« February 2004 | Main | April 2004 »
March 29, 2004
[Misc.] this is a second test.
Posted by jones at 01:30 PM | Comments (0)
March 18, 2004
[Misc.] this is a test
so don't read any farther. Great. Now it didn't work.
Posted by jones at 04:19 PM | Comments (0)
March 17, 2004
[Misc.] Glossary for visit with Pres Samuelson
[Merriam-Webster Online: milieu]
milieu: the setting in which something occurs.
tail end of the Starling curve: not sure, but see this reference. Probably a reference to the knee of the curve.
Malthusian economics: found some refs, but not sure what it means in the context.
Posted by jones at 08:07 PM | Comments (0)
March 16, 2004
[Misc.] Zion National Park with twins
There's a dearth of information on the net for traveling with twins. So we'll do our small part by posting trip reports for traveling with twins. The first one is a roadie to Zion NP (which is about 4 hours' drive from our house). The twins were 9 months old on this trip.
Zion has tons of wheelchair accesible trails, making it an ideal destination for the stroller-bound. Despite being in a canyon, there's plenty of sun. We wished we'd had more than just sunscreen and hats for the girls. Shades for the stroller were essential (but we didn't have them with us). We walked the river side trail. The emerald pools trail would have been another good choice and may have had more shade. We wish we'd have used two umbrella strollers rather than our front/back stroller. The umbrella strollers might have helped us avoid some of the inevitable "oh they must be twins!" discussions. Which are fun for the first two or three times, but get old.
The girls seemed to enjoy playing in the grass by the visitor's center the best. In addition to many wheelchair accessible trails, the park has big grassy lawns near the visitor's center and in front of the park lodge. The men's restroom in the vistor's center has changing tables.
We stayed in the Town and Country Best Western in Cedar City. That worked out well. The hotel was extremely accomodating and the twins had a fun time in the heated indoor pool.
Overall, Zion NP was an excellent southern Utah destination for our family with twin babies.
Posted by jones at 02:34 PM | Comments (0)
March 15, 2004
[patents] Disposable underpants having discrete outer seals
[United States Patent: 4,610,681] One of the 12 patents on the "Little Swimmers" disposable swimpants we used for the girls during our recent vacation. A highpoint of this patent is the Background Art which includes a description of how to change a disposable diaper
"Disposable diapers of the type presently on the market are flat open-sided garments that are intended to be fit about an infant by a parent while the infant is lying down. The rear panel of the diaper is placed underneath the infant, and the front panel drawn between the infant's legs, after which the sides are overlapped and held together by pressure-sensitive adhesive tape. A diaper is meant for use when the child is young and dependent upon a parent for this essential purpose."
Posted by jones at 08:54 PM | Comments (0)
March 11, 2004
[Research] 6811 real time interrupts
I think those numbers on the left are the addresses of the various registers.
But, I was wrong. Tehre is a real time interrupt function. See section 10.2.2 on page 10-11. Period between interrupts is measured from interrupt to interrupt--not from clearing rti flag to next interrupt. You can set the RTI clock frequency using RTR1 and RTR0. Have to set the RTII bit to one to get hardware interrupts on RTI time outs. The question is how to register an ISR for the RTI when the RTII bit is set to one?
the TCNT value lives in $100E through $100F.
Here's some handy code in C that sets up an ISR for RTI on the 68hc11. I want that, but in assembly.
Posted by jones at 09:13 PM | Comments (0)
[Research] Reading from a port in the 6811
[Very simple example] Figure out how to read from a port and how long its going to take.
The address for Port A is $1000, Port B is $1004 and the address for Port C is $1003. There's other stuff around that address too.
Posted by jones at 10:01 AM | Comments (0)
March 10, 2004
[Research] Persnickety details about 68hc11 timing
[Dickens 68HC11 - Documentations and References] we want to know, down to the cycle, what happens in the 6811 using the free running counter to provide periodic interrupts. We want to konw this so we can validate our model of time using a real processor. Turns out the answer is "5". the TCNT value lives in $100E through $100F.
"Another use of the output-compare function is to generate a specific delay. For example, to produce a 10-ms delay to time programming of an EEPROM byte, follow the initial programming steps to the point where the programming supply has been enabled (EEPGM bit has been written to one). Read the current value of the main timer counter and add a number corresponding to 10 ms (if the count rate is 2 MHz, the value corresponding to 10 ms would be 20,00010 or $4E20). Write this sum to the outputcompare register so that an interrupt will occur when the counter gets to this value. In this example, the actual EEPROM programming time started just before the current time was read from the counter and ended after responding to the output compare and turning off EEPGM. The small delays for setting up the output compare and the latency for responding to the output compare are not considered because they only make the EEPROM programming time longer by a few microseconds. A more advanced user of output-compare functions will learn how to correct for these details, although it is often not necessary. Details about the implementation of the output-compare functions and specific examples of how to use output-compare functions in the MC68HC11A8 are included in 10.4 Output-Compare Functions."
the value of the free running counter is stored in a register called the "TCNT" register. There's no way to interrupt it and it sets an overflow bit when it overflows. Its got 16 bits.
