Author Topic: Different die pictures  (Read 81141 times)

0 Members and 2 Guests are viewing this topic.

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1701
  • Country: de
    • Richis-Lab
Re: Different die pictures
« Reply #150 on: September 22, 2021, 07:57:25 pm »


The Mullard 438BGY is an old RF amplifier (>18W between 68 and 88MHz).
First ad for the 438BGY was found 1974.




You don´t find much information about the 438BGY but the schematic of the BGY32, BGY33, BGY35 and BGY36 fits quite well. I added the part numbers, we will need them later.
There is a preamp around the transistor T2 and a output stage around T3. At the inputs and outputs there are some LCs for filtering and impedance matching.
T1 generates a bias so T2 doesn´t switch off completely. It´s interesting that T1 is supplied from the output stage (every stage has its own supply rail). I assume that supply concept should prevent oscillations because the phase shift is "better".




Removing the lid you find a PCB potted with some silicone gel. There are THT, SMT and bare die components.
(Module turned around 180° so the signal flow is from left to right.)




It´s a normal PCB with copper on both sides. You can see the glass fibre.




Vias are connecting the ground potential to the backside of the board. The heat sink acts as low impedance ground.




The preamp is on the left side. The only difference to the schematic is the diode Dx. The inductors are built with loops of copper on the PCB. It looks like the inductance L3 is distributed around the resistor R2.




The smaller capacitors are built with silicon dies. There is a big and two small rectangles. You can adjust the capacitance by bonding the smaller areas to the main area.




T1 is a quite normal transistor since it just does the biasing.




At the base of the preamp transistor T2 there is the diode Dx. It is integrated on a die with three other diodes. The diode protects the transistor against negative base emitter voltages.






The preamp transistor T2 is placed on a BeO ceramic. The BeO dissipates heat much faster than the PCB.






The preamp transistor T2 is a RF transistor with very thin emitter areas. Between the emitters there are the base contacts. A low base impedance allows fast switching.




I don´t think they did some tuning here. In my view it looks more like some bond problems.




On the right side there is the output stage.




That looks like some tuning. The first bond connection was cut and a new bond connection was established at a different place.




L6 has a core because you need a lot of inductance at this place.




The output transistor T3 sits on a second BeO ceramic. The currents are below 1A but you need a lot of connections to get a low parasitic inductance.






T3 is an overlay transistor like we have seen it in the 2N3553 (https://www.richis-lab.de/Bipolar22.htm).
Every vertical emitter line has an emitter resistor and the whole block is connected to one single horizontal emitter resistor. That´s important to get a uniform current distribution.


https://www.richis-lab.de/rfamp01.htm

 :-/O
« Last Edit: September 23, 2021, 03:48:11 am by Noopy »
 
The following users thanked this post: SeanB, TimNJ, doktor pyta, RoGeorge, Miyuki, capt bullshot, ch_scr, iMo, Renate

Offline Renate

  • Super Contributor
  • ***
  • Posts: 1460
  • Country: us
Re: Different die pictures
« Reply #151 on: September 22, 2021, 10:18:23 pm »
Those are very nice photos.
I hope that you took EXTREME precautions with that berylium oxide???
 

Offline Cerebus

  • Super Contributor
  • ***
  • Posts: 10576
  • Country: gb
Re: Different die pictures
« Reply #152 on: September 22, 2021, 11:06:27 pm »
Those are very nice photos.
I hope that you took EXTREME precautions with that berylium oxide???

As long as you avoid breaking it, or chipping bits off it's fine. The biggest hazard is inhalation of dust. However, unlike some of the various relatively minor chemical hazards you can encounter with electronics that sometimes get rather exaggerated, BeO is something that should be treated with proper respect. But it doesn't need a huge song and dance. Careful handling of solids and a decent FFP3 dust mask is adequate. However, BeO powder is something I'd not want to personally come into contact with and would only handle, if at all, in a fume cupboard/glove box.
Anybody got a syringe I can use to squeeze the magic smoke back into this?
 

