>Paul Hovnanian P.E.
>On further contemplation, I can see it as returning to your models to
>see where they don't match the physical realization... I do it all the
>time... refining models and macros.
>
>But calling it "regression" makes me ill...
You appear to be ignoring repeated descriptions of what regression testing is. Is this because you disagree with the definition given?
Regression testing tests functionality that previously worked as desired no longer working the same way as an unintended consequence changes in other parts of a design. Regression testing involves re-running all previously run tests as a batch and checking whether previously fixed faults have re-emerged. It is common in software development and, it appears, in Analog IC design.
I am having a hard time figuring out why calling it 'regression' makes you ill. You are testing for regression (a feature that stops working after a certain event), so what's the problem with calling it regression testing?
I'm of the (very) old school ;-) "Regression" is fitting an equation to a data curve. It appears that the new definition is "go back and tweak repeatedly until it works", then claim "engineered all the way" ;-)
...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems | manus |
| Phoenix, Arizona Voice:(480)460-2350 | |
| E-mail Address at Website Fax:(480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |
America: Land of the Free, Because of the Brave
I am aware of that. The above is but one definition of the word "Regression", and not even the first definition in most dictionaries. Does it sicken you when other English words have multiple definitions, or just this one? Does the word "Linear" as used in art (a style of oil painting) or botony (a type of leaf) or electronics (a circuit or device having an output that is proportional to the input) also make you ill because the same word is used in the phrase "Linear Regression?"
As for whether you are being old school, the etymology of the word "Regress" goes back to the Middle English / Anglo-French "regresse" and then back to the Latin "regressus / regredi." The usage "to go back" predates your preferred definition, which, if the references at the end of the Wikipedia article
formatting link
are to be trusted, dates back to 1805-1809.
Again, I am curious as to why you insist on replacing the dictionary definition with the above -- which you appear to have manufactured out of whole cloth. You didn't get the above definition from the job description that started this thread, and you appear to be actively avoiding standard English word definitions. Why?
Here are some dictionary definitions. It would help if you at least acknowledged that you have read them, even if you totally disagree and wish to redefine the word.
-----------------------------------------
From Merriam-Webster:
Regression
1: the act or an instance of regressing
2: a trend or shift toward a lower or less perfect state: as
a: progressive decline of a manifestation of disease
b (1): gradual loss of differentiation and function by a body part especially as a physiological change accompanying aging (2): gradual loss of memories and acquired skills
c: reversion to an earlier mental or behavioral level
d: a functional relationship between two or more correlated variables that is often empirically determined from data and is used especially to predict values of one variable when given values of the others ; specifically : a function that yields the mean value of a random variable under the condition that one or more independent variables have specified values
3: retrograde motion
Derived terms:
regression testing regression therapy (psychotherapy) linear regression nonlinear regression quantile regression regression to the mean
-----------------------------------------
From Dictionary.com:
regression -noun
the act of going back to a previous place or state; return or reversion.
retrogradation; retrogression.
Biology. reversion to an earlier or less advanced state or form or to a common or general type.
Psychoanalysis. the reversion to a chronologically earlier or less adapted pattern of behavior and feeling.
a subsidence of a disease or its manifestations: a regression of symptoms.
regression -adjective
of, pertaining to, or determined by regression analysis: regression curve; regression equation
The "job description" smacked of some hocus-pocus EDA for "designing".
...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems | manus |
| Phoenix, Arizona Voice:(480)460-2350 | |
| E-mail Address at Website Fax:(480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |
America: Land of the Free, Because of the Brave
It's EDA BS implying you can design without knowing your ass from a hole-in-the-ground.
If all you ever do is paste already-developed cells together... maybe.
...Jim Thompson
-- | James E.Thompson, P.E. | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona Voice:(480)460-2350 | | | E-mail Address at Website Fax:(480)460-2142 | Brass Rat | |
formatting link
| 1962 | America: Land of the Free, Because of the Brave
I can't speak to "smacked of", that being purely subjective, but I once again note the multiple technical papers about regression testing of large analog IC designs that come up in a web search. Are they all hocus-pocus as well?
The above looks like typical marketing fluff designed to sell services to upper management -- tte ones who might actually hire them -- and not geared towards working engineers such as yourself -- who would never hire them no matter what they wrote. That says nothing about whether the underlying methodology is sound.
Again, analog IC design is way out of my range of experience and I have no way to evaluate the actual methodologies they are touting, but I looked at the entire site and could not find anything "implying you can design without knowing your ass from a hole-in-the-ground", and such an interpretation seems to require redefining the phrase "Regression Testing." Can you tell me which section gave you that impression?
