12
\$\begingroup\$

(Apologies upfront if this question seems too vague, I'm trying to formulate the problem statement as I ask this)

I'm thinking about circuit assembly in a high component variability / low MOQ environment, from the standpoint of passive components. Assuming identical footprints, I'd like to be able to substitute:

Resistors

  • tighter % tolerance for looser
  • higher wattage for lower

Capacitors

  • Higher voltage for lower

What I'm getting at is - if I'm prototyping lots of different designs, parts inventory management becomes a substantial issue. If I could aggregate by component value, and use the same parts where possible, inventory requirements go down, likely enough to offset the potentially higher cost of the individual items.

Does an approach like this work, or would I break designs by doing this? If this works, what other components can I try to standardize?

\$\endgroup\$
  • 1
    \$\begingroup\$ For resistors, I hope you mean to substitute closer tolerance for looser - that is 1% in place of 5%. \$\endgroup\$ – Peter Bennett Jan 14 '15 at 21:53
  • 1
    \$\begingroup\$ oops - yes, that's what I meant. edited. \$\endgroup\$ – kolosy Jan 14 '15 at 21:55
  • 2
    \$\begingroup\$ Higher power rated devices are generall y bigger and a design might not tolerate it \$\endgroup\$ – JonRB Jan 14 '15 at 21:57
  • 4
    \$\begingroup\$ Higher rated fuses, though... \$\endgroup\$ – AJMansfield Jan 15 '15 at 2:44
  • 4
    \$\begingroup\$ Within reason. Substituting a 1000W resistor for a 1/4W resistor, eg, would likely introduce significant inductance into the circuit. Using a 1W in place of a 1/4W, OTOH, is very unlikely to be a problem. \$\endgroup\$ – Hot Licks Jan 15 '15 at 17:16
21
\$\begingroup\$

Yes, this is valid and commonly done for one-offs produced from local "lab" stock. There is cost in maintain more parts, which can easily dwarf the savings of using 5% resistors instead of 1% resistors in those cases where 5% is good enough.

There are also costs in production for each different part used. Even at high volumes, the pick and place machine has to be set up separately, different reels need to be bought, kept somewhere, etc. Unless you have a very high volume product, it makes no sense to use a 10 kΩ 5% resistor in one place when you need to use 1% tolerance of otherwise the same resistor elsewhere on the same board.

In other cases you have to be careful that the better spec in one dimension doesn't cause tradeoffs you care about in another dimension. For example, you mention higher voltage capacitors substituted where lower voltage is required. That's OK electrically, but higher voltage caps will be physically larger most of the time. The same is true for higher wattage resistors. Electrically, a 2 W 100 Ω resistor is a superset of a ¼ W 100 Ω resistor, but the 2 W resistor will be significantly bigger, which may incur other costs.

\$\endgroup\$
12
\$\begingroup\$

Yes, this works, and we find it useful for things that are produced in relatively small volume (thousands or fewer). If you can purchase a reel of resistors or capacitors, it can be significantly cheaper and more convenient than trying to deal with several types of parts in smaller quantities (like 200 at a time).

Don't go nuts with it, but things like bypass capacitors, diodes and resistors, jellybean transistors and some kinds of regulators are cheap enough that it makes sense to stock up. Sometimes a bit bigger part can make sense (as in ceramic bypass capacitors) where 0402 may be the cheapest part, but 0603 are easier to handle, and come in smaller reels (4,000 rather than 10,000) and you can get 16V rather than 10V for a similar price, so they work in more places.

I tend to use 1% resistors for most purposes where a network (5%) can't be used, and where a good resistor (0.05% or whatever) is not required. That may mean 1% pullup resistors sometimes, and that's okay. Volume stuff gets the 5% resistors, of course.

If it turns out that something is going to be made in serious quantity, you're going to want to go over the BOM in detail anyway, so I don't see much downside (you should put the engineering in up front anyway to determine the tolerances required). Be careful that the higher spec parts are actually higher spec'd in all respects and that you're not missing something like voltage rating, temperature range, or surge rating. A single field failure can wipe out an awful lot of savings.

\$\endgroup\$
5
\$\begingroup\$

In general there aren't that many issues.

However, higher wattage/voltage ratings usually mean larger components. If you're just doing bread-boarding or have ample PCB space, these aren't that big of an issue (it may even be helpful to have larger components to route things through them).

Additionally when it comes to anything with high frequencies size can start to play a role in both the PCB layout and intrinsic properties of the components such as undesired inductances and capacitances.

\$\endgroup\$
5
\$\begingroup\$

In general, you should have no problems, but always keep your policy in mind.

If you are prototyping and producing pilot runs, then full-scale manufacture will no doubt involve using the minimum specification components. If you produce a PCB design, then it may be well to allow for min-spec as well as your-standard PCB holes for instance, so your standsrd 50V single-ended cap could be replaced by a physically smaller 16V type without bending the component leads on the production line.

Weight hasn't been mentioned, but is unlikely to be a consideration unless you're producing for NASA. But then, you'd be using Milspec, no doubt.

