DACs - die pictures

[Noopy]

It seems that I have to create a DAC-topic. 


Let´s start with the Burr Brown DAC709!




"U.K." 


The datasheet gives you a lot of specifications:

Linearity Error: +/- 0,0015 %
Differential Linearity Error: +/- 0,003 %
Gain Error: +/- 0,07 %
Zero Error: +/- 0,05 %
Power Supply Sensitivity Vcc: +/- 0,0015 %/%Vcc
Power Supply Sensitivity Vdd: +/- 0,0001 %/%Vdd

Total Error Over Temp Range: +/- 0,08 %
Total Full Scale Drift: +/- 10 ppm/°C
Gain Drift: +/- 10 ppm/°C
Zero Drift: +/- 2,5 ppm/°C
Differential Linearity Drift: max + 0,009 / - 0,006 %
Linearity Error: max +/- 0,006 %

Settling Time (to +/-0,003%; voltage output): 4 µs
Slew Rate: 10 V/µs
Settling Time (to +/-0,003%; current output in 10Ω to 100Ω): 350 ns




BB used two dies so they didn´t have to integrate the logic on the same die as the analog part which probably would have degraded the performance of the analog part.
BTW: The Package has two interconnection layers.




I think BB was able to configure the logic die so it acts as a DAC707, a DAC708 or a DAC709.
The two bondwires connected over an isolated square are promising candidates.





That´s an big old gatearray! 




Nice test structures! 




The DAC itself.




Resistor, Laser, Resistor, Laser, Resistor, Laser,... 




Buried zener! Yeah! 


Some more pictures here:

https://richis-lab.de/DAC01.htm

[ali_asadzadeh]

Nice, pictures!

[Wimberleytech]

The test structure brings back memories.  I used to routinely put those on my chips in the very early days. I never did field transistors, however.  Here is one from circa 1985.

[Wimberleytech]

And this from '79

[Noopy]

Such test structures are often very interesting.
I was surpriesed how many test structures they have put on the gatearray!

[Miyuki]

I wonder is today chips also with this test structures ?
And are they used mainly in first tuning of manufacture - like all that diffusion times and temperatures. And then to keep track about process variation over time
Or have they any use in detection function defective chips ? As I expect to function defect are more result of microscopic contamination or crystal growth anomalies. Os is in real manufacturing problem with like material quality variation of all required chemicals ?

[Noopy]

Well this one for example is not extremly old: https://www.richis-lab.de/ADXL213.htm

It has a lot of test pads at the bottom of the die:



The test structures are sawed away but they obviously still use test structures.


My read is that they mainly use the test structures to tune the process or to identify which is the problematic process if yield goes down.
I don´t think they are interested in one defect chip. They simply throw them away. But if you have 10% defect chips in a batch you have to identify the problem quickly!

[SeanB]

Some of the test structures are used in production, IIRC some are plain resistors to measure the thickness of deposition of dopants, others are used to time the etching processes so that they can stop when the layers are removed, and others will be used in between to see that you got a viable wafer without too many failed parts, so that you can scrap it at a low price point, see why it failed there if all, and repair your machine. No good to make a whole wafer all the way through, and find out after you are done a week down, that a mask in step 3 of 5000 was misaligned.

[Gyro]

Buried zener! Yeah! 

I didn't realize BB were using buried zeners back then, NS must have had a patent on them. Maybe it had expired by 1990 or they licensed it.

[iMo]

It seems that I have to create a DAC-topic. 
Let´s start with the Burr Brown DAC709!

 
Such a beautiful chips, I would never ever decap them, I want them working!! 

PS: I wonder how much you are going to pay for the electricity when all the chips ever produced will pass through your oven 

[Wimberleytech]

Here are some wafers that have collected throughout my career.  Notice in the small wafers (early years) that there are "test plugs" on the wafer.  The structures on these plugs are for the mfg guys to characterize the wafer at least at final so that critical parameters (VT, sheet rho...) can be extracted.  These test plugs are not needed in modern wafers (the big ones) because they, instead, put the test structures in the kerf (between die).

The test structures I showed earlier were for my benefit (as designer).  If I had some anomaly when I got the chips back, I could check to see what process corner I was in by measuring the test structure.

It is a bit impractical for the designer to be putting in test structures that the mfg guys use.  Not saying it does not happen but other than alignment keys and fiducials, I have never seen them.

[Noopy]

It seems that I have to create a DAC-topic. 
Let´s start with the Burr Brown DAC709!

 
Such a beautiful chips, I would never ever decap them, I want them working!! 

PS: I wonder how much you are going to pay for the electricity when all the chips ever produced will pass through your oven 

Well basically you are right. But after all I don´t simply destroy the chips. I do some "research". Perhaps someone who wants to use these old chips needs some internal Information and can do a better job with my analyses. 

My ofen is very efficient, I can do a lot of decapping… 

[Wimberleytech]

It seems that I have to create a DAC-topic. 
Let´s start with the Burr Brown DAC709!

 
Such a beautiful chips, I would never ever decap them, I want them working!! 

PS: I wonder how much you are going to pay for the electricity when all the chips ever produced will pass through your oven 

Well basically you are right. But after all I don´t simply destroy the chips. I do some "research". Perhaps someone who wants to use these old chips needs some internal Information and can do a better job with my analyses. 

My ofen is very efficient, I can do a lot of decapping… 

That is a side-brazed ceramic package...just glue the lid back on!!