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1701
  • Country: de
    • Richis-Lab
Re: Different die pictures
« Reply #153 on: September 23, 2021, 03:47:29 am »
Thanks!  :)

I agree with Cerebus. As long as you don´t grind the BeO it´s not that bad.

In the 438BGY the BeO is even potted. Parts like the 2N3375 are more problematic:



You have to be careful with these components.  :-/O
 
The following users thanked this post: doktor pyta

Online RoGeorge

  • Super Contributor
  • ***
  • Posts: 6136
  • Country: ro
Re: Different die pictures
« Reply #154 on: September 23, 2021, 10:29:20 am »
I agree with Cerebus. As long as you don´t grind the BeO it´s not that bad.
...
You have to be careful with these components.  :-/O

True that.  When not grind, only mild side effect are observed, for example the exposed subjects develop an urge to eviscerate electronic parts, and an attraction to big lens and microphotography.  These side effects have no known explanation yet.   :-//

However, the community knows about the effects, so the community keep sending to the affected ones small sacrificial electronic parts.  This way the affected can eviscerate the little gifts, and thus tame their urge to turn during the nights with full moon into what it is known in mythology as an ausgeweidetelektronisch (electronicoevisceratorus in latin).  ;D
 
The following users thanked this post: ch_scr

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1701
  • Country: de
    • Richis-Lab
Re: Different die pictures
« Reply #155 on: September 23, 2021, 01:02:29 pm »
Bravo!  :-+ ;D

Offline mawyatt

  • Super Contributor
  • ***
  • Posts: 3180
  • Country: us
Re: Different die pictures
« Reply #156 on: September 23, 2021, 01:34:28 pm »
During my career I worked at Honeywell which produced navigational systems based upon an Electrostatically Suspended Gyro, or ESG. The ESG used a thin hollow Beryllium ball about the size of a golf ball electrostatically suspended inside a BeO cavity. The entire area where these were produced was inside a somewhat sealed environment (negative pressure I recall) and everyone wore gowns and masks like you see in the semiconductor areas. Never had a desire to go into the production area, but did work around some of the gyros.

https://www.ion.org/museum/item_view.cfm?cid=2&scid=4&iid=30

Also recall some of the early days where thermal paste or grease was BeO in a silicone gel.

Best,
Curiosity killed the cat, also depleted my wallet!
~Wyatt Labs by Mike~
 
The following users thanked this post: doktor pyta, RoGeorge, MegaVolt

Offline mawyatt

  • Super Contributor
  • ***
  • Posts: 3180
  • Country: us
Re: Different die pictures
« Reply #157 on: September 23, 2021, 02:11:37 pm »
I agree with Cerebus. As long as you don´t grind the BeO it´s not that bad.
...
You have to be careful with these components.  :-/O

True that.  When not grind, only mild side effect are observed, for example the exposed subjects develop an urge to eviscerate electronic parts, and an attraction to big lens and microphotography.  These side effects have no known explanation yet.   :-//

However, the community knows about the effects, so the community keep sending to the affected ones small sacrificial electronic parts.  This way the affected can eviscerate the little gifts, and thus tame their urge to turn during the nights with full moon into what it is known in mythology as an ausgeweidetelektronisch (electronicoevisceratorus in latin).  ;D

Funny  :-DD

Must have got contaminated over 20 years ago, ever since been imaging chips and designing/collecting custom imaging fixtures and lens assemblies ;D

These were/are used to image the chips we designed in very high resolutions into the gigapixel region.

