Most of our electronics now days include integrated circuits, but what are these mysterious black boxes and how do they work?
We have often used ICs in Diyode Projects, and in fact, most of our projects use ICs rather than discrete components. So, what is meant by an IC? Integrated Circuits have been around for much longer than the ICs we currently think of.
In valve times, a Septode was a valve with seven electrodes, which actually was an integrated circuit with both a Triode (three electrodes) and a Pentode (five electrodes) in the one glass ‘Tube’ (aka Bottle or Valve). Now hopefully at least one reader has done the math and recognised that 3 + 5 = 8, not 7. (...recognising that the word ‘Septode’ suggests seven electrodes).
That’s where the word ‘Integrated’ comes in. The Anode of the Triode is also the Cathode of the Pentode! With one shared electrode serving between two components, we end up with seven electrodes.
One example, the 6BE6 valve, was a “Mixer” valve, including both an oscillator and an amplifier acting together to make the mixer. The bottom line is that the 6BE6 acts as “more than one component already connected together to perform a defined function”, which is a pretty good and simple definition for the term “Integrated Circuit”.
WHAT IS AN IC?
Modern ICs are more complex, in general, than the aforementioned Electronic Valve. However there are simple ICs that we would not even consider today as integrated circuits, such as the LM394 (and it’s brothers and sisters of similar numbers) being a “Supermatched Pair” of two transistors in a single package. The original package being a 6-pin T05 package which looks like a metal capped transistor.
The LM394 was used in Instrument circuits where it was essential for a differential amplifier to use very well matched transistors in the input. In fact, the LM394 might be considered a predecessor to the now ubiquitous ‘Operational Amplifier’.
The LM741 Operational Amplifier is an example of a group of ICs known as ‘Linear’ ICs which simply means they have predominantly ‘Analogue’ functions. Analogue or Linear databooks listed all manner of ICs including Op-Amps (e.g. LM741 etc., Linear Power Supplies (e.g. LM723 etc.), Timers (e.g. LM555 etc.) and yes, the ‘Transistor Array’ ICs such as the LM394, ULN2003, and many others.
Another group of analogue or linear devices sprung up including Power Amplifiers and Radio Function Blocks that are often gathered together into a Data Book labelled as ‘Special Function’.
While the possible ‘functions’ would indeed require a data book to explain, they include Phase Locked Loop ICs, Radio Mixers, Radio Power Output Modules (RF-Blocks) commonly called Bricks, and so many other Radio operations they in turn got their very own Data Book of “RF Devices”.
Perhaps I should point out that at one time I had over thirty Motorola Data Books, and those of many other manufacturers as well including the almost essential National Semiconductors three volume set of Linear Devices.
Data Books are a great way to keep up with what is available, but unfortunately not commonly available today. Of course the internet has taken over, but another issue is that many of the electronics manufacturers have been taken over by less well known brand names, and in some cases, less reputable manufacturers.
Most data books have been replaced with an online store of data sheets, and application notes, with at times restricted access, or only as paid downloads.
The request for this article came from our discussions on Digital Logic, and the fact that we mentioned so many ‘families’ of logic devices. As the logic families came out at different times, a Data Book would be produced for each family, e.g. TTL as a major digital logic family had sub-families of differing technology that while still TTL, was perhaps a faster type, lower power type, advanced type etc. and all had their own Data Book. We will go into the families of Logic Devices in another article however.
Although many early ICs did come in TO5 metal capped form, even 8-pin as the metal cap LM741 shown below attests, it became obvious that a new and simpler standard for integrated circuit packages was needed.
Many audio and RF ICs were made in a Single Inline Package (SIP) form. Although they took up less horizontal space, they had weight and vibration issues in sometimes violent applications such as industrial and military, where oscillations (vibrations) and shock loads could cause failure of the soldered joints to the pcb.
The SIPs did however start a standard that is still around today having lasted for decades, the 0.1” pin spacing, meaning that the legs of the ICs were spaced 0.1 inches apart, or an odd sounding 2.54mm spacing; for those of you that missed the era of inches and cubits!
Another option became more popular as being far more stable and also allowing for greater number of connections to the electronic ‘chip’; the actual Silicon semiconductor.