"The count value that is returned on a double-byte read corresponds to the value of the free-running counter at the second-to-last cycle of the double-byte read instruction. This and other subtle timing details related to the main timer are discussed in 10.5 Timing Details For The Main Timer System."
[From section 10.5, Timing details]
This is the timing diagram for a reset. Not sure if its reset from interrupt or what, or just a reset of the continiuous timer reset call.
The timing diagram for a "read the timer value as a double" instruction:
Later, we learn that if you do a read right after a reset, then you get like a 5. To wit, from page 10-41, "The information in Figure 10-13 and Figure 10-14 can be combined to find the value that would be returned if the LDD instruction was performed as the first instruction after reset — that is, if the reset vector pointed directly to the LDD instruction. The value read will always be $0005."
In the examples, most of their interrupt timers are accurate to between 27 and 70 cycles and are limited to 2^16 becaues they don't deal with counter overflows. That means they deal with about 8.38 seconds.
This example code may be helpful in coming with a very precise periodic interrupt.
Posted by jones at 03:50 PM | Comments (0)
March 09, 2004
[Paper reviews] "Using multiple levels of abstraction in Emb. Software Design" by Burch et al
[Using Multiple Levels of Abstractions in Embedded Software Design] The most concrete model is a continuous time, continuous values extentional model with interleaved semantics. Ours is discrete time, discrete values, an intentional model also with interleaved semantics. The intentional model with discrete time and values is closer to the behavior of embedded devices modeled at the machine code level and more amenable to automated analysis.
Its mostly about a framework for comparing and contrasting different formalisms. But, there's a set of three trace algebras for embedded software. Of those three, the "Metric time" model is most similar and interesting relative to our timed semantics for machine code.
The metric model:
- has a continuous model of time in which time is modeled on the reals.
- allows real and integer valued variables.
- A partial trace is a sequence of triples gamma, delta and f. delta is the duration. gamma is the signature and f is a function, with domain [0,delta], that: gives values as a function of time, and maps input or output events to times.
- the mapping of a variable to a value wrt time must be piece-wise continuous and right-hand limits must exist at all points.
- f(a)(t) == 1 means that event a occured at time t.
A complete trace is a pair, gamma and f, where gamma is the signature and f is a function. The function f is defined as before except f is defined over the non-negative real numbers rather than the range [0,delta] .
Then you can do projection, renaming and concatentation.
So, relative to our model, our model is discrete time and discrete valued on the inside but allows time varying functions of real time over real values in the environment. Our model starts with a specification language (I suppose that's what we'd call it) then gives the interpretation of a system. That's a little different because their algebra gives a formalism for manipluating traces that might result from a system such as ours. So, we can specify signatures then their formalism could represent the behaviors of our system and be used to do projection, renaming and concatenation.
Seems that their formalism is more powerful than ours. They have real time and real valued variables. We don't. I think we could express all of our models in their trace algebra. We can't write models that generate all ofthe traces they can model.
Open questions: what verification algorithms and properties can be defined on the traces?
In Smolka's classification scheme this would be an interleaved extensional model and ours is an interleaving intensional model. Smolka claims, and I agree (for what its worth) that "Extensional models provide a useful basis for explaining system behavior, while intensional one are more amenable to automated analysis." And so it would make sense that we are going extensional.
Posted by jones at 07:49 AM | Comments (0)
March 07, 2004
[patents] method for perspective transform in computer using multi-pass algorithm
[United States Patent: 5,808,623] Yet another Photoshop Elements patent. The splash screen lists a ton of patents, I'll look a new one up each time I start up elements. I didn't think perspective transform was a big deal, but I guess this algorithm, er, I mean method, is.
Posted by jones at 08:32 PM | Comments (0)
March 05, 2004
[patents] displaying and printing multitone images derived from grayscale images
[United States Patent: 5,146,346] Also an algorithm, er, I mean method, used in Photoshop elements. The description is a fine piece of technical writing. I could follow it just fine. It used "her" instead of "him" in the narratives.
Posted by jones at 10:06 PM | Comments (0)
[patents] High speed data compression and decompression apparatus ...
[United States Patent: 4,558,302] My version of Photoshop elements uses the LZW algorithm, er, I mean "apparatus" for compression. Probably in jpeg encodings.
Posted by jones at 10:01 PM | Comments (0)
[Misc.] Last post on vaccines and autism
[mercola.com: Scientists Retract Vaccine-Autism Link?!] Dr. Mercola's reaction was to point out that "the system" has become amazingly corrupt and "convince[d]" the researchers to withdraw their claim. Interesting. I will now resume my normally scheduled life. You do have to give it up for Dr. Mercola for using a blog.
Posted by jones at 07:36 PM | Comments (0)
[Research] motorola 68hc11 docs link
Dickens 68HC11 - Documentations and References
Posted by jones at 04:16 PM | Comments (0)
March 03, 2004
[Misc.] Vaccinations, seizures and autism
[More Research Supports MMR Link to Autism 7/10/02] I wonder Doctor Mercola will revise his position on the research of the "brilliant young British gastroenterologist" Dr. William Wakefield now that we know that he had an undisclosed conflict of interest and 10 of his 12 co-authors (1 couldn't be reached to sign the retraction) have withdrawn key parts of their Lancet article.
Posted by jones at 09:12 PM | Comments (0)