Honestly don't know how Noopy does this with so many excellent images, I know how difficult this is and how much time it takes to create a quality chip image, and they just keep coming.....so hat's off to Noopy :-+

Best,
Curiosity killed the cat, also depleted my wallet!
~Wyatt Labs by Mike~
 
The following users thanked this post: RoGeorge

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1701
  • Country: de
    • Richis-Lab
Re: Different die pictures
« Reply #158 on: September 23, 2021, 05:43:00 pm »
Honestly don't know how Noopy does this with so many excellent images, I know how difficult this is and how much time it takes to create a quality chip image, and they just keep coming.....so hat's off to Noopy :-+

Thanks!  8)
I spend quite some time to extract the dies, take the pictures, process them and put them in a context that is interesting to read.  :-/O
...and beside that I lead a very normal life...  ;D
« Last Edit: September 24, 2021, 05:54:31 am by Noopy »
 
The following users thanked this post: MegaVolt

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1701
  • Country: de
    • Richis-Lab
Re: Different die pictures
« Reply #159 on: September 28, 2021, 07:08:07 pm »






Texas Instruments BQ24040, a Li-Charger for one cell and a supply of 5V. Up to 1A of charge current is possible with an option to switch to 500mA or 100mA (USB current limit). Voltage accuracy is 1%, current accuracy is 10%. The BQ24040 doesn´t need an external shunt.
WSON package => "Very, very thin Small Outline Non-leaded"  ;D Edge length is 2mm.




Datasheet shows a block diagram too.  :-+








Now that is interesting: There is a very thick metal layer above the highly integrated circuit. This top metal layer is used to form the bondpads so you have more area for the circuit. In addition the high cross section is important to conduct the relatively high current.
The structure in the right area looks like the metal layer contacts the two power transistors. And it looks like the upper transistor is smaller than the lower one. The upper one is probably the one that stops current flowing back to the input. While the lower one does the current regulation and has to stand more power loss.
There are four unused bondpads. The die is probably used in the BQ24041 and BQ24045 too. These chargers are very similar.




Aha! It looks like the BQ24050 and BQ24052 use the same die. These chargers are watching D+/D- to draw a high current if the connected USB device can supply it.
Revision A3?






The die is 0,2mm high, the top metal layer is ~20µm.


https://www.richis-lab.de/li02.htm

 :-/O
 
The following users thanked this post: RoGeorge

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1701
  • Country: de
    • Richis-Lab
Re: Different die pictures
« Reply #160 on: October 10, 2021, 07:57:00 am »
BQ27220, a single li cell fuel gauge in a small flip-chip package (DSBGA-9 - 1,62mm × 1,58mm).
[...]
https://www.richis-lab.de/li01.htm

Texas Instruments BQ24040, a Li-Charger for one cell and a supply of 5V. Up to 1A of charge current is possible with an option to switch to 500mA or 100mA (USB current limit). Voltage accuracy is 1%, current accuracy is 10%. The BQ24040 doesn´t need an external shunt.
WSON package => "Very, very thin Small Outline Non-leaded"  ;D Edge length is 2mm.
[...]
https://www.richis-lab.de/li02.htm


By the way: These two chips were originally built by the company Benchmarq Microelectronics which was acquired by Unitrode which was acquired by Texas Instruments.
Because of Benchmarq Microelectronics there are the letters BQ which originally were uncapitalised.
 
The following users thanked this post: MegaVolt

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1701
  • Country: de
    • Richis-Lab
Re: Different die pictures
« Reply #161 on: October 16, 2021, 05:55:51 am »
I have put together a new picture calendar with the best pictures of 2021:

https://www.meinbildkalender.de/richis-lab

(Sorry, there is just a german speaking calendar.  ;))
 
The following users thanked this post: capt bullshot

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1701
  • Country: de
    • Richis-Lab
Re: Different die pictures
« Reply #162 on: October 18, 2021, 06:03:17 am »
I had created a list with all "my" semiconductors sorted by name of the part:

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


Now here we have a list sorted by manufacturer:

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

 :-/O
 
The following users thanked this post: capt bullshot, MegaVolt

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1701
  • Country: de
    • Richis-Lab
Re: Different die pictures
« Reply #163 on: October 20, 2021, 04:31:35 am »