The DIP, (Dual Inline Package) was very flexible, yet still a good standard with two rows of pins 0.3” or 7.62mm between two rows of 0.1” spaced pins.
Note: when larger ICs such as ROMs, Microprocessors, and other ‘Large Scale Integration’ (LSI) ICs required more pins and a larger semiconductor chip, a Wide DIP; the WDIP of 0.6” width was introduced. The term is not usually necessary as the part number is almost always one or the other package.
PCB software may refer to WDIP, or perhaps define packages as DIP3 and DIP6.
Of course, our components have become more complex and grown many more legs so after the M68000 microcomputer was introduced with 64 legs in a WDIP package, other packages began to pop up and although they are standard packages, there are now many more ‘standards’ than necessary.
The DIP package can have as few as 6 legs usually representing an opto-coupler, an IC with an LED on one side and a phototransistor on the other side to provide electrical isolation between a high voltage and a low voltage circuit. Our new friend the LM394 can also be bought in a 6-pin package.
DIP devices with 8-pins include most commonly available Op-Amps, the LM555, and a lot of devices even including the ATtiny and some PIC micro-controllers.
Although not all combinations are used, DIP ICs up to at least 28 pins, on 0.3” spacing, are common enough with WDIP packages beginning at 24-pins and going up to the common 40-pin devices, and extending to the aforementioned M68000 (Motorola 68k) of 64-pins.
With two rows of pins, or even on SIP devices with one row, the problem arises of how to make sure you know which pin is which, so you can place the IC the correct way around, or as we say in it’s correct orientation.
The original SIP devices were a slip of ceramic with the silicon chip grafted to the middle, legs attached to the ceramic and then wired to the chip itself by tiny strands of silver or gold wires.
Then to protect the device the whole thing, sparing the attachment parts of legs of course, was dipped in a ceramic slip and fired to set the protective coating. Then the part number (etc.) was printed on the coating, including a small round dot to indicate pin-1; which is typically the pin on the left when you are reading the printed information.
When epoxy or plastic variations were made, the dot became a dimple or round dent in the moulded plastic case, just above pin-1. Moulded SIP devices may also have been identified by a chamfer on the top left corner of the case, or along the corner between the front face and the left end of the device.
While some DIP devices have a similar dimple or painted dot in the corner near pin-1, and sometimes ICs are found with a similar chamfer on the corner beside pin-1, the most common marking is a ‘Notch’ being a half round notch missing from the end of the IC, or more commonly today, a half round dimple at the end of the IC between pin-1 and the highest numbered pin.
Every IC identifies pin-1 in some way, but the other pins are never numbered, and I sometimes feel like that! (Joking!)
When I first started working with ICs I asked why (looking down on the IC with the text readable!) pin-1 was in the bottom left corner and the pins were numbered anti-clockwise around the IC. I was told by those more learned than I, “That’s just the way it is!”
It wasn’t until I started making PCB patterns, originally from stick on pads and black crape tape, that I recognised that those people that numbered the ICs were looking from below making pin-1 at figuratively the one o’clock position, (i.e. with two vertical columns of pins) and the pins were indeed numbered clockwise around the IC. This is even more obvious when looking at an 8-pin TO5 Metal Cap op-amp from below.
Surface Mount Devices are those intended to be soldered directly to the pcb without having any legs that require passing through the pcb. That’s great, and in itself solves several issues in making a pcb at home. No need to drill holes for example, but the technology also ignored hobbyists by ignoring 0.1” pin alignment, and instead started a competition into how small the gap between legs could become, and even for those with good eyesight, the next issue is the chip itself is almost too small to have a label.
Still, we sometimes need to identify a pin before we can predict what signal should appear on that pin. While some tiny ICs use the Notch/Dimple ID methods, some have a slight chamfer on the corner next to pin 1 and others have a painted dot. Get out the Jeweller’s Eye Glass!
There are too many square IC types with pins on each side to get into great detail, but it is worth pointing out that they are not always numbered from a corner as might be expected. Some micro-controller ICs are numbered from the centre of one side of pins, if odd numbered, or the pin to the left of centre if even numbered.
PIN GRID ARRAY
Other ICs have no legs but a pattern of gold coloured raised dots on the bottom surface of the IC. They are intended to be surface mounted to a PCB, and unless you are designing with them you would have no reason to know the pinout.
