Intel Edison generic calculator shield photo journal

[MarkHaysHarris777]

Greetings folks, today I'm pumped because my Intel Edison board (with Arduino breakouts) arrived from Hungary (Arduino Store) by UPS; along with my new Arduino mug, and a very cool blue 20x2 character LCD display. My Edison shield will have a blue|red|white theme, blue 20x2 LCD, white lettering, red pcbs (still in the mails).

The Arduino shield is going to make a great development shield, but I really think the Intel Edison is the micro controller that actually has the stuff needed for a commercial shield for a scientific calculator (supporting full comms, printing, IR, wifi, all on chip).

I have not unpacked it all yet, but I'll get some photos up on this journal sometime today. Did I mention that its tiny? Oh, man, it redefines tiny.


PS. Thanks for pointing me to this board, Pauli !


cheers,
marcus

[MarkHaysHarris777]

Greetings, well, first I really need to thank Pauli one more time for pointing me to the Edison with the mini-breakout board-- absolutely astounding! It has been Christmas around my house, complete with lights. We've got a couple Raspberry PIs glowing red from some corner, and now an Intel Edison glowing green from the other corner (just a fabulous computer science day for me).



The big surprise for me is that this little gem (which may be powered three ways) is a full Yocto gnu/linux system complete with all of the busybox tool suite, and out-of-box gnu/linux server functionality. I had it on the network via wifi in minutes, complete with a web server and full accessability via minicom terminal logon, or ssh logon via remote... a geek can play with this jewel from Intel for 'hours' and never physically connect it to anything; but man, solder a header on this thing and you have full access to the GPIO bus voila!-- you have a controller the size of a book of matches running a full-blown gnu/linux system and all of the pinouts you might expect from an Arduino!

The terminal pic is a screen shot of 'top' running over the Intel Edison via ssh logon from over the network. This board may be powered by a lithium ion batt on the J2 connector, or a 7-12v batt (9v similar to arduino) on the J21 connector, or it may be powered from the 5v pins of the micro usb connector (my choice tonight). Tomorrow we'll try the lithium and 9v options (with measurements); keep you posted.

This board is definitely the way to go. I don't think its going to run on coin cells; I'll know more about that tomorrow... need to study the datasheet and do a little playing.

Well, can't say enough... I'm pumped about the Intel Edison... thanks again, Pauli!


PS... Python is on-board; as well the gcc compiler! ... whoohoo, toys!

Cheers,
marcus

[Paul Dale]

Also, it has quite a lot of flash and RAM included on board -- far more than a calculator will need even including Intel's decimal mathematics library

I really think this board is hard to go past for many applications.


- Pauli

[Gerald H]

Looks fantastic!

[Paul Dale]

(03-19-2015 08:06 AM)Gerald H Wrote:  Looks fantastic!

So is the power draw

Okay, it isn't really but I couldn't resist....


- Pauli

[Les Bell]

(03-19-2015 07:08 AM)MarkHaysHarris777 Wrote:  PS... Python is on-board; as well the gcc compiler! ... whoohoo, toys!

That's the killer language for number-crunching professionals; I've said before that I love the 71B, but if building something in that format today, I'd build it around Python and not BASIC.

Code:


In [1]: (0+1j)**(0+1j)
Out[1]: (0.20787957635076193+0j)

In [2]:

With NumPy and SciPy, it would be hog heaven. And it looks like the Edison would be a great platform.

[Claudio L.]

(03-19-2015 07:08 AM)MarkHaysHarris777 Wrote:  Well, can't say enough... I'm pumped about the Intel Edison... thanks again, Pauli!

Please use this as a base for your hardware calculator designs. It's "big enough" for many other projects, unlike the Arduino boards. If you can put a kit together with keyboard and LCD display I'm sure you'll sell quite a few here.
Just wanted to encourage you to continue with this project.

Claudio

[MarkHaysHarris777]

Greetings, well, my edison mini-breakout has been updated with the latest Yocto gnu/linux image 1.6.1 (this is an absolute requirement, words to the wise) and it has spent much of this day sitting peacefully on a shelf stand-alone battery operated (9v power source on J21 +round -square) running sshd, web server, and some other stuff; how much fun is this?! I made a parallel power source (four 9v batts each via 1N4002 diode) to a Berg connector on J21. The outside power pins are (-) 'square' solder dots, and the inside pins are (+) 'round' solder dots. I have a lithium ion pack ordered. My power draw measurements follow the hi-res pic. This is the tiniest gnu/linux server I've ever supported:



The unit measures (bounces) between 100-120 mA at 8.3v during boot-up; after which it settles in at about 50 mA idle with wifi 'ON'-- 10 mA with wifi 'off'. When the unit is processing keyboard entries (like starting python, or top) the current spikes to 100 mA and then settles in at about 50 mA again... about 415 mW. I do not know yet how to turn off the two bright LEDs on the mini-breakout board (one is board power, and the other is system on, or active). The LEDs are pulling about 10 mA, which jives with my hearsay evidence that says wifi draws 40 mA.

