Transistors - die pictures

[Noopy]

The famous BF862.






It´s an genuine one! 
The design is very interesting. Two areas contain a lot of small transistors.




Well, Zeptobar´s pictures are better...
There is a very thin gate grid surrounding the drain and source areas contacted by the metal layer.
There are three round areas without contacts and gate grid. Would be interesting what these round areas do. Do they lower the current density? Do they optimize the manufacturing? Some weird noise reduction? 


https://www.richis-lab.de/FET17.htm

 

[Miyuki]

Well, Zeptobar´s pictures are better...
There is a very thin gate grid surrounding the drain and source areas contacted by the metal layer.
There are three round areas without contacts and gate grid. Would be interesting what these round areas do. Do they lower the current density? Do they optimize the manufacturing? Some weird noise reduction? 

 

Do I understand it right. Those tiny electrodes connected to bond pads are D/S and the die itself is Gate?

[Noopy]

Well, Zeptobar´s pictures are better...
There is a very thin gate grid surrounding the drain and source areas contacted by the metal layer.
There are three round areas without contacts and gate grid. Would be interesting what these round areas do. Do they lower the current density? Do they optimize the manufacturing? Some weird noise reduction? 

 

Do I understand it right. Those tiny electrodes connected to bond pads are D/S and the die itself is Gate?

The tiny contacts contact a lower layer where there are small drain and source rectangles.
The channel is controlled by the substrate (below) and a very thin grid that is integrated around the drain and source rectangles (above). The substrate and the grid are the gate.

[Noopy]

2AW - internet says that´s the BF862 manufactured in China while the 2Ap is manufactured in Hong-Kong. I found no official statement to the 2AW. In the datasheet you just find the 2Ap.
The 2AW was one of the cheapest BF862 I found on Ebay (0,5€ while the 2Ap was 2,74€).
The surface of the package looks a little strange like it was sanded.






Surprise, it´s a counterfeit BF862! 
I´m sure that transistor does behave very different than a BF862.
There is an unused gate bondpad. Gate potential comes through the substrate.


https://www.richis-lab.de/FET18.htm

 

[magic]

The 2AW was one of the cheapest BF862 I found on Ebay (0,5€ while the 2Ap was 2,74€).
The surface of the package looks a little strange like it was sanded.

I'm shocked

Do I understand it right. Those tiny electrodes connected to bond pads are D/S and the die itself is Gate?

Right. Basic structure shown below (Siliconix, "Designing with Field Effect Transistors"). In BF862 the gate is a grid rather than a single line or multiple lines. It seems that the substrate is typically (always?) connected to the top gate and also participates in pinching the channel.



I wonder if those "holes" in BF862 could be connections between the top grid and the substrate.

[Noopy]

I wonder if those "holes" in BF862 could be connections between the top grid and the substrate.

That´s possible... 


Someone told me what transistor we probably have found here:



(National Semiconductor FET Databook 1977)

 

[Noopy]

Now let´s take a look into a BF861. It has less forward transfer admittance than the BF862 and was sold in three different bins.
M34 is the old Philips naming.






The edge length is 0,32mm. The structures are similar to the structures of the BF862 but they are bigger. Cgd is bigger than Cgd of the BF862 (2,1-2,7pF vs. 1,9pF).


https://www.richis-lab.de/FET19.htm

 

[Noopy]

Today something special.
The К1HT291Б (K1NT291B) is a dual transistor built by ALFA (Latvia). Because the transistors are integrated on one die they are ideal for differential amplifiers => similar temperatures, similar specifications
The special package was used to develop hybrid circuits (the DAC 32 is an example for a hybrid circuit: https://www.richis-lab.de/DAC03.htm). Over the valve like pins in the bottom glass plate you are able to test the transistors. Then you cut the wires near the pins and solder them into your new hybrid circuit. 






The bondwires hold the die in place. There is some potting protecting the die and the bonds.






The die was cut out of the wafer quite crude.
In the sawing lines there are test structures to monitor the mask alignment and the manufacturing process.




The two transistors are isolated from each other with frames.
The big bondpad connects the collector. It seems like there is no buried collector connection.
You can see the base area and on the right side you can even spot the small emitter area.


https://www.richis-lab.de/Bipolar71.htm

 

[RoGeorge]

What on Earth is that?!   

[Noopy]

A special package for a special purpose. 

It absolutely makes sense.
In first place you can test the transistors. Perhaps you even test your whole circuit. Date code is 1972. I´m sure there were a lot of tube sockets around. 
Then you cut the wires and just have to solder them in place without the need of bonding wires. 

Hybrid circuits, a world on its own. In Russia they have built a lot with hybrid circuits.

[RoGeorge]

It absolutely makes sense.

[Noopy]

 

YMMD

[mawyatt]

Nice images!!

