Quality components

[Mr. Scram]

Anyone who has seen a number of EEVBlog videos, will have heard Dave talk about cheap Chinese capacitors. View a few more and you may have heard about Nichicon and Panasonic. It is clear that within the huge pile of available components, not all are created equal. What is less clear, however, what these differences entail and what good strategies are to ensure that good, or even the best components end up in your product. Is brand any or a good indication? What role do model numbers and specifications play? Do differences on paper mean anything in the real world? Are resistors in those regards different from capacitors, plugs or cables?

This thread is intended to share experiences and strategies employed to ensure a high quality product. A lot of products are built down to a price, but sometimes that just is not good enough. What do you do when you need the best you can get? Let's compile a list of what to look out for and take into acocunt:

Sourcing:
- Obviously, buying your components from a reputable reseller rather than from a Chinese bargain site is a good start.

Capacitors:
- Reputable brands: Nichicon, Panasonic
- High hour rating
- High temperature rating
- Avoiding non-solid capacitors?
- Low ESR?

Resistors:
- Accuracy indicative of quality?
- Metal film?

Connectors:
- Reputable brands: Amp/TE Connectivity

Switches:

Cables:

PCB:
- Material?
- Layers?
- Finishes?

[DerekG]

What do you do when you need the best you can get?

If your design generates heat, then 105C electrolytics will generally always have a longer life than 85C & choosing a capacitor that is rated at say 12000 or 14000 hour operation at the rated temperature is the best way to "put off" any failures. Low ESR from new is often important depending on your design requirements.

Billion have received a canning with customers avoiding their modems & routers after using poor quality caps in them over the past 10 years.

[dmills]

It starts with design.
No matter how "good" a component is it will have poor reliability if run to hot or too near the abs max spec.

For me, electrolytic caps are possibly the obvious place that buyers like to play games, the good stuff is expensive and the poor stuff looks like the good stuff if you don't get it.  I have a mental short list of the acceptable in a quality product, it features the usual half dozen or so suspects. 
I would note that there is more to electrolytic caps then people often think, **READ** the datasheet carefully, ripple ratings vary with frequency, and remember that it is not just the fundamental, but also the harmonic currents, a properly applied 85 degree part can last longer then a poorly thought out application of a 105 degree unit.
Low ESR is NOT always a good thing, pick the part to have the appropriate characteristics for the job in hand, a 1000uF 40mR polymer elco on a big power plane can be trouble.

MLCCs vary far more then you would expect between manufacturers, especially in some of the high K ceramics, read the long form datasheet and pay attention to what happens as the package size drops (Also pay attention to the crack prevention guidelines).

Resistors are sometimes also a lot more complicated then you would think, thick film often has measurable excess noise and some linearity issues for example, if designing stuff where that matters specify the part and manufacturer, often it is a small number of parts in a much bigger BOM, but that set of 10 resistors being expensive thin film ones can lower distortion by a factor of 10.

Sand, not all versions of the 5532 (or any other chip) are created equal, take the time to qualify what you use.

MOSFETs, you can safely ignore the first page current rating on the datasheet, it is almost always fantasy, also the RDSon figure only applies at specified junction temperature and gate voltage, if you are lucky some work with a calculator and the graphs further into the datasheet will tell the real story. Be careful of substitutions here as well.

Board materials, not all FR4 is equal, mainly a high frequency thing, but the higher Tg boards have lower thermal expansion, Tg150 is not much more expensive then Tg130.

Design for test, zero ohm links between sections of power planes makes fault finding so much less annoying.

Basically it comes down to use suppliers (and supply chains) you are comfortable with, do not skip on the fine print in the datasheets, and be conservative.

Regards, Dan.

Regards, Dan.