Here we have the next part of the GDR telephone system (https://www.richis-lab.de/phone.htm), the U3230, the so called "PCM-Koppelfeld". No datasheet, no more information just like with the U3220.  :(






But one thing is interesting. I have taken pictures of a U3230 wafer (the upper one). There are more testpads and there are small circuits integrated near most of the testpads. I assume that is some kind of buffer. On the die of the U3230 they removed these circuits.


https://www.richis-lab.de/phone11.htm

 :-/O

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1701
  • Country: de
    • Richis-Lab
Re: Different die pictures
« Reply #164 on: November 24, 2021, 09:19:17 am »

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1701
  • Country: de
    • Richis-Lab
Re: Different die pictures
« Reply #165 on: November 26, 2021, 04:40:21 am »


Here we have the next part of the GDR telephone system (https://www.richis-lab.de/phone.htm), the U1500PC001, the so called "PCM-Sender". Again no datasheet, no more information...  :(
It seems like the size of the die was caused mainly due to the number of bondpads needed.






The U1500PC001 is based on the standard cell family U1500 designed by Carl Zeiss Jena.
They first used shorter names for U1500 circuits so here we see the U1501.




There was enough free area for a big test structure that even contains input protection structures!




Here we have two interesting circuits. In the upper left corner of the picture there is an additional input protection in a second line. The two bondpads above this circuit are outputs but it seems they wanted to use them as inputs too.
The U1500 system was designed to build big logic circuits but here there is quite a big push-pull output driver. There are even additional bondpads for the outputs.


https://www.richis-lab.de/phone12.htm

 :-/O
 
The following users thanked this post: RoGeorge

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1701
  • Country: de
    • Richis-Lab
Re: Different die pictures
« Reply #166 on: December 01, 2021, 08:01:50 am »


We had the first graphic processor µPD7220 (https://www.richis-lab.de/GraKa02.htm) now let´s take a look into the Intel version D82720.
M82720 was the military grade.






There is one more bondwire than in the µPD7220. It connects the die with the bottom of the package through a block.




The edge length of the die is 5,6mm. It´s a little bigger than the µPD7220 but that is due to the bigger edge area. All in all the D82720 and the µPD7220 use the same design. Well no surprise since NEC and Intel worked together to develop this graphic processor.




Yes, it´s an Intel 82720...




...but it´s also a NEC 7220.  ;D
In the NEC µPD7220 there is no Intel logo.










There are no test structures in the middle of the die but Intel has integrated a lot of test structures in the cutting areas.






The bias generator looks a little different in comparison to the µPD7220. Since it it connected to the bottom of the package we can try to measure the bias voltage. => -2,6V  :-+




With just a supply connected to the D82720 I was able to meassure a very small 33MHz signal between the supply pins. Shorting the leads exactly at this point didn´t show the signal so it´s quite likely we see here the switching frequency of the bias generator.




Numbers for the bondpads as in the µPD7220. Nice!  :-+




Output pads with predriver and push-pull transistors.




Input circuit - There is a series resistor limiting the input current and a grounded gate NMOS to drain pulses.
Under the input bondpad in the outer frame there is a contact to a rectangular area. Substrate is at a negative voltage so it is no substrate contact. I assume that´s a diode so if the local substrate get´s more positive than ground current can flow into the ground structure.




Memory is a little clearer than in the µPD7220.


https://www.richis-lab.de/GraKa03.htm

 :-/O
 
The following users thanked this post: RoGeorge

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1701
  • Country: de
    • Richis-Lab
Re: Different die pictures
« Reply #167 on: December 03, 2021, 07:08:05 am »


Here we have the next part of the GDR telephone system (https://www.richis-lab.de/phone.htm), the U1500PC002, the so called "PCM-Empfänger". Again no datasheet, no more information...  :(
Just a lot of logic in the last parts of the telephone system but I want to complete the collection.
One interesting detail is the great number of capacitors. These capacitors make the supply more stable.




In the first place the name was U1502. It seems like this design is a second revision.




