In previous articles with photographs of crystals of microcircuits ( 1 , 2 , 3 ) - they wrote in comments that it would be nice to disassemble a simple microcircuit in details - so that it was clear “what is what” at the lowest level, and where there is “magic smoke "Hiding. I could not choose a microcircuit for a long time, in the scheme of which one could figure it out in a few minutes - but finally the solution was found: ULN2003 - an array of Darlington transistors .

Despite its simplicity, the microcircuit is still widely used and manufactured. ULN2003 consists of 21 resistors, 14 transistors and 7 diodes. It is used to control the relatively powerful load (up to 50 volts / 0.5 amps) from the microcontroller foot (or other digital circuits). Canonical application - to control the powerful 7-segment LED indicators.

Crystal itself

As follows. The colors are somewhat reinforced with respect to natural ones; under the pads, the metal is damaged by acid (and has acquired such a brown color):
')


As we see, 7 channels are absolutely identical, therefore we will consider only one. Fortunately for us, the scheme of each channel is known to us - and we can pry into it:



And now 1 channel with marked elements. Comparison of specific elements of the scheme - I leave it as homework for the reader.



But how is the transistor itself made? It is known that the internal structure of a planar bipolar npn transistor during production is as follows:



Thin base - "dives" under the emitter. Despite the fact that both on the collector and on the emitter - silicon is doped to type n, the dopant concentration and thickness differ: this is done in order to optimize the transistor for “current amplification” in one direction.

Knowing this - we can take a closer look at 1 transistor, and understand where that is. Silicon alloyed in a different type - a little different in color. To the naked eye it is almost negligible - but here the saturation of colors and contrast are twisted to almost the maximum. Let 2 emitters connected in parallel do not confuse you - they work like 1 large area.

To ensure that the connections do not “short-circuit” what is not needed - the silicon surface is covered with a layer of transparent glass (SiO ₂ ), in which there are holes directly above the places where the output is connected to the right place on the transistor. This is clearly seen in the next photo, because the depth of field on this lens is smaller, and for example the connection to the base is no longer in focus, because located above, above the glass layer.

The collectors of both transistors are actually a single whole, since according to the scheme they are connected. Neighboring channels are isolated by a pn junction, you can see a rectangle of slightly different color around each channel in the general high-resolution photo .



As you can see, there is no magic inside :-)