From my own experience, the process to select the different circuit elements starts with a good list of specifications. With this information you can define different architecture alternatives, and based on it, rough requirements that will allow to select the components.

If you are lucky enough, you will find the desired parts that will fulfill the requirements at least for one architecture. Then you can start the design using this parts and complete the process.

If you cannot find the parts, you would need to continue to the design with those that provide the most promising characteristics, and see which is the more limiting devices. Then it could be possible that the devices selected reach the required specs. In this case you can finish the design.

It is also possible that you find that one characteristic of the preferred device does not allow to reach the requirements. In this case, the solution could be select a new device that fulfills with this requirement and the others.

If not, you can try to find a new architecture, new devices, or revise the specs and see if they can be relaxed.

So in brief, depending on the design characteristics and complexity, the process of selecting can be straight-forward, or an iterative solution. But in any case, you need to go back and forth between the design and parts selection in order to optimize both.

I totaly agree with José María Gómez Cama. The only factor that I want to add is that you have to consider your budget. Thus, there is a trade of between the following parameters:

1- Budget

2- Specification

3- Result accuracy (which depends on architecture and quality of elements)

First requirement is the purpose (Testing or Real Implementation). Normally for testing purpose the final design will be used for say 1 minute to 1 hour with all in perfect environment. But if the design is to be used somewhere else by someone else, then the design must be perfect.

The safety range value of all the parameters of all terminals, inputs and outputs should be set at first. Accordingly the best matched component should be choosen based on its datasheet. Working Environment should also be taken into consideration for supply variation and load variations.

Dear Rohit,

Your question is interesting and the colleagues gave satisfactory answers to it.

However my comment is that there is no best and worst circuit elements and components.There are devices and components having certain specifications and performance parameters. The system level design results in the determination of

of the specifications of its components. Once these specs are known you can choose from the component list or library the best match to that required.

When making design by analysis, you can build your circuit from common components and make estimation of the unknown elements and then analyzing using a circuit simulators

I think you want to build a stock of circuit elements and components in your laboratory. Normally, this stock is formed from the most common integrated circuits such that op amp 741, 747, the pll 565, the TTL, CMOS logic families,..etc.

The high performance components can be purchased after the circuit design and simulation.

wish you success.

Dear dr Vasile,

You rarely up vote an answer. Thank you for voting up my answer and for your valuable additions to the answers. I agree with you to build the circuit with reliable operation for long time and all possible operating circumstances and conditions.

The simulation programs are very useful and are used for automated fast analysis of the complicated circuits.They also save the experimentation work. In addition, the circuit simulator are advanced so that they give results in close agreement with experimental results. Integrated circuits can not be designed without cad tools including circuit analysis and lay out. Theory and experiment are two faces for same coin.

Best Regards

Most of the important aspects have been answered but I appreciate the comment made by Abdelhalim Zekry that theory and experiment i.e. the theoretical/simulated results must be complimented by hardware/practical realization and these are in-separable sides of the same coin. It means that, the development process has to take place concurrently.

Coming back to selection of components, primary consideration should be given to local availability and costs.

It is not the components alone contribute to the best possible circuit performance, but "clean" power supply, environmental aspects such as operating temperature, electrical noise (EMI/EMC) and proper grounding and shielding (means proper PCB layout) are equally important factors that one cannot neglect.

Prasanna hits on another key point--availability. The part may meet the requirements/specs but has a long lead time of say 10 months but your project due date is 6 months from authority to proceed. Also, along the lines of availablity.. be aware of maturity of the part. The part might meet the requirements but little reliability data is available and/or instructional data on the use of the part has issues.

George,

I do agree, but when we say "LOCAL AVAILABILITY" that means either off-the-shelf availability or an acceptable lead time - 1-4 weeks, depending upon the quantity and production schedule. Therefore any lead time in excess of 4-6 weeks is indeed a matter of concern. Certainly then in a project time frame of 6 months end-delivery, we cannot consider a part having 10-months lead time. It is not even wise to list such a part.

Secondly, the maturity of the product - For initial studies and small prototype quantities this may not affect as far the part is available for atleast a year's time ahead. When we want to launch a production volume, in a long term market, then YES, the PRODUCT MATURITY comes into role. This can be always sorted out with the silicon manufacturer or its distributor. And I believe silicon vendors to provide sufficient notices to its regular customer of advance purchases in case of a product reaching its maturity or the foundry services are planned for a closure.

Thanks any way for interesting dialogue!

Regards to all in this discussion!

-Prasanna Waichal

And George,

Sorry for missing the point. At NASA, you may want MIL833 or radiation hardened SPACE-GRADE parts and for a space probe scheduled to to launched, the reliability data IS A MUST.

In academic research environment, cost and budget constraints always must be limiting to selecting the parts of mostly commercial grade or of the industrial grade.

Here, focus is more on "PROOF-OF-CONCEPT" rather than releasing any qualified finished product to be deployed in harsh / extreme environments.

Hope you agree with me!

Regards,

No problem. I was merely responding to the question from my prospective. The question did not qualify from what stage of development and industry. Yes, NASA is a bit stringent on the environment. However, whether it is a prototype or a flight system, one must be cautious on the availability/maturity of the parts. I have been bitten on several occasions with commercial parts that were sent out the door with many issues. It can cause many delays. We are dealing with that now with a part that the literature is nebulas and incomplete.

Rohit,

This is an interesting question and some aspects have been answered by others. Your choice of components would largely depend on your design goals- as expected that would involve some trade-offs and you would need to make a judgement based on your design criterion. I find it very useful to simulate the circuit in first instance using a free software like LTSpice or if you have licence for PSpice or Mulitisim. However, I agree with others that availability of components is important together with their costs (if cost is one of the critical design issues). You can then change the components and analyse the effect.

This could be followed with a breadboard evaluation of the circuit.

Hope this helps,

Tahmina