11
\$\begingroup\$

I was wondering what developments are done on electrical components today.

Do manufacturers try to make the electrical components they sell as close to the ideal model as possible? So for example, is the goal for their fixed resistors to behave according to Ohm's law for the widest range of voltage values and frequencies? Or are there benefits to non-linear behavior that differs from the ideal model in some use cases?

| improve this question | | | | |
\$\endgroup\$
4
\$\begingroup\$

Often, but far from always, the aim is to replicate the behavior of an ideal component, at least over some range of frequency, voltage, temperature, whatever.

Sometimes, however, manufacturers intentionally stray away from the ideal because a certain degree of "non-ideal" behavior is desirable for the typical application of a component. Consider bypass/decoupling capacitors. If you have worked for long in electronics, you know of the need for capacitance between the power and ground of your circuit.

For example, from a manufacturer's perspective, TDK has a line of controlled-ESR ceramic capacitors intended for power supply bypassing/decoupling. Although an ideal capacitor has zero equivalent series resistance, the ESR of these capacitors is intentionally moderate. Indeed, they have actually spent more money on each component in order to raise the ESR, and thus the cap is even further from the supposed ideal than their other MLCC caps. If you have ever designed or specified the performance of a power distribution system, you'll know that too high ESR means your bypass caps are not effective, but too low of an ESR can create resonances in your power system, increasing voltage ripple. MLCCs often have problematically low ESR, so TDK is trying to make components which solve this problem.

From the perspective of an engineer applying bypass caps, it's better to choose lossy ones (e.g. X5R, X7R dielectrics) than the high-Q C0G types: your power system will have less ripple. Were you making an RF filter, maybe the high-Q caps would be a better tradeoff.

So sometimes components are intentionally non-ideal because that's what's best for the typical application circuit. I have found it best to understand the types of non-ideal behavior exhibited by particular components and try to "design it in" to the circuit.

| improve this answer | | | | |
\$\endgroup\$
  • 1
    \$\begingroup\$ IMO a controlled-ESR cap should be considered a capacitor and a resistor in one housing, each of which is manufactored towards the ideal values. \$\endgroup\$ – leftaroundabout Jun 2 '19 at 21:53
21
\$\begingroup\$

Yes - but to a budget.

For example, in the case of resistors, there are various tolerances available which tell you how much the actual ohm value may differ from the stated value. 5% tolerance used to be standard, these days 1% is not significantly more expensive. If you want a 0.001% tolerance resistor you'll have to pay more. Similar things apply to the temperature coefficient of resistors.

| improve this answer | | | | |
\$\endgroup\$
  • 6
    \$\begingroup\$ Right, that's a slightly different question; would the thermistor count (a resistor which deliberately has a large temperature coefficient)? Some semiconductor devices might count too: tunnel diodes, avalanche diodes. Various audio designers seek out nonlinearity if it "sounds good" - use of vacuum valves, certain types of transistor. \$\endgroup\$ – pjc50 May 31 '19 at 12:34
  • 1
    \$\begingroup\$ In an ideal world, manufacturers make their components behave as close to the datasheet as possible. But real world data sheets never describe "real world" components - if they did, there would be no need for the data sheet at all! \$\endgroup\$ – alephzero May 31 '19 at 20:39
  • 6
    \$\begingroup\$ @blue The classical example of a non 'Ideal' component actually being very much superior to an 'ideal' one is a capacitor used across the power input to a board. You actually WANT some series resistance here to avoid the LC network formed by the supply cables inductance ringing (The resulting voltage can destroy the regulator on the board). An electrolytic with a few ohms of ESR is actually vastly superior to a low ESR part here. You can also buy ceramic MLCC caps with a certain amount of deliberate resistance, handy for decoupling with a bit of damping. \$\endgroup\$ – Dan Mills May 31 '19 at 21:51
  • 2
    \$\begingroup\$ @ThreePhaseEel Indeed, also using inductor ESR to sense current in a current mode switcher. A decent designer understands the parasitics and uses them to bring in a simpler design. \$\endgroup\$ – Dan Mills Jun 1 '19 at 6:21
  • 1
    \$\begingroup\$ @DanMills -- there are also designs that use FET Rds(on) for that job, for that matter:) \$\endgroup\$ – ThreePhaseEel Jun 1 '19 at 13:37
9
\$\begingroup\$

Resistors have a temperature coefficient of resistance. Wire-wound resistors have inductance. Composition resistors have inductance also, but more like any piece of wire has inductance.

Capacitors have series resistance, leakage and temperature sensitivity.

Inductors have series resistance and may have significant shunt capacitance and magnetization non-linearity.

All passive components have a value tolerance. All are sold in various grades and types at various prices to offer solutions to non-ideal behavior for applications that require something better.

Active components and devices have similar shortcomings with many product variations and design methods used to compensate.

| improve this answer | | | | |
\$\endgroup\$
  • 3
    \$\begingroup\$ I remember using non-inductive wirewound resistors (counter windings to minimise inductance). \$\endgroup\$ – Peter Smith May 31 '19 at 13:06
  • 1
    \$\begingroup\$ Capacitors not only have series resistance, they have series inductance too. \$\endgroup\$ – Uwe May 31 '19 at 22:34
4
\$\begingroup\$

Do manufacturers try to make the electrical components they sell as close to the ideal model as possible?

