18
\$\begingroup\$

I'm curious why some IC manufacturers like Maxim, Analog Devices, and Linear Tech's parts are so expensive? I know they are much higher quality (better electrical characteristics, etc), but for some parts it still doesn't make sense. For example:

SD Flash Media Controllers (not identical, but to do the same thing)
Maxim MAX14502: $21.69 Mouser
Microchip USB2244: $2.18 Mouser

RS-232 Transceiver (not pin-to-pin identical, but do the same thing)
Maxim MAX3232: $5.92
TI MAX3232: $1.91
ST ST3232: $1.06

Accelerometer. This might be an extreme example. They're a little different, but both are 3-axis, digital accelerometer with the same sensitivity.
Analog Devices ADXL362: $9.73 (!?)
Freescale MMA8653FC: $1.09

And there are tons of more examples.

I'll try to frame the questions so this doesn't become a discussion. As a design engineer why would you want to select a part that is 5 to 10 times more expensive? What could be some technical reasons?

Obviously there are markets that require very high quality components, but is the demand for pricey, quality components even that big? How can the expensive-chip manufacturers compete with others that can make it so much cheaper?

\$\endgroup\$
  • \$\begingroup\$ An "Approved Vendors" list. \$\endgroup\$ – Ignacio Vazquez-Abrams Sep 10 '14 at 23:08
  • 4
    \$\begingroup\$ Didn't check the full list but a MAX14502 is end of life and even small quantities of the MAX3232 are much cheaper directly from Maxim. This really amounts to how do Mouser determine their pricing and what price breaks manufacturers have between the MOQ and a million pieces. \$\endgroup\$ – PeterJ Sep 10 '14 at 23:17
  • 7
    \$\begingroup\$ In the case of AD and LT, it's because they can. It's worth the extra money to buy parts that you know are going to meet or better the published spec. For example, I've never assembled an LT SMPS controller circuit to their design and had it NOT work, perfectly, first time, and meet (for example) their quiescent current spec. The same can not be said for other manufacturers... \$\endgroup\$ – markt Sep 10 '14 at 23:58
  • \$\begingroup\$ The price published by Mouser isn't what everyone is paying. As a design engineer it can be frustrating that you need to have several meetings to establish real prices. Pricing can be very flexible for silicon, even for customers buying as little as 1000 pieces. Often 2 customers buying the same volume of devices can be paying very different prices. You may use a vendor where you have a contact who you know will give you a decent price straight away. \$\endgroup\$ – Will Sep 29 '14 at 17:16
26
+50
\$\begingroup\$

Disclosure: I currently work for one of the manufacturers mentioned, I completed an internship with a second, and I know current and former employees of a third. I can't reveal specifics but I can give some general reasons why ICs have variable costs and prices. I also can't speak about the specific ICs mentioned -- even if I knew why my company's version is priced the way it is and could reveal that information, I couldn't possibly know why other companies priced theirs differently.

There are many reasons -- both technical and non-technical -- why the price of one manufacturer's IC may be significantly higher than another's. Below are some of the major ones. Some or all of these may be true for a particular case, and manufacturers may be in different price positions for different IC types (e.g. op amps, ADCs, voltage regulators, etc.).

Technical Reasons

Fab/Assembly

Every manufacturer has a different fab process (actually many processes), and the process used by a manufacturer may have better performance for a particular application or the process may be more expensive (which of course drives up the cost and price of the final IC).

Even with identical processes and circuitry, though, the materials used in the assembly of the IC can vary in performance and cost. For example, a higher quality mold compound reduces stress on the die and therefore improves performance over temperature...at an increased manufacturing cost. To further reduce stress on the die a polyimide layer may be added. Another material choice that can affect performance and cost is the wire bond material -- for example, it is easier to meet or exceed qualification standards (e.g. temp cycle) with gold wire but gold is more expensive than copper. The added cost of higher quality materials may be important for applications which require long lifetimes, severe temperature swings, etc., but would be an unnecessary expense for shorter term applications with little temperature variation.

Test

Production test also has a large effect on overall cost and quality. Virtually every IC requires some sort of trim (e.g. laser trimming) for at least an internal bandgap voltage reference or oscillator, and possibly for offset reduction, gain correction, etc. Adding additional trims and/or trim bits can improve the performance of the trimmed IC at the cost of increased test time (which is increased test cost). Trim may also require the addition of non-volatile memory, which may require additional data retention tests that also increase the test time. The fab process may even dictate whether trimming is done at wafer probe or final test (i.e. after the die is packaged); wafer probe generally has higher throughput (so it's cheaper) and allows the manufacturer to throw out bad die before spending money packaging it, which of course reduces overall cost of test.

Also, while every IC is tested (at the very least for continuity) overall test coverage can vary. Some applications like defense, automotive, or medical require very low or 0 DPPM, which requires the manufacturer to fully test all electrical parameters (possibly over temperature, which significantly increases test cost). Other applications do not require such low DPPM and the manufacturer may choose not to test in production certain electrical parameters which demonstrated a high \$C_{pk}\$ during characterization, especially if those parameters have a long test time or require more expensive tester equipment. Skipping these tests can result in a significant cost and price reduction with very low but non-zero risk (due to the high \$C_{pk}\$) of passing a die that does not meet the spec, which may be worth it to customers in less critical applications.

