Beginners guide - Electronics
Created; 05/02/2015, Changed; 18/01/2024, 14/12/2023
Previous page; Electronics main page
Getting the Knack for Electronics
As a youngster, I struggled with circuit understanding (perception of symbols) and the application (construction) of the circuit. Learning to be patient, taking a long time and effort eventually leads to success was another lesson. Do play with circuit design, copy small examples of circuit function blocks, and play with them by changing things, but don't leap into a bigger constructional project at first. A small constructional project such as a kit is a good way of learning assembly and reading instructions I had an American crystal set Radio kit with a 50-foot (15M) iron wire aerial when I was a boy, it did not use a cat's whisker crystal but used a conventional glass point contact diode. The magnetic headphone was sensitive, high impedance, but after trying to get something from the radio, for a long period, this was the most worthwhile part left of it. There was virtually no tuning, and you heard all AM stations scrambled together but very quiet.
The key to understanding and predicting outcome is to appreciate that circuit, PCB layout, electromagnetic behaviour and performance are interlinked. This understanding will become intuitive (a learned semi-automatic process) in time. You need to understand the whole, but to get to understanding the whole system you need to work on the pieces and so get an understanding of each function. Therefore, periodically step back to look at the project as a whole and then go back into working on another or the same detail.
The ICL7117 had an auto-zero offset calibrations stage with a high quality capacitor, dual slope analogue to digital conversion stage and another high quality capacitor and a voltage reference and a third good quality capacitor, therefore it is not simple and elegant compared to alternatives but uses more components.
Digital panel meter
This Digital Panel meter is based on ICL7117 Intersil Integrated Circuit, works very well and was regarded as the best integrated circuit for that function of its time.
The double-sided PCB was drawn using crêpe tape and scalpel on stable plastic drafting film.
This digital voltmeter is a non-multiplexed design with a smooth linear performance compared to a bought in multiplexed unit which had errors around where more or fewer segments of the display were used. I design it when I was 20 and revised the design 8 years later when capacitors were packages became more standardised at 2.5mm, 5mm, 10mm etc. I have always spent enough time on a design so that it is right as this design is.
So this IC is not like the Agfa camera on the previous page (it appears) in that each function block is separated. For example other manufacturers made a DVM IC's that used a techniques called triple or quad slope which first measured the input offset then subtracted that from the measured input voltage, this part eliminated the auto-zero correction function block by reusing the analogue to converter block. That is, the digital processing was more complicated than the Intersil part.
Carrying out worse case design on parts of this era was not easy, but you can guess from in a J-FET design the leakage current doubles for every 7 to 11'C. BJT input op-amps including the very low input current super beta transistor types tend to have little bias current change over temperature, but the input resistance should be the same applied to each input. FET types have an exponential input current over temperature, they have a higher input voltage offset and voltage noise. MOSFET and chopper stabilised types, often called zero offset, compete well with traditional types.
In 1983, I evaluated the new Ferranti ZN450 digital voltmeter (DVM) IC for an application. This part required fewer external components than other IC's of the time (such as the one pictured above) because of its principal of operation. It used delta sigma, analogue to digital conversion, which is based on digital averaging. Although one of a broad range of successful Ferranti semiconductors, this particular part, although sampled, never became a commercial part and the company was wound up by the government. Delta Sigma and Sigma Delta (which may be a patent work an around for the same) for high accuracy A/D conversion is very widely used now for high accuracy low to medium speed, wide range or audio applications. Cirrus logic were and still are makers of some of the best available (when I last review in 2012) Cirrus logic - Delta-sigma A/D Converters the company marketed them under the brand Crystal and held key patents in the early 1990s.
The Digital Panel meter pictured works very well, uses star point 0 volt design strategy within a metal enclosed instrument. There are no power planes on any of the PCB's in these designs, and the analogue inputs are filtered by series resistors and the capacitors are placed close to the IC pins. This strategy of providing resistor filtering between analogue functions or approximately transmission line strategy has been carried out throughout the design.
The instrument used low speed random logic, linear power supplies and worked straight from leaving the drawing board without any issues despite there being no extra filtering other than two Y-class capacitors at the mains input and decoupling.
All circuit function blocks have filtering and decoupling, so each is self-contained. I have also used gentle curves on the routes to minimise RF susceptibility (as lightening conductor design) but straight lines with 45' chamfered corners would have been neat easier to trace and work fine - I learnt later. When my design started using microprocessors with buses then 0V plains and power supply filter between PCB became necessary, this was a useful lesson learnt by experience.
Even so, I do not advocate star point 0V strategy - but I will explain how to use that strategy successfully.
Intersil parts like the 1970s industry standard ICL7117 IC, as I said. The Digital Voltmeter IC data sheet only specify the IC's performance at room temperature. The notes may have suggested how the IC would perform under other conditions, but it was not until about 1990 that commercial product data-sheets started to give worse case information.
In the 1970s it was usual for 2% of power transistors within a batch to not meet all specification points such as voltage rating, but by the 1990 this had fallen to fractions of parts per million under many manufacturer's programmes such as the one called "Zero defects".
The engineering sample ZN450 worked perfectly, but unfortunately needed a current source power supply because there was a 5V Zener diode embed in it. That Zener diode was not inconvenient. Ferranti advised me that there was some noise at 10uV sensitivity, though I had not intended to use it at that high sensitivity (+-19.99mV).
I do not know if the 5V Zener diode was a parasitic component made as a side effect of the processor or more likely a part included for a customer, but most parts are designed for a specific customer. IC's usually are designed for a specific customer or industry, but then have general purpose features that make them more widely usable.
CVDT (or Linear Differential Capacitive Variable Transducer) processor,
14-bit ADC, EPROM curve fitting, and 4-digit BCD counter display. There were good ICs for designs that did not use a microprocessor. A CVDT or LDCVT linear sensor doesn't wear out but is much more linear than LVDT, but must be used in a clean and dry atmosphere, unlike an LVDT. Both types of linear sensor are acurate. Although this circuit worked perfectly, it was costly and untidy, and I did not like the project, because it was a step back from what we were already doing, but a quicker but expensive solution. The LH0062 was particularly fast at the time 15MHz, 70V/uS hybrid op-amp late 1970s part. Other parts; ICL7217, ICL7109, 2708-2732 (EPROM not fitted), and random logic.
Observation - Opinion
Mullard (Philips) manufactured many modular components; RF sections for Radio and TV, Pre-amplifiers with tone and volume controls, and Power amplifiers. These were manufactured in the 1960s and 1970s but were displaced by Integrated Circuits by 1980.
These modules used a minimal number of components and relied on component parameters such as gain to be tightly controlled for such a design to be mass-produced reliably. This was not how most other manufactures made transistors, particularly, for example. As an almost universal rule, there is a high degree of negative feed back designed in to every stage of transistor amplification, but this is not the case with the Mullard modules circuit.
Philips probably was one of the leading company's to manufacture consistently close tolerance semiconductors. But this is a trait of European manufactures to produce or design for high volume, high quality and often short production run components.
The UK company Ferranti was very much in the same business developing the I2L low voltage technology for very low voltage wristwatch type applications and a Texas Instruments microcontroller used in the best amateur camera of its time, 1976 the Cannon AE-1. The development exercise for this leading edge in 1972 technology was demonstrated with the ZN414 TRF radio happened to be a popular very cheap IC that would have competed with the Mullard radio modules I mention for the IF (Intermediate Frequency) section of an AM radio.
British company Plessy made some commercial radio section integrated circuits in the 1960s. It is said that British industry spent a lot on research then. This company made the highest speed microwave frequency operating ICS until early in the 1990s.
Data sheets for discrete semiconductors tend to have a few specified parameters, but not all. The same die (chip), for a transistor may have another designation but then have a different set of parameters specified. You will find for example with LEDs manufactured in high volume for high volume users that are specified for light and colour, but low volume buyers get what is left these can be the very high light output LEDs or in the case of automotive microcontrollers the early versions that have more unresolved errata's. You chose the component that specifies the parameters that matter to you.
Developing manual dexterity and physical understanding;
You can put wires together then solder them - the trick is to hold all the wires together bring the soldering iron to the joint then the solder briefly, but long enough for the flux to work and the metals to fully wet, then take the iron away before the junction oxidise. A bad oxidised solder joint looks dull grey, but a good solder joint looks bright silver and smooth but not bulbous (too much solder and not well wetted) when finished. You will burn your fingers a bit, and you must not let the wires move until the solder is cool. It requires, it seems, a person to have four steady hands, and it will always be a difficult task. Some technicians use a small multiple clamp tool, I find holding wires down on a desk with any heavy tools and holding other wires with pliers also helps with the soldering to avoid burning your fingers.
The point is to get a feel in a physical, pictorial and symbolic, spoken and written understanding of electronics. Veroboard and some of the other prototyping products are good. Power supply shown is 0-16V using a uA733 regulator and power transistor and the circuit is constructed on veroboard. The circuit which seemed novel at the time (about 1975) but is not novel uses an input signal offsetting resistor network to ensure the differential input to the regulator IC's amplifier is proved with voltage within its common mode range, although the output feed back in goes down to 0V that would be below the inputs common mode range of the error amplifier.
You can see that my style is to put the effort into the design, not the appearance of the box or biscuit tin that it goes into. I work with others who will do styling to ensure that functionality and presentation is just so. That it to ensure style does not overly compromise function either. My circuit diagram will be clear functional and well documented, and my PCB will have had time spent on it to an artistically functional neatness. But we all have different ways these work well for me.
Think about component function and what you want. The circuit behaviour can be understood by finding an analogy, pick your own, but these can work;
Rubber band for a resistor,
A rubber band for a soft magnetic low Q choke and a spring for a high Q choke. (because a rubber band absorbs energy and turns it into heat like a soft choke magnetic).
Sting for a diode or conductor - you can pull, but you can not push string,
Pendulum or tuning folk has two main components Kinetic - movement (inductor), Potential - height or spring tension (Capacitor).
For a differential input, you might think about holding a tray of cups with one hand under the middle of the tray. Just picturing the function of a long-tail pair in a transistors circuit works, for a differential input, well enough for me.
AL-0045-01B Pendulum analogy of a tuned circuit graphic
Useful links:
I have used the following tools on these pages or have used them in my work;
Electronics Workbench National Instruments - Circuit diagrams - but I am not recommending this tool National Instruments make very good logic programming tools, Lab view.
OrCAD Capture https://www.orcad.com/ - is mostly excellent (still the best circuit capture tool and truly if what you have drawn looks right it is right). It is used with its own PCB Layout now Allegro tool or another PCB layout tool, but that may be less practical or beneficial now. The PCB tool is different to other tools in that you select what you want to do then display, find and select what you want to do it to. More to learn, and you will need to use the technical support, but very capable of fast superb work depending on version and issues. A free capacity limited version is available for download and is slightly different to the Professional version but with manual routing that leaves things very nice tidy.
CADSTAR http://www.zuken.com/en - is comparatively easy to get started with and does a very professional job. Upgradable to high-end tools. Free capacity limited version available (it's not a demo).
https://www.kicad.org/ is a free Linux based Circuit and PCB design tool. I have not used this CAD.
https://www.rs-online.com/designspark/home Have many tools and blogs called designspark, for hobbiests and design help. I have not used these.
More basics in Electronics; http://www.instructables.com/id/Complete-Guide-for-Tech-Beginners/ - the author has made his presentation very clear and kept to the basics.
To discuss these electronics pages, see; Blog page Electronics
Next Page; Engineering Management & Fault finding