SMALL - espritmodel.com SMALL - Telemetry SMALL - Radio
Jack Crossfire's blog
Posted by Jack Crossfire | May 28, 2014 @ 12:23 AM | 2,207 Views



Doubled the amount of aluminum between the LM317's & the board. Ran 3 laps around the complex, giving a rough estimate of the speed. 7.5V was now well above what the human can sustain.

8m4s
7m48s
7m49s

At 7.2V, it was

8m53s
8m20s
8m11s

Tested 7.3V last. It was

8m52s
9m11s
8m57s

It was definitely inconsistent. GPS seemed accurate enough. The Lipo may be at the end of its life.

The I feedback made a small improvement in the steering. It still occasionally veered off course, but it managed to correct after a certain time. It's probably the best compromise between going crazy & not steering at all.

Finally moved it to C. Random failures abounded. Computing power went way down, manely owing to the required stack size. SDCC ended up not initializing the stack pointer. The stack pointer was 1 of the few registers the PIC didn't initialize. Manually put it at a reasonable address. There were probably also stack overflows. Managed to get the variables as compressed as possible, overwriting the gyro initialization variables. It didn't have any more issues. It ran fine at 3.3V, despite being rated for 5V.

SDNM doesn't work, making for a lot of guessing for the stack size. The free, 10 year old, through hole 18f1320's have probably made their last project. There are 2 more 5V ones, 1 more 3.3V, as probably dedicated spares. The 18f1220's are all gone. There are some exotic ones: 3 x 18F4685, 3 x 18f6585, 3 x dsPIC33fj128mc802, a lot of 18f458's intended for crash damage.
Posted by Jack Crossfire | May 27, 2014 @ 03:47 AM | 1,664 Views





Another day banging on the GPS analysis software yielded an easier way to get interval training times, manely basing pace on a fixed distance into the past instead of fixed time. The calculation was more complex. The result was no more accurate than fixing time, but easier to read.

http://heroinewarrior.com/workout.html?interval5.kml

The interval training app turned out to not work unless GPS was in keep on mode. If it's closed & reopened, it can't restart the alarm. What would be nice is if Android had a provision for detecting if an alarm is already set, so the alarm can be set wherever there's a chance it may need it.
Posted by Jack Crossfire | May 23, 2014 @ 02:34 AM | 2,740 Views
Since Google's interviewing formula became accepted as the proven way to hire everyone, algorithm tests on codility.com have all but replaced the interview. The 1st codility test was 15 minutes for 2 problems & a total failure. Assumed it would be equivalent to programming tests in the past, but it was drastically harder, with a much clunkier interface. Thus ended any chance of ever working at Airware.

They might have actually configured it wrong, because the guy said it would be 70 minutes. The 2nd codility test felt better, but also had 60 minutes for 2 problems. There is never as much time as a real job. The trick to winning in Google's world is more about mastering the algorithm tests than being good at the job.

For codility, you have to read the instructions in the sample tests, take the sample tests or at least formulate how you would solve them, then read as many blog posts about the solutions as possible. They require passing the complexity requirement, passing all corner cases, handling the very large numbers in their entire stated range, very small numbers, very large data sets, & very small data sets. Their example data sets don't test the corner cases. Cleanliness of code doesn't matter & there isn't enough time to clean it up, anyway. Sometimes a clue is given in the problem, like the range of a variable being the size of an array, thus you should make a lookup table, or order sqrt N complexity, thus a for loop of sqrt N...Continue Reading
Posted by Jack Crossfire | May 22, 2014 @ 12:14 AM | 2,522 Views




In another 10 miles of driving, the voltage regulator rework definitely reduced the amount of power. The fastest mile today was 8m2s. Things dragged on at 8m30s per mile for most of it. The recommended pace for a tempo run is 8m22s. Staying close to the car made it easier. Had nothing but salad before it. The fiber seemed to prevent crashing. Tried the most aggressive sugar content in the recovery drink.