Some ICs, including many CPUs, use BGA (Ball Grid Array) packages, which are dropped into an IC socket. This method makes them easy to get in and out if you need to replace or upgrade them.
As we don’t yet have superconductor ICs, every IC has losses in operation, resulting in heat generation. Heat is energy. EH = I2RT, meaning that the heat generated and stored increases with time. Energy stored in a space results in an increase in some form of pressure or stress, whether pneumatic or hydraulic pressure, voltage, or in the case of Heat Energy, an increase in Temperature.
Thankfully heat flows from a position of high temperature to a position of low temperature, just as electrical energy flows from a high voltage to a low voltage. The thermal flow is indeed just like the electrical current, and perhaps you might have guessed that therefore electrical laws also work for Heat!
Very good if you did, Ohm’s law for example works for heat (now we’re using Fourier’s Law) as long as you remember which parameters are alike. Temperature(T) equates to Voltage(V), Heat(Q) equates to Charge(Q), Thermal Flow (ask your physics teacher) equates to Current(I) and Thermal Resistance (ask your physics teacher) equates to Resistance(R).
Essentially, the thermal flow depends on the Temperature difference between the hot spot and the ambient temperature of the air, or wherever the heat has to go, and the quantity of heat energy, and finally the thermal resistance of the path the heat must travel through, which like electrical resistance, depends upon the area, the length, and the material of the conductor.
IC HEAT DISSIPATION
In an IC, the heat generated is therefore limited by the opportunity the package has to dissipate the heat, and therefore to limit the temperature rise. It is the temperature of the silicon chip that is most important, and the temperature is greatest within the silicon, and of more concern, within the semiconductor junctions themselves.
That is where the semiconductor layers are most prone to “meltdown”. NOT that the silicon will totally melt, but the junctions may get hot enough that under current flow, the junctions themselves may dissolve together into either a conductor or insulator, rather than a semiconductor.
If just one junction is effected, the IC may not perform as desired and may simply result in lower performance, or higher noise levels. In some cases the IC may simply cease all operations even though externally it looks fine. Of course, in heavy abuse, a real meltdown can occur resulting in visible damage to the package itself which at least tells the repair guy where to look. See where the smoke came from!
To assist in cooling the device, the IC designer uses all of the factors at hand, from reducing heat generated to increasing the opportunities for heat to dissipate through the shortest path, over the greatest area of the path, and with the best conductive properties of the path.
IC HEAT DESIGN
The first option is to use a thermally conductive material for the case, one that is of course not electrically conductive. The material should be a minimum thickness for the voltage protection requirement, and present as large an area as possible to both the IC chip and the external environment.
Of course, there are other ways to convey the heat from the IC. You may have noticed IC pins designated as ‘NC’, meaning ‘Not Connected’. Mostly this is simply because those pins are not used for that particular version of an IC, pins 1,4 and 8 of some op-amps, for example.
However extra pins are often a way to transfer more heat out of the IC and ‘Audio Power’ ICs (without obvious heat sinks) may have four or more pins closest to the middle of the IC connected to ground, as a thermal sink, allowing heat to transfer to the copper ground plane which is also designed as a heat sink by using as much area of the copper layer as possible.
SPEAKING OF HEAT SINKS...
SIP packages are often used for power devices, allowing all of the electrical connections on one side of the IC and a large Aluminium “flag” on the other side that can be mounted to a much larger Aluminium heat sink at the top.
The whole heat sink circuit can be engineered and calculations made using formulae very familiar to what people working with electrical circuits already use, except you will not need a return path.
For DIP devices, instead of pins, some power ICs in DIP format have two large ‘legs’ that are used as heat sink connections replacing any number of narrow pins with the one wide ‘flag’ either side of the IC. They are sometimes referred to as ‘Butterfly’ packages.
Small glue-on heat sinks can also be attached to the top to increase heat dissipation.
Another danger that ICs face is Electro-Static Discharge (ESD), or Static Electricity. In fact static electricity is fine by me, as long as it doesn’t get out. When static electricity is let loose it may have a significant current flow (i.e. no longer static!) and a significant voltage to push the current.