Well, we knew it was going to be higher than 'coin cell' range ( its a dual core 500 Mhz Atom processor for crying out loud) but its somewhat lower than I was expecting ( my Raspberry PI b draws somewhat more than half an amp just sitting there idle!).

I'm excited to try the lithium ion pack /

I will be putting together a getting started guide for this mini-breakout board to help folks get up and running a fast as possible. There is some misinformation out there, and the doc from Intel is kinda sparse. One thing everyone must know is that the edison ships with downlevel Yocto gnu/linux which is very buggy (really irritating, I spare you the details now). All of those problems vanish after the system upgrade.

Cheers,
marcus

[MarkHaysHarris777]

(03-19-2015 09:03 AM)Les Bell Wrote:  

(03-19-2015 07:08 AM)MarkHaysHarris777 Wrote:  PS... Python is on-board; as well the gcc compiler! ... whoohoo, toys!

That's the killer language for number-crunching professionals; I've said before that I love the 71B, but if building something in that format today, I'd build it around Python and not BASIC.

Code:


In [1]: (0+1j)**(0+1j)
Out[1]: (0.20787957635076193+0j)
In [2]:

With NumPy and SciPy, it would be hog heaven. And it looks like the Edison would be a great platform.

Yes. I will be testing pdeclib on this puppy soon... I have been waiting a year and a half to embed pdeclib in a real calculator platform (other than my phone, where it has been running for almost two years). One of the thoughts I have had for the improvement of hand-held calcs is to allow the user to specify arbitrary precision (within reasonable limits). The decimal library I use on my linux systems allows for that easily; it might be time to move that into a calculator.

Calculators of the past allow for configuring the number of digits the user may display... but not the number for intermediate results. WP34s comes close to this with double precision, allowing for scroll. But, what if I allow for 200 digits, and scroll?

I'm not sure why, but they shipped the edison with python2. I wish vendors would stop that! Just install python3 and be done. The decimal library does not have the speed in python2. I'll fix that.

Cheers,
marcus

[Paul Dale]

Try the Intel decimal library. It's large but that isn't a big issue on the Edison. It is also very fast -- many functions are supported my table lookups and interpolation, others use the built in binary instructions to give an initial estimate of the decimal result.

It isn't variable length though.


- Pauli

[Paul Dale]

(03-20-2015 04:33 AM)MarkHaysHarris777 Wrote:  I do not know yet how to turn off the two bright LEDs on the mini-breakout board (one is board power, and the other is system on, or active).

Wire cutters?


- Pauli

[MarkHaysHarris777]

Greetings, well, I moved pdeclib into the edison and ran some voltage|current tests, as well some performance tests. The following hi-res pic is a screen shot of python calculating PI to 1000+ digits using an AGM routine. The import of pilib imports pdeclib as well, which uses Pythons new Decimal module; very fast, arbitrary precision. The highlighted area is the signature for the 1000th place ...201989 python on edison called this out in a matter of a few hundred milliseconds:



Also, the datasheet says that the edison will normally not exceed 200 mA, with occasional current spikes as high as 600 mA during wifi transmit. I have never seen the current jump above 120 mA. I started three wifi sessions to the edison from a VT102 terminal and started a python PI job on two of them... set the dscale(20,000) to ensure that both of them ran for a while... then I took a look at /proc/cpuinfo and the output from 'top'. Both cores were clicking at 500 Mhz with the system showing 99% CPU... all three terminal ssh sessions remained active and responsive and the current never jumped above 120 mA ! The edison board barely gets warm... so, based on an alkaline 9v battery being about 586 mAh (my four pack is then 2344 mAh) means that clicking away at PI for several hours my 9v four-pack will keep this thing powered for 19.5 hours. A lithium 9v four-pack will power the edison for 40 hours under full load; for 96 hours under normal mostly idle load. I have not seen the lithium ion shield pack yet, so I am not sure its precise mAh specs; keep you posted. The bottom line is that battery operation (while not coin cells) is viable for this board-- this is NOT the Raspberry PI load monster!

Cheers,
marcus

[Claudio L.]

(03-20-2015 08:45 AM)MarkHaysHarris777 Wrote:  The edison board barely gets warm... so, based on an alkaline 9v battery being about 586 mAh (my four pack is then 2344 mAh) means that clicking away at PI for several hours my 9v four-pack will keep this thing powered for 19.5 hours.

That's encouraging. A pack of 4 rechargeable AA's (2000 mAh) should keep it running at full throttle for about 8 hours then. That's not bad at all, considering in a calculator it will be idling a lot.

[Claudio L.]

(03-20-2015 07:13 AM)MarkHaysHarris777 Wrote:  One of the thoughts I have had for the improvement of hand-held calcs is to allow the user to specify arbitrary precision (within reasonable limits). The decimal library I use on my linux systems allows for that easily; it might be time to move that into a calculator.
...
But, what if I allow for 200 digits, and scroll?