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems | manus |
| Phoenix, Arizona Voice:(480)460-2350 | |
| E-mail Address at Website Fax:(480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |
America: Land of the Free, Because of the Brave
I have an associate digital guru who does all my control logic and SPI or I2C buses.
I know where my skills lay, and their limitations. I use other contractors extensively.
...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems | manus |
| Phoenix, Arizona Voice:(480)460-2350 | |
| E-mail Address at Website Fax:(480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |
America: Land of the Free, Because of the Brave
Read page 20, "Serial Interface." This is the typical analog-datasheet masterpiece of obsfucation. There's no particular distinction between edge-operated and level-sensitive things. There's obviously some clock gating going on inside - a mortal sin in the digital world - but they don't say what.
They thoughtfully provide a command (that must be clocked in using falling edges) which changes the chip to using rising edges. And the "RESERVED" commands should be relabeled "FOAD", because some of them seem to kill the chip.
We have this on an SPI bus with an equally bizarre BB/TI serial DAC, which has its own obscure, badly documented, very different protocol, using the opposite clock edges of course. We spun the driver FPGA four times this week before we got both dacs to load reliably.
At least chip select is active low on both. Some people don't even do that.
Analog chip designers seem to hack the digital interfaces horribly. And randomly.
Analog-wise, once you get it loaded, the 5432 is pretty good, although the tc is terrible at low mdac codes; I characterized that and fixed it in firmware.
Same here. I don't do layouts and I don't do software, except in a pinch or where it's really RF-critical. The good thing is that my layouter quickly understands which areas are "hot".
Interesting that you mentioned this. I found out the hard way that a converter from them didn't like the falling edges on SPI. They worked nicely in rising edges. During a lengthy discussion with an AD engineer he finally said "Well, then just use the rising edge." But their converters sure have a great performance in the analog domain. Unless people split grounds ;-)
Do you wish an introduction to my guru? He's a product of the excellent graduate courses in ..HDL at Ohio State.
...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems | manus |
| Phoenix, Arizona Voice:(480)460-2350 | |
| E-mail Address at Website Fax:(480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |
America: Land of the Free, Because of the Brave
I use a cute little TI 12-bit SAR ADC, ADS7866. $2.65 at 100. When you assert CS, it freezes the s/h, and as you clock it it outputs serial data and does the successive-approximation thing. Clever. But the digital interface is barely described. It works fine as long as you clock it one more time than what they seem to suggest.
I suppose the safe way to load a serial dac is....
Most of them work that kind of conversion scheme. But yeah, WRT the digital part AD's datasheets can become iffy.
That's the belt and suspender method. Mostly ok because they often allow up to 20MHz speed. Some can even do 30MHz but the SPI usually has to be throttled to around 3-4MHz because stuff like many of the EEPROMs can't stomach anything over 5MHz.
OTOH what's a little datasheet bug in the SPI part compared to the problems people experience with I2C. Such as devices waking up with a hang-over and hogging the bus until the uC kicks it a dozen times. For some chip designers a proper BOR design appears to be beyond comprehension. I never understood why.
Eeew! That *is* nasty! If you have engineers doing that, you need more help than some testing methodology can give you. You need replacements or possibly remedial instruction...
We did an actuator test set once where we froze the clock to the (all-CMOS) digital section, did an ADC conversion, then let the logic do some more work. Even then we *still* separated analog and logic.
The last time you said you didn't like something I posted, I gave you a full apology and stopped doing it. You killfiled me anyway, leaving me at a loss as to what it takes to satisfy you. It will be interesting to see if history repeats itself.
As happened last time, I am choosing to respond to your stated dislike of what I post by giving you what you appear to want, and will not ask any further questions of you or post any further responses to this thread or topic.
If you require an apology, I am willing to give you that as well, but doing so would require some assistance to help me to understand why you dislike the above question so I know what to apologize for.
My only dislike is the EDA jerks implying design by regression.
...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems | manus |
| Phoenix, Arizona Voice:(480)460-2350 | |
| E-mail Address at Website Fax:(480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |
America: Land of the Free, Because of the Brave
Since the thread has (as usual) lost its sense of direction, I again quote the first line...
"... CAD experience in automating analog IC design"
Note the "automating"!
...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems | manus |
| Phoenix, Arizona Voice:(480)460-2350 | |
| E-mail Address at Website Fax:(480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |
America: Land of the Free, Because of the Brave
I often have logical controls within the analog. But I leave the bus controls to an expert.
I always ask :-)
...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems | manus |
| Phoenix, Arizona Voice:(480)460-2350 | |
| E-mail Address at Website Fax:(480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |
Only Hillary Clinton can spin a long-winded lie,
repeat it regularly, then declare, "I misspoke".
Pinocchio lives on ;-)
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.