Some test structures to check the alignment.


https://www.richis-lab.de/phone13.htm

 :-/O
 
The following users thanked this post: RoGeorge

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1701
  • Country: de
    • Richis-Lab
Re: Different die pictures
« Reply #168 on: December 14, 2021, 08:56:06 am »




Here we have the last two ICs of the GDR telephone system (https://www.richis-lab.de/phone.htm), the U1500DC007 and U1500FC008. But hey, that are the same circuits!? Yes, they are!  ;D




Both circuits share the name U1503. The U1503 is an IC to transmit data by optical fiber.
There is no information about a U1500DC007 and a U1500FC008.  :-//




At the right edge of the die there are two "big" push-pull-structures.




That´s an interesting structure. Besides the supply there is just one connection.
Two big capacitors, two big transistors, perhaps an internal oscillator?


https://www.richis-lab.de/phone14.htm

 :-/O
 
The following users thanked this post: RoGeorge

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1701
  • Country: de
    • Richis-Lab
Re: Different die pictures
« Reply #169 on: December 17, 2021, 03:44:52 pm »

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1701
  • Country: de
    • Richis-Lab
Re: Different die pictures
« Reply #170 on: December 20, 2021, 04:49:54 am »
I have some more analog switches. I will post them here https://www.eevblog.com/forum/testgear/replacement-for-fluke-700013-ic-(quad-spst-analog-switch)/msg3884996/#msg3884996 because I started with putting them into this topic. :)
So if you are interested in more analog switches click notify in this topic.

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1701
  • Country: de
    • Richis-Lab
Re: Different die pictures
« Reply #171 on: December 21, 2021, 02:11:03 pm »


Just a small Update: I added some package pictures to the GDR telephone system circuits.

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

Online T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21567
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Different die pictures
« Reply #172 on: December 24, 2021, 11:04:01 pm »
No picture, as I don't have anywhere near good enough magnification, or even much of a specimen, but --

Cracking open a failed FQA9N90C today, I notice an iridescent blue glow.  Even in the middle of the chip (it broke more or less down the middle, some stuck to the tab, some stuck to the plastic).  I wonder if I'm seeing the trenches here, or even actual N/P pillars (the "super" in "SuperJunction" tech)?  Also, doesn't seem to be polarized (appearance doesn't change under polarizer in front of light or view).

Die is 5 x 6 mm BTW.  So the power rating seems reasonable.

Also, failure occurred in linear operation, though I suspect the problem was overheating from poor thermal grease.  It's a bit inconclusive.

So, take this as a point of interest, if you run across such a MOSFET, maybe melt / crack it open and see if there's any funny color in the die? :)

(Also, not that just a bare die shot will be all that interesting -- as examples in this thread I think likely already show?  Acid to remove surface metal/oxide should prove interesting though.)

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline TimNJ

  • Super Contributor
  • ***
  • Posts: 1649
  • Country: us
Re: Different die pictures
« Reply #173 on: December 25, 2021, 05:27:19 am »
On the note of "SuperJunction" and "SuperTrench", are there any die shots of these types of modern MOSFETs?

Here's a 600V super-junction: IPA60R280P7XKSA1

Here's a 100V shielded gate trench / supertrench (whatever): AOT66920L

There are countless others, but if you were uhh...looking for ideas? (Seems you have plenty of ideas already  ;))
 

Offline NoopyTopic starter

  • Super Contributor
  • ***
  • Posts: 1701
  • Country: de
    • Richis-Lab
Re: Different die pictures
« Reply #174 on: December 25, 2021, 08:08:48 am »
I take the FQA9N90C on my list.  :-+
Could be interesting with some HF and some HCL.

Is the STP3NB100FP (https://www.richis-lab.de/FET15.htm) already a superjunction MOSFET? Not sure...

I definitely have to examine more Power-MOSFETs since I have started to do some delayering.
I already have a lot of parts and ideas but new suggestions are always welcome.  :-+
 
The following users thanked this post: TimNJ


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf