Design Con 2015
Home    Bloggers    Blogs    Article Archives    Messages    About Us   
Tw  |  Fb  |  In  |  Rss
Maxim Integrated - Integration Nation
Joshua Israelsohn

The Other Side of the Integration Coin

Joshua Israelsohn
Joshua Israelsohn
Joshua Israelsohn
3/29/2013 1:44:53 PM
User Rank
Blogger
Re: Is an op amp "discrete?"
Brad-

Yes, there are indeed three terminal ICs. Examples that come most readily to mind are three-terminal voltage references such as those exploiting the Brokaw Bandgap topology and three-terminal regulators.

But, beyond their abilities to invoke nostalgic sighs, those devices were quite distinct from other three terminal devices, such as discrete bipolar or MOS transistors, because of that barrier of abstraction thing I mentioned. That's a high-value attribute and, I'm thinking more and more, a compelling differentiating attribute.

50%
50%
Brad Albing
Brad Albing
3/28/2013 4:34:37 PM
User Rank
Blogger
Re: Is an op amp "discrete?"
Joshua - I agree w/ that assessment - we'll ID the discrete devices as the ones w/ 2 or 3 leads or terminals - tho' of course there are ICs w/ 2 or 3 terminals. So, simple passive components, diodes, and transistors.

50%
50%
Joshua Israelsohn
Joshua Israelsohn
2/18/2013 2:33:28 PM
User Rank
Blogger
Re: Is an op amp "discrete?"
In my book, neither op amps nor converters are discrete devices. Op amps may be small enough and cheap enough to sprinkle like pepper, but that doesn't change their nature.

Vendors design, verify, fabricate, and package these devices as they do more complex ICs. In that regard, they are subject to the same kinds of challenges and benefit from the same kinds of advancements as do more complex analog and mixed-signal blocks.

Beyond that, I think we make an important distinction between discrete components and integrated devices: Integrated devices provide barriers of abstraction that relieve designers from a great deal of technical detail compared to discrete designs. This changes the way we design with integrated functional blocks, be they as conceptually simple as an op amp or as complex as as a complete AFE.

100%
0%
TheMeasurementBlues
TheMeasurementBlues
2/18/2013 12:16:34 PM
User Rank
Teacher
Is an op amp "discrete?"
With all this integration on ICs going on, dos it change what we call a "discrete" component? I remember in college I had a required take-home exam in my senior year. You had to pass to graduate. The design problem was written "using discrete components." Even then, we students asked if op amps were considered discrete components. The answer was the "discrete" meant no op amps, aonly transistors and diodes.

Is that different today? Would you consider an op amp as an integrated device or is it now so basic as to be considered more or less a transistor? Could ADCs and DACs even be considered discrete components.?

BTW, the requirement for this take-home "competency exam" was dropped several years after I was required to take it. I think that was a wise decision on the part of the school. Some who went through it considered it a rite of passage and that the school was making it easier to graduate. But in truth, just because you can pass an exam and get up in front of three professors to defend your design dosn't mean you'll be a competent engineer.

100%
0%
Joshua Israelsohn
Joshua Israelsohn
2/15/2013 8:53:53 PM
User Rank
Blogger
Re: The Other Side as a game
Oh the life, indeed!

I don't know about the current curriculum, but when I was pursuing my engineering degree (@# years ago), game theory was not on the menu for EE students. Neither was project management or engineering risk assessment. Those "business of engineering" issues were left to one's first employer to teach. In retrospect, I think university engineering programs would benefit by including these issues in a coordinated set of lab courses if they don't already.

Fortunately, my summer interning job included what might be called Engineering Pragmatics 101 in the form of a generous and personable, but strict, supervisor.

100%
0%
Joshua Israelsohn
Joshua Israelsohn
2/15/2013 8:44:37 PM
User Rank
Blogger
Re: The Darker (other) side of the integration coin
Yes, this is an example of a broad range of complex analog and mixed-signal ICs for which the design team must have carefully considered issues of segmentation to appeal to as broad a range of customers as possible. I think medical-electronic AFEs are often prime examples because the required functionality for various devices tend to be both well defined and reasonably constrained. Another application example is communication PHYs (physical-layer ICs).

100%
0%
Joshua Israelsohn
Joshua Israelsohn
2/15/2013 6:39:36 PM
User Rank
Blogger
Re: The Darker (other) side of the integration coin
There are painful exceptions, but my experience has been that complex analog and mixed-signal devices offered as catalog parts typically come with a reasonable complement of documentation (though often distributed amongst several items such as data sheets, application notes, and the like). Remember it's in the manufacturer's best interest to ensure that the available documentation is sufficient to make users successful with minimum fuss and futzing because the alternatives--shipping an application engineer with every device or losing the customer to a competitor--are both far more expensive than generating sufficient documentation in the first place.

That said, devices that start out in life as CSICs (customer-specific integrated circuits) migrate to the catelog after an excusivity period elapses, and the vendor must re-evaluate the quality and completeness of its documentation in advance of releasing the part to the catalog. Suffice it to say, some vendors are more practiced in this regard than others.


Of course circuit and system designers have a responsibility to review the documentation vendors make available. I don't know what the track record is for this activity, but I note that RTFM (read the manual) is an acronym that dates back at least three decades  possibly twice that  and has probably survived for good reason.

100%
0%
eafpres1
eafpres1
2/3/2013 4:21:54 PM
User Rank
Blogger
The Other Side as a game
Hi Joshua--I liked your thoughts on the tradeoffs.  It occured to me you could write this down in standard game theory notation and solve it for equilibrium.  What I would expect is happening are a lot of non-equilibrium choices where things like lack of complete information, over or under-estimating risk of either solution, pressure from a boss or colleague to go with one solution, vendor concessions on price, etc. create anomalies.  Oh, the life of a design engineer.

100%
0%
eafpres1
eafpres1
2/3/2013 4:19:19 PM
User Rank
Blogger
Re: The Darker (other) side of the integration coin
@WKetel--There are off the shelf AFEs which if used might be considered a component, and if you then look at the total risk as a probability function, might lower the expected value of a stock-out.  These would be previously characterized, already in production parts, not new custom designed parts.  An example (I picked randomly) might be this TI front end:

TI AFE

100%
0%
WKetel
WKetel
1/31/2013 9:57:29 AM
User Rank
Artist
The Darker (other) side of the integration coin
One trhing that some of those organizations producing the very application specific chips may not have considered is the possibility that others might possibly use the device if an adequate amount of information about it was available. Of course there may be some sort of proprietary agreement in place about the device that would prevent marketing that device to others.

As the engineer responsible for designing a variety of systems for much smaller companies over the years, I can point out that on the opposite side, there are some very big advantages to using only those chips listed in the "commodity" catagory. The very first, and very big, benefit is part availability. When some manufacturer has a production quality or yield problem the alternate source is a lifesaver, even if the price is not quite as good. That saved me once on the '4038 one-shot, as one producers process somehow set the trigger threshold to a different piont. An alternate supplier's part still functioned as intended. So while it may be beneficial to put all of the system functions in one single sourced custom chip, it may be much wiser to use a set of the more common devices. And one other benefit, which is only to the purchaser of the product, is that it becomes possible to repair that product, since replacement parts are actually available. I have never been able to purchase a replacement for any of those custom chips, which has been a source of irritation on several occasions.

100%
0%
More Blogs from Joshua Israelsohn
The integration of optical and electrical devices on silicon, and the ability to convert from electrical to optical and back, are paramount in the design of high-speed communications channels.
New power modules raise the question of what exactly is integrated. How much functionality must be integrated before the device becomes a system?
Nearly a half century later, and in the industry's eagerness to cling to his prediction, we seem to have forgotten about Moore's insights.
Why is it that, whenever someone invokes Mooreís Law, I start getting angry?
flash poll
educational resources
sponsored content by Maxim Integrated
follow Planet Analog on Twitter
Planet Analog Twitter Feed
like us on facebook
Planet Analog
About Us     Contact Us     Help     Register     Twitter     Facebook     RSS