First, I must explain how Static Electricity arises in a circuit or component. If two insulators are rubbed together, the electrons on the surface of one insulator are transferred to the other insulator, having no option but to settle in and relax!
It must be said that the two insulators must be different materials, and as far apart on the ‘Tribo-electric’ scale as possible. Tribo-electric is a scale that determines a materials propensity to attract or reject electrons.
One insulator has more electrons than it should have while the other has less. As the number of electrons increase, the insulator builds an electronic charge, increasing the negative voltage on that insulator while increasing the positive voltage on the other insulator, and twice that voltage between the two insulators.
An employee in an electronics shop can generate tens of thousands of volts simply by shuffling along in plastic shoes on a wool carpet, for example. The employee literally has lethal fingers!
The moment the person reaches to pick up an electronic component, the electrostatic charge may discharge to the component resulting sometimes in an audible ‘crack’ or a visible spark, but often no indication at all.
Electric charge (Q) is determined from current and time (Q=IT) or from Capacity and Voltage (Q=CV), giving us a double formula of Q=CV=IT. You might have noticed that to carry a charge a person must become a capacitor for a period, and insulated shoes are enough to achieve this. Dry air as found in many sealed electronics labs, and many parts of Australia, helps greatly.
Now for any person to charge up to 10,000 volts is not difficult, and may not be noticeable though not desirable around electronics (or flammable vapours!)
Engineers have specified a “Simulated Human” for testing purposes as a 100pF capacitor in series with a 1500Ω Resistor. So let us charge our simulated human to 10,000 volts (DC of course), resulting is a charge of Q = CV = 100e-12 x 10e3 = 1µC, 1 micro Coulomb.
Now you might agree that 10,000 VDC would be sufficient to cook an IC, but to be thorough, let’s imagine that 10,000 VDC would actually cause the IC to conduct, but without current flow, there is no damage!
Of course, once the IC conducts, current flows, limited of course by the 1500Ω resistor.
Now we should do exponential calculations to be precise and we could do that, but even as a ‘Guesstimate’ using DC values, 10,000 V limited by a 1500Ω resistor, using no more maths than simple Ohm’s law, results in a maximum current flow of 6.667 Amps.
This only happens for a very short time, but through such sensitive electronics, is enough to do damage to very small circuit elements even if the current only lasts for T = Q/I = CV/I = 100e-12 X 10,000 /6.67 = 150nS! (Remember over simplified but useful!).
PROTECTING ICS AGAINST ESD
Electro-Static Discharge (ESD) is therefore a big issue for small devices. Internal protection is possible but mostly to improve the defences rather than completely protect the device. The technology of the IC further complicates matters with CMOS being more sensitive than more conventional technology, but therefore tends to be designed to be better protected by internal diodes and such.
When testing devices in circuit, a good habit is to connect the ground of the PCB and of the testing device together, and connect both to a common bench ground if available.
Operators may wear a wrist strap attached to that ground, or regularly touch that ground to discharge themselves, especially before touching circuitry.
Tools such as the soldering iron may be double insulated with no ground, which in effect means a floating ground. Some workshops have used a technique of sliding an appropriate size of light spring over the soldering iron barrel with an attached flexible wire returning to the bench ground. Of course a ESD safe tool would be better.
Touching the soldering iron tip to a grounded point, bench ground, makes sure that an un-earthed soldering iron is at least temporarily discharged.
Note: In some workshops, it is common (but not recommended) practice to solder live circuits, mostly low voltage and low current circuits, and therefore an earthed soldering iron would actually present a hazard!
Other unexpected shorts can result from direct earthing/grounding, so one way to prevent unexpected shorts is to place a 1MΩ resistor in series with grounding leads. The resistor allows discharge while limiting any unintended shorts. Any good wrist strap and lead will also feature a series resistance just in case the lead comes into contact with dangerous voltages.
We have looked at ICs from a general perspective, from a distant perspective in fact without going into great detail on any one IC. However we looked at different groups of ICs, and where they are used or how they may be different, and how they are standardised.
We looked at issues like Thermal and ESD protection, and how manufacturers attempt to protect their devices. As you can see however, the number of IC devices with many varying requirements and technologies makes standardisation very difficult.