[james_s]

Fascinating pictures. If you're looking for more stuff to de-cap I have a faulty Seibu custom chip from a 90s arcade game that I'd love to see the inside of. It's a plastic QFP though and I'm not sure I want to play with fuming nitric acid nor do I have a suitable microscope to look at it.

[Noopy]

That is a side-brazed ceramic package...just glue the lid back on!!

I can do that. Right now the lid is taped on the package.

Fascinating pictures. If you're looking for more stuff to de-cap I have a faulty Seibu custom chip from a 90s arcade game that I'd love to see the inside of. It's a plastic QFP though and I'm not sure I want to play with fuming nitric acid nor do I have a suitable microscope to look at it.

I have success rate of ~90%. But some packages resist every treatment. If you are ok with that you can send the chip to me.
I´ll write you a PN.

[magic]

Fascinating pictures. If you're looking for more stuff to de-cap I have a faulty Seibu custom chip from a 90s arcade game that I'd love to see the inside of. It's a plastic QFP though and I'm not sure I want to play with fuming nitric acid nor do I have a suitable microscope to look at it.

Noopy bakes the chips to open them, acid is so 2018

I have success rate of ~90%. But some packages resist every treatment.

What kind of chips you had problems with?

[Noopy]

I have success rate of ~90%. But some packages resist every treatment.

What kind of chips you had problems with?

I infrequently have problems with epoxy packages. Particularly with the last few micrometer on top of the die.
First I can remove all the epoxy without problems but on the die itself remains a still very stable residue. This last rest of package doesn´t look different. I´m pretty sure that is nothing different than epoxy but it sticks to the die very well.
Since scratching too much damages the die the only way to deal with it is another backing process. This time I need some more temperature and duration: ~450-500°C. The die on the surface of the ceramic is probably hotter. After that procedure the residue is ash and can be removed. Unfortunatelly sometimes the higher temperatures damage also the die.
It seems to me that mostly newer packages have this "problem". Perhaps they have a surface treatment that sticks more to the epoxy of the package or something like that...  Perhaps it has something to do with a polyimid finish... 

And of course sometimes I have no luck and the die breaks apart while dissecting it. Meanwhile that´s very unusual but it can happen...

[Noopy]

NEWS!


I have decapped a DAC80:






Quite an old part but it´s very interesting because you can identify every single part.




There are three Intersil ICL8018 current switches which generate the analog current with a resistor array and a reference voltage. The current switches are different in size (1:2:4:8 ) to match the current densities.




Did some reverse engineering. More Pictures on my website.




Of course the resistor arrays are laser trimmed.




The generation of the reference current is very smart. The current only depends on the ground and the reference. The example circuit in the datasheet of the ICL8018 is based on the negative supply voltage.





There are two different Opamps: PMI OP-01 and OP-02
The OP-02 (the top one) is slower, the OP-01 has a feedforward path.
They only had to change the metal layer to switch from one to the other. 





Never seen such a buried zener. Interesting!


More pictures here:

https://www.richis-lab.de/DAC02.htm

Have fun!

 

[Miyuki]

Wow
Did they have bond wires even on capacitor 
It looks like a hand placed and wired job, just imagine that amount of manual labor put into IC manufacture, wonder how insane expensive it was back then

[Noopy]

Wow
Did they have bond wires even on capacitor 
It looks like a hand placed and wired job, just imagine that amount of manual labor put into IC manufacture, wonder how insane expensive it was back then

Yes they did! 
It particularly surprising because they were able to connect a transistor to a trace by soldering it. Nevertheless they bonded the capacitor... 

[Noopy]

I have a new DAC for you. It´s a DAC 32 built by KWH (Keramische Werke Hermsdorf).








A nice hybrid with glass resistor arrays and a THT zener.




It´s a R2R design with equal currents.




The die sinking the highest currents is placed on a second die for less leakage.




One of the three current sink dies. 06K720? 




Fusible Links for marking the quality of the current sinks.




Three Tesla MAA741 opamps.






A different kind of voltage reference. 


More pictures here:

https://richis-lab.de/DAC03.htm

 

[TheUnnamedNewbie]

Wow
Did they have bond wires even on capacitor 
It looks like a hand placed and wired job, just imagine that amount of manual labor put into IC manufacture, wonder how insane expensive it was back then

Few brands actually make capacitors with gold plating for wirebonding onto. I know Murata makes the WBSC/WTSC/WXSC series, which are used sometimes in RF stuff, and don't look like classic SMD caps. But they also have the GMD series which is a MLCC with AuSn terminals. You can solder these caps on a regular PCB (I think - never used these myself), and then wirebond to the top of them, so you can have your decoupling right up against the dies or in package. Pretty cool stuff.

[TomS_]

Wow
Did they have bond wires even on capacitor 
It looks like a hand placed and wired job, just imagine that amount of manual labor put into IC manufacture, wonder how insane expensive it was back then

I seem to have only seen BB used in military electronics and avionics so far. If those were their target markets, or their biggest consumers, then cost may have been only a secondary concern.

[TheUnnamedNewbie]

I seem to have only seen BB used in military electronics and avionics so far. If those were their target markets, or their biggest consumers, then cost may have been only a secondary concern.

BB also made a lot of consumer stuff. I believe they were big in the audio sector. Made some good audio-oriented low distortion opamps, ADCs and DACs. Bought by TI in 2000.

[Noopy]

I have seen something like a complex sound machine that worked with a DAC80.