Adjustable parts. Why I avoid them when possible.

Adjustable parts will eventually require adjusting. Those words of wisdom were given to me some decades ago. There are some examples of where adjustable parts were used but they are not really necessary in many cases with modern components. Sometimes there is no choice, of course, but my view is that those occasions should be few and far between.

In a previous post, I mentioned that many university and college labs appear to have a stash of 741 opamps and while they were really good for their time (it was the first 'superstar' component according to many) it is not something I would be using for a modern design.With an input offset voltage (Vos) of up to 5mV at 25C, any precision designs with high gain really required us to use the offset null terminals. Modern amplifiers of modest frequency response, such as the ADA4522 have astounding specifications when compared to those older devices. The single part has the capability of nulling the offset but in the majority of cases it is simply not necessary. Interestingly, the 741 is still in production and features quite large in some of the equipment I work with to this day.

RF historically had many adjustable parts, particularly inductors and capacitors in filters and tuned circuits and while those type of system exist today, precision RF can be achieved without them. The advent of frequency synthesis and ultra stable oscillators have, in many cases, obviated the need for them. Synthesis has been with us for several decades although the parts available today are far better than those of even 10 years ago.

That is not to say adjustable parts are going to disappear; for a simple, low cost circuit they can be ideal but if I am designing a precision circuit where cost is not a major issue, some extra circuitry can achieve results equal to or better than those with adjustable parts.

For circuits that require calibration, it might seem counter intuitive to eliminate adjustable parts, but by using a microcontroller and a lookup table, circuit errors can be characterised and corrected in software. This is particularly true in test equipment but is not limited to that market. Precision hardware can be realised this way for many areas of electronics. With a bit of thought, a system can actually be made to be self-calibrating using this method.

On the offset null, nulling can give a false sense of security as it does nothing to address the temperature dependence of the circuit and this is true of many circuits; adjustable parts also have their own temperature dependencies.

Adjustable parts are often used to vary the brightness of LEDs (as an example) but PWM drive of the device yields the same result (provided the |PWM is fast and produces no visible flicker) without the wasted power inherent in resistive loads.

In many cases practical alternatives to adjustable parts exist and when used properly mean that production items should never have to be, well, adjusted.

Comments are welcome as always.