Of course not. Doing so would be an enormous waste of time, effort, and money. They only make parts that are good enough to do the job their customers need them for - any more would just raise the price and make their products uncompetitive.

Take a simple resistor for example. What are its ideal characteristics? Zero tolerance, zero capacitance and inductance, stable and linear to infinite voltage, infinite power dissipation, infinite current handling etc. But even if such a device was possible it would be grossly over-specified for most designs. Some people might need a resistor that can handle 1MW at 500kV, others might only need 1/4W at 5V, but nobody wants to pay more than they have to.

In all cases the circuit is (or should be) designed to work with practical components that have non-ideal characteristics - sometimes very much so. And sometimes the circuit is actually designed to take advantage of it. A transistor doesn't work like any 'ideal' component - but it's still useful. Transistors usually have wide tolerances, and all have unwanted characteristics that would make an idealist cry. A typical circuit may have dozens of other parts whose sole purpose is to compensate for the transistor's 'flaws'. But that's still cheaper than trying to make a more 'ideal' component.

The main reason for wanting 'ideal' components is to make circuit design easier. However in practice they don't have to be perfect, just good enough that the circuit works as intended. Op amps are often used in circuits that could work better with discrete components, but would be more difficult to design. Many products use older 'industry standard' parts simply because the designers are more familiar with them, and manufacturers continue to churn them out in the millions despite being superseded by more modern parts with better characteristics.

| improve this answer | | | | |
\$\endgroup\$
3
\$\begingroup\$

If there is a reason why someone would want a more ideal version of something and it's possible to improve on an existing design. Even if the new thing would be really expensive, so long as there is enough demand for it, then of course it will be made.

Though many new designs and improvements are based around making things smaller, more energy efficient, etc while maintaining the current levels of ideal-ness.

There is also the problem that many components have multiple characteristics and what you think is more 'ideal' in your application could be less so for other applications.

Then you have components that were not designed with a use in mind, but turned out to be good for some use. And of course people do use the flaws in things as a feature to do what ever crazy nonsense they dreamed up.

So it starts to get blurry as to what an ideal version of something actually is. We also cannot produce exact copies of things, there is always some amount of variation. So there always has to be some level of tolerance.

I think the most obvious examples of things that do keep better are power conversion and electric motors. Where your input to output ratios have improved a lot over the years and also power consumption, we keep getting things that take less and less power to do the same thing.

| improve this answer | | | | |
\$\endgroup\$
2
\$\begingroup\$

Of course they do.... Most of the time. Reputable manufacturers will always want to provide the best they can. However, some things just cannot be achieved. Take op-amps for example, there is just no way you can get an infinite open loop gain for example. It is also impossible to have infinite input impedance or zero output impedance. But manufacturers will try and get as close as they can.

Fixed resistors will always behave according to Ohms Law. All the manufacturers can do it make it as close to the specified resistance. That is why they have tolerances.

Trying to make ideal components, or extreme precision components cost money, so there will always be some trade-offs that means nothing is ever going to be idea.

In short, a reputable manufacturer will do their best to try and give the best product they can, within an allowed budget. The better the specs, the more expensive the product is to make, and then the more expensive it will be to buy.

| improve this answer | | | | |
\$\endgroup\$
  • 8
    \$\begingroup\$ Fixed resistors don't always behave according to ohm's law. They generally increase resistance as current increases because the resistivity increases with temperature. \$\endgroup\$ – Charles Cowie May 31 '19 at 12:30
  • 3
    \$\begingroup\$ Manufacturers will always want to provide the best they can I think that statement is really not realistic. \$\endgroup\$ – Huisman May 31 '19 at 12:43
  • 1
    \$\begingroup\$ I have made a few small edits to try and address the comments \$\endgroup\$ – MCG May 31 '19 at 14:48
  • 2
    \$\begingroup\$ Most engineers want to provide the best product possible. Manufacturers mostly want to make money. A project manager I worked with had a saying: "At some point in the project you need to shoot the engineers". Meaning that, at some point, you need to stop the engineers from tinkering with the product trying to make it better. \$\endgroup\$ – Mattman944 May 31 '19 at 15:39
  • 3
    \$\begingroup\$ Fixed resistors certainly don't behave according to Ohm's law for high power, low frequency signals. The temperature of the resistor cycles up and down at twice the signal frequency in sync with the power dissipation, creating measurable harmonic distortion. \$\endgroup\$ – alephzero May 31 '19 at 20:36
1
\$\begingroup\$

Consider the "Ideal" opamp: enormous gain-bandwidth, huge output drive current==huge output transistors, zero standby current, zero settling-time, stable for all positive dB and -dB gains, zero cost == zero die area.

Notice any conflicts in these "idealities"?

Thus there is no one single "ideal" opamp.

| improve this answer | | | | |
\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.