Wonder if this is something like Fluke did way back then when they developed a RMS detector utilizing a pair of transistors that were thermally isolated from the case. By isolating the transistor pair, this had low thermal mass. The idea was based upon having a integrated resistor pair (50 ohms I believe) near each transistor, one resistor was driven with the measurement waveform, the other resistor driven from an integrating feedback loop to "balance" the two transistors thru heating the resistor to balance the Vbes. This feedback was the equivalent of the heating effect of the waveform and thus the RMS content.

Best,

[Noopy]

Thanks! 

Would be a good solution for a RMS detector. 

The more modern version is the LT1088:
https://richis-lab.de/LT1088.htm

[magic]

The slightly less obsolete version

[T3sl4co1l]

It's... just fucking floating in space?!  What was it rated for, 10mW?

Tim

[Noopy]

15mW 

[mawyatt]

Thanks! 

Would be a good solution for a RMS detector. 

The more modern version is the LT1088:
https://richis-lab.de/LT1088.htm

Think LT copied the idea much later, I recall the Fluke note from ~1970. Fluke even thinned the die to reduce the thermal mass and suspended it with wire bonds just like in your images (why I immediately thought of the Fluke concept). At that time the only effective lab means for a somewhat True RMS measurement was the Calorimetric Wattmeter from HP, the analog techniques employing the transistor exponential characteristics worked OK, but didn't handle high crest factors well. Of course the usual absolute value averaged RMS calibrated method back then was only accurate for pure sinusoidal waveforms, and you had to apply a correction factor for other common waveforms.

Edit: Thinned die was often used for various purposes, recall when we were working with Draper Labs a demo of an 200mm thinned wafer you could see thru!!! Then this wafer was wrapped around a Coke can

Best,

[Noopy]

Tesla KT110, a fast power thyristor. This one holds 400V ("/400"). Other bins isolate 200V to 750V. Continuous current is 3,2A, peak current is 30A, once 50A. The switch off time is less than 40µs.




We know this red potting Tesla used.
The cathode wire is thicker than the gate wire. 




The wires were soldered to the connector pins... ...but that tin ball doesn´t look good. 






The edge length is 2,8mm. The cathode area was flooded with tin for lower resistance.
You can spot a step in color between the gate and the cathode area. The construction is probably similar to the well known thyristor structure that is shown with the ST103: https://www.richis-lab.de/Bipolar13.htm




They etched the edges of the die (a funny sentence  ) to get a clean gate cathode junction edge.


https://www.richis-lab.de/Bipolar72.htm

 

[Miyuki]

The wires were soldered to the connector pins... ...but that tin ball doesn´t look good. 

Oh, just imagine that poor workers soldering wires to this 

[Noopy]

2N3700, a transistor that was produces by a lot of companies. I don´t know which company made this one. There is just 2N3700 on the package.
You can get the 2N3700 still today, you can get it with MIL spec and there is even a radiation hardened 2N3700HR. It seems the 2N3700 was integrated in ballistic missiles. 
Max ratings: 80V, 1A, 1W






With an edge length of 0,68mm the die is quite big for a small signal transistor. The contact structure of the metal layer looks like a power transistor. With that setup ft is just 100MHz.


https://www.richis-lab.de/Bipolar73.htm

 

[Noopy]

KT808BM, the worse KT808 (https://www.richis-lab.de/Bipolar48.htm). Vce is just 100V (130V for the KT808A).






The old KT808 design...








Not the cleanest structures but the other KT808 (old design) look similar.




Base emitter breakdown occurs at -7,5V (50mA).




(1A) There is an artefact...










There is a small dot with round about 15µm diameter besides the base emitter border. The light emission shows us that the junction goes around this impurity. Due to the higher field strength the breakdown occurs early at this point.





A second KT808BM with the same date code but the thickness of the base plate is nearly twice that of the first KT808BM: 2,5mm vs. 1,4mm.








Nothing special inside... ...but the die attach looks a bit strange. There is quite a lot of solder around the die.




The die looks a little stressed where the connection wires were soldered.




In breakdown (-8V, 1A) the current distribution is very uneven. Most of the light occurs above the base electrodes.


https://www.richis-lab.de/Bipolar74.htm


 

[Noopy]

SF137, a small HF transistor built by the Halbleiterwerk Frankfurt Oder.
20V 200mA 300MHz




Now that is a big current amplification range! 




LM, built in September 1972.








The die is 0,49mm x 0,49mm.
The design is similar to the SS109 (https://www.richis-lab.de/Bipolar01.htm).


https://www.richis-lab.de/Bipolar75.htm

 

[RoGeorge]

It reminds of the rod and crankshaft from the Steam logo:


Photo from:  https://1000logos.net/steam-logo/

[Noopy]

It reminds of the rod and crankshaft from the Steam logo:

Indeed!
Opening the next TO-3 I will be afraid of small headcrabs! 
https://half-life.fandom.com/wiki/Headcrab