Steering with AUTO_STEERING_PERIOD reduced all the way to STEERING_PERIOD / 4 was very ineffective. After 86.2 miles, the steering mechanism has deteriorated to the point of requiring I feedback.


For a good time, the GPS log can now be analyzed online.


http://heroinewarrior.com/javascript.html?tempo2.kml
Posted by Jack Crossfire | May 20, 2014 @ 05:20 AM | 2,020 Views
http://www.forbes.com/sites/georgean...nifty-job-ads/

There is going to be a quad copter/drone delivery service, but not what you think. It would be impossible for a quad to fly from the distribution center to a house, with current technology. What is going to happen is a truck dispatching a quad the last 50 ft to deliver to a front door or some kind of target the customer prints out on an 8.5x11" sheet of paper & sticks to the door.

Current technology is good enough to just barely make it more efficient for a very highly maneuverable flying robot to do it than to have the driver get out of the truck. It would require another robot in the truck to load the copter. The navigation sensor of choice would be Vicon on the truck guiding it to a landing site within visibility of the street.

It would require another layer of route planning beyond the current street algorithm. It would need a database of every floorplan or lidar on the truck to do final route planning to the door. As unlikely as it is, the same thing happened to garbage men riding on the backs of trucks. That became automated, decades ago.

The ultimate plan of flying from the distribution center would need something much more reliable & bigger. A current, gas powered copter could probably do it at great expense & with a few crashes. It would fly above the destination & dispatch a smaller copter the last 50 ft. The problem is going to be solved, but it's hard enough to ensure a few years of job security.
Posted by Jack Crossfire | May 17, 2014 @ 04:55 PM | 17,193 Views
Netscape Navigator 7.2 (5 min 17 sec)


Things were so different, in those days. The Netscape server room had about as much computing power as a modern phone & far less bandwidth. The PC was king. Jobs were for life.

The big ass PC application was king. Thousands of menus had to be literally driven by the user. Everything a developer wrote was visible. Nowadays, it's all hidden & automated. There are maybe 2 menus.

The Facebook app has virtually no user interface, but 54 services in the background. Each service has probably 50 developers. All of its functionality is automated. The Google Play app has 22 services, but virtually no user interface. All the stuff that would have taken a user going to a menu in 1999 now only becomes apparent in the rare case a service can't figure out what to do & pops up a notification.

It doesn't feel very individualized, in a world where what everyone writes & thinks is now determined by an algorithm.
Posted by Jack Crossfire | May 16, 2014 @ 03:30 AM | 2,191 Views



When used carefully, it doesn't show annoying animations, popups or make you wait for additional bits to load, when you scroll down the page.

Accessing it through jquery has been required for any kind of day job, since 2011. jQuery adds around 300kb to every page load & is another layer of indirection, explaining why web apps are so slow. It theoretically reduces the amount of porting between browsers, but trying the various jquery demos showed similar browser incompatibilities to pure javascript.

Javascript behaves so differently on each browser, it's not really portable. There is a subset of math functions which seems to be portable. The DOM trees are vastly different. Chrome seems to have the most bits working.

It's very slow for analyzing just 32000 data points & the user interface is always going to be clunky, no matter how much tweaking you put in. It's a practical way to analyze small sets of data, when polish doesn't matter.


After much refinement, a GPS analysis tool emerges. It does everything you need, but no-one else ever implemented. It mostly works on any platform, with no need for installation, libraries, compilation, or waiting.

Then, the reality of why we have a cloud & all the investment money in cloud startups emerges. 1st, it has to be called cloud instead of web to avoid another .com collapse. 2nd, getting the data from a phone to the web app is a real pain. The PC...Continue Reading
Posted by Jack Crossfire | May 14, 2014 @ 10:52 PM | 3,021 Views
The RC car empty & 7.5V did 8 minutes on the fastest downhill & 8m24 on the uphill. It had increasing numbers of steering glitches. Didn't completely discharge the battery or survive the full 10 miles, this time. These RC car runs are intended to go until the human drops, rather than battery depletion. The temperature was 80F at starting time & 77F at finishing time, with a strong north wind. The Danville run was 50F with no wind. This route had fewer stops than the Danville route.