There will inevitably be the odd special-circumstances that you'll trip over. Here's a few war-stories, not necessarily related to passive components:

In the early days of microcomputing, programs were stored on and retrieved from cassette-tapes. Long-forgotten technology now, but at the time, a cheap-and-cheerful recorder was king. The "professional" standard machines were so intent on compensating for this, that and the other thing that they doctored the recording to the point of unusability.

Then there was the incident where a pilot run was produced with kits supplied by the design company from their standard stock. There was 1 drop-out from the 600 produced (and that was an incorrectly-mounted diode) against an expected 1% drop-out rate. The manufacturer went into full-scale production and rang in a panic with 30%+ failures. The cause was traced to low-spec transistors. Only just within specification and not following the expected specification distribution. It appeared that what was being supplied was the low-end "hobbyist" grade rejected by the big Asian manufacturers and sent to marketing backwaters. Our local manufacturer aded an extra filter - processing the low-grade devices through a beta-tester and returning the very lowest in the range to be sold off at the local electronics stores...

Finally, there was a device that consistently failed when one manufacturer's chips were installed at a particular point. They were consistently counting between 2 and 3 pulses when only one was sent. The lightbulb moment was when the manufacturer's representative exclaimed that they only ever released devices that were "three times that clock-speed." The trigger pulse was very short and the failing devices were counting both the pulse and its reflections...

\$\endgroup\$
4
\$\begingroup\$

I see this question has been answered but I'll mention a couple of things. Firstly SMT X7R, X5R capacitor voltage ratings and how capacitance changes with applied voltage. Take a look at this: -

enter image description here

It's telling you that the actual capacitance of a lot of common capacitors varies significantly with applied voltage level. If you need capacitance to be stable with varying voltages then use a better capacitor or one that is rated for higher voltage (% change in C is much less).

Article that shows this is here

MIL-HDBK-217F (here) is still a pretty significant piece of text for determining MTBFs of electronic circuit boards and what underlies a better MTBF is the de-rating of components to reduce their circuit stress. Table 3-2 (page 3-4 and 3-5) tells you the types of environment to pick for analysis and ranges from "Ground, Benign" thru "Space, Flight" to the highest level, "Cannon, Launch" - each "service" category places a significantly different weight on the de-rating value for all commonly used components.

For instance, the diode - the type should be chosen and the basic reliability can vary by a ratio of up to 6.57:1. Then there is voltage stress factor to consider - if the applied voltage is less than 30% of the rated voltage then the "stress figure" is 0.054 whereas at 100% rated voltage the "stress figure" is 1 - basically this means that you can expect s diode to last 18.5 times longer when it only has 30% (or less) of its rated voltage applied.

I'm not going any deeper into this but it is a very significant reason to choose components that are seriously over-rated.

\$\endgroup\$
  • \$\begingroup\$ Out of curiosity, do you have any idea whether it would be possible to produce a cap whose capacitance was maximized as some voltage other than zero? An XSR cap is analogous to a water tower which is very fat at the bottom but narrow at the top; it's much more useful, however, to have a water tower be fat at the top and narrow everywhere else. Putting a battery in series with the cap would reverse things, but none of the electrons from the battery would have to go very far (net charge flow through the cap would be zero). Perhaps using different metals for the anode and cathode... \$\endgroup\$ – supercat Jan 15 '15 at 20:35
  • \$\begingroup\$ ...might yield behaviors somewhat analogous to having a battery in series? \$\endgroup\$ – supercat Jan 15 '15 at 20:36
  • \$\begingroup\$ @supercat I'm no expert on 'em - I just use them and get caught out by 'em every now and then! \$\endgroup\$ – Andy aka Jan 15 '15 at 20:40
  • \$\begingroup\$ @supercat Should be possible if you apply a high voltage as the ceramic cools through the curie temperature, but then the cap would be polarized. \$\endgroup\$ – Spehro Pefhany Jan 16 '15 at 4:34
  • \$\begingroup\$ @SpehroPefhany: Polarization would be implied in such a scenario; I wouldn't generally expect a cap to have a low capacitance when the differential voltage was small, but a high capacitance when biased in either polarity interchangeably (though thinking about it, such a thing could occur with microphonic effects, though probably not in such a fashion as to improve efficiency). For intermittently-powered circuits, applied voltage would go between zero volts and nominal voltage, so having a polarized part wouldn't be a problem. \$\endgroup\$ – supercat Mar 11 '15 at 17:16
2
\$\begingroup\$

I don't see any reason why not, but keep in mind a few things. I doubt you'd be able to beat the pricing of a 5% resistor by buying lots of 1% or .1%s, maybe depends on volume. Also these are not the only characteristics of these components to worry about. For capacitors the dielectric material may matter for example. Also the designer (or you) may have carefully picked a resistor that meets environmental requirements. Further parts from a reputable manufacturer do not always equal parts from another... Nothing like chasing down an issue that turns out to be a counterfeit part or the like.

Just pointing that kind of stuff out, many times have I had to hunt down purchasing or manufacturing because someone thought it was no big deal to substitute the cheaper part they saw or one that was more available...

\$\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.