden explained some bits of what I understand is / was his job. He repaired those machine before becoming actual operatir. Am I right?

 

I spent several decades working with hybrid circuits so it's nice to see some again, though they leave me itching to get a probe and pull up some of the downed wirebonds.

Yes, I saw a lot of damage including broken wires. Never get these parts working again

I have been opening chips for years. Many, many years ago we opened a CCD460 memory chip that was used in a digital TV image memory. After opening it up we put it back into the circuit. The chip was light sensitive and we could see effects on the TV monitor when we changed the lighting on the chip.

Later, we convinced one of the engineers to give us a ceramic packaged Motorola MC6800 microprocessor to open up. We told him he couldn't trust it anymore because it had been in and out of circuits too many times for testing purposes.

After opening it up we passed it around for everyone to admire. After everyone had a look, someone asked if it might still work so we plugged it back into the PCB. It did not work. Then one of the technicians speculated that a bond wire might have been bent b all of the handling. When he put it under a stereo micoscope he actually did see a bent wire! At this point he asked someone to give him a "cobar". He used a 30 gauge wire-wrap wire to pry up the bent bond wire without breaking it.

After the repair, we put the part back into the board but it still did not work. Then someone pointed out that maybe it was light sensitive like the CCD memory chip. Sure enough, the microprocessor worked when it was shielded from the room light.

 

Inside 6502 http://visual6502.org/welcome.html

I am familiar with this. MUCH better pictures than mine! Good enough to reverse engineer the circuit!

 

Here are pictures of the fixtures I made:

The metal plates clamp together with the TO-5 or TO-8 can in between. This keeps the can from spinning when the rotating tool is applied to grind or cut away the top. An LH0032 is in the TO-8 fixture ready to decap.

The fixture on the microscope is for a camera that has a "flat" lens -- The lens does not extend out from the camera. The positioning of the camera is very critical. Even a very small misalignment ruins the picture.

Wow, I really need to clean those microscope lenses.

 

"Back in the day" we had a fixture to open "can" type devices we called the "can opener."

Sure enough, it had the same sharp disk you would find in any can opener in any kitchen anywhere, plus two holding (non sharp) disk to bank the part against.

There was some sort of crank to turn the part then using a screrw you would advance the sharp wheel against the part just like a plumber's pipe cutter.

I do not remember if we purchased the unit or it was a custom made device. I had fun opening every TO-3 I ever blew up.

 

I am familiar with this. MUCH better pictures than mine! Good enough to reverse engineer the circuit!

Maybe you also know for Sergei Skorobogatov work.

http://www.cl.cam.ac.uk/~sps32/mcu_lock.html
http://www.cl.cam.ac.uk/~sps32/phd.html

 

Great pic's! This reminds me of a visit to the muesum of holography in Chicago. About 1998 or 99, (now closed). I saw a hologram of a microscope. When you centered your eye over the eye-piece you saw a beautifully magnified IC. I wonder if I can hunt it down and see it again.