That brought it to 76.2 miles & its 1st broken TO220.



The rebuild was the crustiest circuit ever. Dave Jones would have a fit. All that weight to avoid paying up. Not sure how we ended up in a world where the only BEC with fully adjustable voltage is from Castle Creations & requires a proprietary protocol which no-one ever bothered to reverse engineer.

Fully built buck converters with the required parameters are rare things. It might be a very unique application to want constant output power, variable speed, making the car slow down on hills.
Posted by Jack Crossfire | May 12, 2014 @ 08:46 PM | 2,025 Views
So the US of F offered admission in Fall. Depending on how the GPA is calculated, they would award an MSEE in 3 years either barely or easily. Mathematically, it would happen.

The alternative is Cal State Bakersfield, which would award a BS in computer engineering in 2 years. It would be cheaper, but require another 2 years somewhere else to earn an MS. CSUB hasn't produced any admissions decision yet & traditionally given decisions only after it was too late, but you never know. The difference between waiting until August for a decision or moving back to USF now is a matter of a certain amount of money.

The US of F is not Stanford, Georgia Tech or even the U of F & it's emotionally harder in the short term. The internet streams endless, random, contradicting noise on these matters. There is a big difference between what the top 1% puts out & what the bottom 99% puts out.

Generally, the more broke & miserable you are, the more advice you give. In studying many businesses formed after 1973, it would seem the top 1% never follow the popular advice & they're all products of rich dads. 1973 was the cutoff for a worker to accumulate wealth or be utterly dependent on a rich dad or extremely lucky lottery tickets.

A wealthy man once said "Listen to everyone. Follow no-one." That sounded like the best advice. Lacking a rich dad, accumulating any kind of wealth is all but impossible, but you are definitely going to be working for a long long long long time, so you better make sure the time is spent doing what you want to do.
Posted by Jack Crossfire | May 11, 2014 @ 01:46 AM | 2,637 Views




The search for a better throttle stick led back to the picco-z encoder. It definitely had enough reinforcement to do the job.


...Continue Reading
Posted by Jack Crossfire | May 10, 2014 @ 01:10 AM | 2,382 Views
http://books.google.com/books?id=byg...20Mohr&f=false

There was an interesting story about a solar powered refrigerator that used direct solar heating instead of a photovoltaic. Daytime heating & night time cooling powered 1 cycle of a vapor absorption cycle every day. The cycle was at 1 pressure. The vapor absorption cycle was the dominant method, in those days. It was far less efficient than a photovoltaic turning a compressor, but much simpler.


That was actually a pretty depressing magazine, not in how obsolete the technology was, but how they were trying to solve the same problems that we're still trying to solve today & how their means during the great depression were not unlike modern times. There were miracle tools which promised to revolutionize manufacturing & put every man in business for himself.

The dominant DIY project involved resourceful use of very low cost junk, without any money or tools. Despite so many decades of new tools & the economy supposedly being a step above the great depression, most of today's DIY projects are still made by resourceful application of the cheapest junk from China & no money or tools.

Gender roles were the same as today. Gadgets, inventing, science, & tools were all aimed at the men. Cooking, cleaning, & home making was all aimed at the women. They were a lot more transparent about it, but comparing gizmodo.com to jezebel.com reveals the same separation behind the...Continue Reading
Posted by Jack Crossfire | May 09, 2014 @ 08:44 PM | 2,618 Views

During another 5.5 miles, tried moving the camera farther back to get the front tires to wear less. It made steering unusable & didn't affect speed. Either have to recalibrate the steering or keep the weight centered. A wood piece as a landmark for the steering buttons was worthless. It was impossible to differentiate the wood from a button. That brought the car up to 66.5 miles.

An RC car exercise coach could probably make money. It would have to also collect location data, uploading it to a cloud service for sale to real estate developers. That's the big thing. Exercise tracking startups make a living by selling all the data.