Non-technical Reasons

One non-technical factor affecting price is which manufacturer was the first to market. This manufacturer has a temporary monopoly or near monopoly and can command a higher price. This manufacturer may spend less time optimizing their production for lower cost in order to be first to the market. Manufacturers which enter the market later tend to optimize for cost to undercut the manufacturer that was first to market since a customer will not switch to a different manufacturer for an identical or nearly identical IC at the same price. The manufacturer who was first to market may still be able to command a higher price if they have established design wins with large customers who do not wish to qualify a new manufacturer's IC even if the new IC is offered at a lower price.

Also, a manufacturer's prior relationships with major customers and perceived reputation can allow it to charge a higher price. Major customers may be willing to pay extra if they have an established relationship with a manufacturer's support teams and/or if the customer(s) have had quality problems with a different manufacturer in the past.

In short

Ultimately, a manufacturer's price depends on which market it is targeting: some customers are relatively low volume but have very high quality needs and are willing to pay for it (e.g. military, automotive, and medical) whereas other customers have much higher volumes and every penny counts. ICs manufactured for critical applications depend on higher margins to make up for relatively low volume, use better quality materials, have more extensive test coverage, etc. ICs manufactured for less critical but higher volume applications optimize cost to deliver lower priced ICs which make up for the lower margins with much higher volumes.

\$\endgroup\$
7
\$\begingroup\$

You have to be careful when you say "they do the same thing." True, the same basic function may be there, but different uses measure quality in different ways.

For example, the LM324 opamp is dirt cheap and the LT1014 of roughly the same vintage costs 10x more. They're both op-amps and will both "do the same thing" but the LM324 offset voltage, for example, is 10's of millivolts while the LT1014's is measured in microvolts. If I'm amplifying a piezo knock sensor signal to tell when a box is shaken, I'll use the 324. If I need to amplify a thermocouple, I'll use the 1014.

I took a glance at the accelerometer data sheets and the Analog one specs things that the Xtrinsic one doesn't, like linearity and cross-axis specification. If those things are important in my application, I can either just buy the Analog part, or call up Xtrinsic and try to track down the info. And then I don't have assurance that it won't change in a later batch, because it's not a property they guarantee.

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

Q. Why would I pick a part 10 times more expensive?
A. I'm building to military specs, for aircraft / space / submarine / whatever, and I need parts guaranteed to work under extreme conditions.

Those other parts MIGHT do the job, but Jupiter's orbit is a long way away [Satellite], or people's lives are a stake [submarine emergency systems].

[Edit] A compny guaranteeing that kind of performance is doing a lot of Quality Control checking on more samples per batch. They may even test EVERY chip. If a spec sheet gives a rating of [say] 100 volts, the actual part is probably tested to 120 volts.
That kind of quality control testing adds a lot to the cost of a component.

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

There are a couple of reasons: The 1st one is that these companies offer various parts that comply with very strict standards which are mandatory in certain industries. At some cases, you can see this in the datasheet or on the website.

2nd reason is that these companies tend to meet the specifications from the datasheet every time. This is something you could not care less about in low-tech products but in some special cases you base your entire accuracy and / or reliability on these spec. so you want to know you can trust the vendor.

3rd reason is that these companies are "boutique" among IC companies. They offer the highest quality products and the best support you could wish for. I used to work for a large company who worked a lot with ADI parts and we had a very good working relationship with them, which was very different from the relations we had with different parts makers (which is little to none).

The last reason is mainly because they can! They tend to make parts that target industries like medical, military, aero-space and such... Usually these companies make very expensive products so they could not care less paying $10 for a good quality op-amp because the rest of the system might come up to several hundreds of thousands or more.

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

Price is negotiable and arbitary. If you are ordering 1- 10 off's per annun as an engineer price may not be an issue for you.

If you are ordering low volume, < 10,000 per annun, price is going to be an issue and you would need to select best pricing options from the major distributors. Also working closely with the regional sales representatives to ensure you get good pricing. different companies have different policies and you may need to go through procurement and deal with approved vendors.

If you are working on projects of high volume > 100,000 per annun you will have each and every sale rep from every semiconductor manufacturer knocking on your door weekly, trying to get their parts designed in early. In these situations the price you pay is the price your purchasing people can negotiate. It bares no resemblance to the low volume pricing you see on digikey/farnell/mouser etc.

So the difference in pricing you are seeing for similar parts is arbitrary. Having said that good op amps from Ti, Analog devices and Linear cost more than average op amps from other manufacturers. But you get what you pay for there ...

It comes down to how many parts your are buying ... the more units you buy the better negotiations you have on price, and the better support you get from the Field application Engineers (FAE).

Having worked in consultancies for small clients, large clients and worked in mid volume production quantities I have seen the range of support from the manufacturers vary greatly.

\$\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.