Merely 20 years ago, junk mail factories spent a lot of money for databases of addresses. Once actually spent a lot of time trying to extract a database of addresses from a CD-R that was used for fixing & translating addresses to barcodes. It was a way to learn C, but it could also have made a lot of money, if it worked. Now, it's the same game, but with databases of absolute locations created by the population itself instead of a database of addresses created by the post office.


Have never put anything expensive on the car without a lot of shielding. It has to withstand a bike or car impact. A phone would be very expensive.
Posted by Jack Crossfire | May 08, 2014 @ 01:51 AM | 2,598 Views

So the hand controller disintegrated. There was a too much horizontal stress on the throttle controller for it to not have the reinforcing enclosure it used to have.


...Continue Reading
Posted by Jack Crossfire | May 07, 2014 @ 06:02 AM | 2,872 Views
So the blocking item in the lightsaber drone game was a lightsaber that could sense a laser impact.





The dollar store had exactly 2 of the cheapest tubes of plastic.

...Continue Reading
Posted by Jack Crossfire | May 05, 2014 @ 11:19 PM | 2,655 Views




They remind me of Red Hat & VA Research, long long ago, in a galaxy far far away. They're revered corporations, by the community. Their employees think they've all hit paydirt & they're always going to be there. They're not officially rivals, but when the sh*t hit the fan with good buddies Red Hat & VA Research, heads still rolled.

Unlike Red Hat & VA Research, I don't think any of the rank & file employees have any stock options. They definitely don't have the benefits we had. The motto for generation Y is your reward is doing something influential that benefits the community, not making money. It was very different in the old days, when we were there to get rich. Our generation was still heavily influenced by the baby boomers.

Both CEOs work their asses off. They look like it in their videos. Nate rarely goes on camera, but Limor goes on several times a week, for hours after already putting in full days. She must be sick of widget boards, but still manages to muster some enthusiasm while rattling off some teardowns & emails. Her new England accent still sounds a lot like, "Ghostbusters whaddya want!" by then.

It's definitely true that in toys & widget boards, you can't make money in design & fabrication. All the money is in the retail part. That might not be true for the highest end hobby grade solutions, because Horizon Hobby still has some money coming in...Continue Reading
Posted by Jack Crossfire | May 04, 2014 @ 11:56 PM | 2,520 Views
http://www.cnn.com/2014/05/04/us/cal...ir-show-crash/

Debated whether or not getting up early, driving 50 miles, waiting in line to see a bald guy in a C-5 & seeing a static display of an F-22 was worth it. Considered just making videos of the propeller flights. Washed the car, then decided there was too much to do down here. It would have ended up a short stay, but it would have produced video of the crash. Between a crash at SFO, a stowaway in SJC, a crash over San Pablo bay, & this, there have been a lot of incidents in this area lately. They could all be symptoms of a lack of money in 1 particular area.


Now some large diaphragm microphone testing.

Coding Sucks: Why a Job in Programming Is Absolute Hell (13 min 33 sec)




So the decision was made to try to make more progress on the human vs. drone concept that has been proposed since 2009. The idea has revolved around an autonomous flying opponent fighting a human. The problem has been finding a game which doesn't destroy parts. The ages old idea of a drone vs. lightsaber from Star Wars was never practical, because what would the drone shoot that a human could intercept fast enough?


In order to make it playable in real life, the drone clearly needs to shoot a sustained laser beam & measure the time taken for the human to intercept it with a lightsaber. Then it needs to move around & point the laser somewhere else. The trick all games use is to make the pattern...Continue Reading
Posted by Jack Crossfire | May 04, 2014 @ 12:58 AM | 2,770 Views

Goo Tube recommended some electroluminescent paint videos. The idea of entire everyday objects glowing was intriguing. It was like an episode of Ducktales where Gyro Gearloose tried to make the lightbulb obsolete by inventing a paint that made everything else glow. It looked too good to be true because it probably was.

A company called Lumilor made a lot of noise in 2013, claiming to have invented a sprayable EL paint, but has never sold it or shown it used in large quantities. The largest applications have been motorcycle tanks. There are no videos of it being applied, only videos of someone spraying something on top of an existing phosphorous coat, then blowdrying it to life.

Lumilor is probably rebranded Dupont Luxprint. The only way to get a part coated is to send a part to Lumilor for them to paint. That's your biggest clue that it's probably a rebranding of something old.

Dupont Luxprint appeared in 2010 & hobbyists began using it to paint small EL panels. Once saw Jeri Ellsworth trying to use it & wondered what the point was. After seeing the world change in the last 4 years, the ability to have entire buildings glow could be 1 of the things which is going to change the way the world looks.

Most likely, the phosphor has to be brushed on. The part they're spraying in the videos is a translucent conductor. That has to be coated again in a clear coat to protect it. The order of layers in a glowing motorcycle tank would be

1) insulation from bike frame
2) silver paint conductor
3) multiple dielectric layers
4) phosphor
5) translucent front conductor
6) clear coat

The translucent conductor needs to have wires every 5cm. The phosphor is extremely expensive & not sold in small quantities to individuals.
Posted by Jack Crossfire | May 03, 2014 @ 01:07 AM | 2,129 Views
After 6 months of waiting, it finally opened.


Many an asian kid looked inside, in the months leading up to it, but in the end, it wasn't what you think.


It was nothing but another home decoration store.

...Continue Reading
Posted by Jack Crossfire | May 02, 2014 @ 03:02 AM | 3,012 Views


The Google Ara is going to be a big deal. They have a good record of actually delivering the development kits they announce, even if they never become real products. The Nexus phones, Nexus Q, Chromecast, & Google glass were always seen as nearly finished products sold as development kits intended for hackers. The same model of releasing nearly finished products as development kits was followed by Occulus Rift to great success, as least for stock market valuation. It's probably going to be a lot more common.


To be sure, the Ara is not going to be as miniature as a normal phone, but it is going to be extremely miniaturized compared to buying the same modules from Adafruit or Sparkfun & building a phone that way. Making phone level miniaturization accessible to robot projects is going to make autopilots a lot smaller.


It’s definitely a case where the intended cell phone user is just for marketing & the real consumer is going to be someone who wants to connect phone modules to microcontrollers. A standalone LTE transceiver that can plug into a microcontroller would be a game changer. There’s never been any standalone module beyond GSM.


A standalone camera with the quality of a phone cam would be huge. A standalone Linux board for a phone would instantly blow past any current single board computer in miniaturization. The devil is in how much interfacing the unipro bus will cost & how much those phone breakout boards are going to add to the size.



The only disappointment is for all the years we’ve had standalone GSM modules, no-one has ever gotten one to talk to another over its 2 mile range without requiring a cell phone plan. Any LTE module would be equally crippled.
Posted by Jack Crossfire | Apr 30, 2014 @ 04:22 AM | 2,253 Views
In looking for an ideal language for analyzing biometric data, have now tried Javascript, Java, C++, & Python. The problem is the data needs to be graphed on a large monitor for the best results, but should also be graphable on the phone it's recorded on for a preview. Javascript offered the best hope of portability. It clearly is more portable than any other option, but it's going to need a lot of glue to load kml files on the phone. It might need to suck data directly from the logging app. Would say the javascript will never run on a phone, but have only the potential.

The mane problem is navigating a timeline showing curves for instantaneous velocity & altitude. The timeline has to zoom horizontally & vertically. The curves need variable amounts of averaging. There are billions of javascript chart libraries, which all suck in some way or don't work.

It's going to be nowhere near as smooth to navigate as C++. If it was 2011, a simple C++ program would have been the obvious choice. It could have been written in a day, reusing 20 years of past knowledge. It wouldn't have run on the phone, but nothing probably ever is.

It's clear now that all substantial software projects are going to run in web browsers. The non portable stuff is going to be limited to small data collection apps. Most every job is going to require javascript. We live in a world where 1 language can't do everything. It's not very efficient to need to stay on top of Javascript, Java, & C to move data from an embedded system to a phone to a web browser.