I agree, it's time to do that on a calculator. newRPL actually uses libmpdec, the same core library powering Python 3's new "accelerated" decimal library (accelerated is quoted because there's nothing hardware accelerated last time I checked, it's just a faster library). A demo with variable precision has been available since May 2014 on a PC, and has actually been running standalone on a 50g calculator since October 2014 (though not publicly available).
newRPL uses 2000 digits as "reasonable limits" for its transcendental functions and for overall memory management. But that limit could be raised with an Edison based calculator.

Sorry, I didn't mean to take your dream of being the first to implement it on a calculator, just wanted to encourage you to follow up on your line of thought, and show you that you are not the only one going in that direction (so your ideas aren't that crazy, it's perfectly doable!).
Keep going, we want a finished open-hardware calculator based on Edison that we can all play with.

[MarkHaysHarris777]

(03-20-2015 02:01 PM)Claudio L. Wrote:  Sorry, I didn't mean to take your dream of being the first to implement it on a calculator, just wanted to encourage you to follow up on your line of thought, and show you that you are not the only one going in that direction (so your ideas aren't that crazy, it's perfectly doable!).

Claudio, thank you, good to hear from you... well like I've said before on this site someplace, there are other people out there building, or trying to build, calculators; I'm not so much seeking 'first' as much as I honestly believe that this community deserves the chance for a lasting legacy and an opportunity to move forward (in a fun way) to take an active role in the development of their own platform/ there are enough DIY and other interested folks on this site to make the project fun and satisfying.

(03-20-2015 02:01 PM)Claudio L. Wrote:  Keep going, we want a finished open-hardware calculator based on Edison that we can all play with.

Yes; the more the merrier--seriously! Thanks for the encouragement. The more I am playing with this board the more convinced I become that the Intel Edison is the winner hands down in every category. I got some more time in on the board today, and I am very impressed with the ease of programming, the organization of the gnu/linux system that manages everything, and the overall flexibility of this platform from an engineering standpoint--EE and comp-sci.

I don't have the appropriate level shifters yet, for testing on my actual proto-board (that will happen next week) but I did write a couple of sketches, essentially blinker sketches without lights, and then tipped that puppy over and took some volt readings with the good 'ol analog VOM to watch the virtual 'lights' and to figure out how the breakouts work... very nice, I might add!

The C|C++ code is compiled and downloaded to a gnu/linux folder called /sketch.

The compiled file 'sketch.elf' is then executed as a root process along with everything else running on the system. In other words, the 'sketch' just becomes another system process managed by the linux kernel. How cool is that?

For those of you who do not quite get the implication of all of this... that means that 'many' sketches can be loaded into this controller at one time and run appropriately from a crontab (automated timed scheduler) and or based on interrupts or other factors. ... this means not only a super calculator, but also a super controller capable of commercial state-of-the-art systems automation and control; all in a teeny tiny chunk of pcb and foil the size of a book of matches!

The only down side to the board (and this is minor) this is a board for grown-ups; to use Walter's phrase, and that means engineers and serious electrical engineering hobbyists. The board uses 1.8v logic, will only stand 3 mA per pin, is NOT forgiving, and requires some thought and know-how. With the right design, correct level shifters, and a solid shield, it should prove to be a truly remarkable platform for hand-held calculation well into the future (and Intel is in the business of providing just such capability well into the future); so I am encouraged. This is going to be fun.

Pleasant evening to everyone... I'm totally pumped!

Cheers,
marcus

[BarryMead]

(03-21-2015 06:49 AM)MarkHaysHarris777 Wrote:  The more I am playing with this board the more convinced I become that the Intel Edison is the winner hands down in every category

Mark: A few weeks a go I posted a thread on this forum about using an "Off the shelf" microcontroller as a way to reopen the worldwide marketing potential for the WP-43S project. They had complained that with the FCC restrictions on importing high-speed electronics it would be impossible to obtain expensive FCC licences for every country, so they had planned on selling/marketing the WP-43S only in the United States. This decision severely cut the potential sales numbers worldwide, and further hindered the economic viability of the whole WP-43S project. I suggested that the WP-43S team should still sell the calculator Worldwide, but that the calculator be a Kit minus the Microcontroller assembly which would be added by the user eliminating the need for FCC testing.

It seems like you (and Pauli) may have discovered the perfect solution to this arrangement with the Intel Edison! It is tiny, powerful, embeddable, reasonably low power, and well priced.

[MarkHaysHarris777]

(03-21-2015 11:47 AM)BarryMead Wrote:  It seems like you (or Pauli) may have discovered the perfect solution to this arrangement with the Intel Edison! It is tiny, powerful, embeddable, reasonably low power, and well priced.

hi Barry, good to hear from you... actually, a couple of months ago I didn't know an Intel Edison from a post hole/! Pauli pointed me to the edison device (I just picked it up and started running). I'm an Arduino fanatic, so I'm relatively new to both the Raspberry PI community, and brand new to the Edison community of engineers and hobbyists. Barry, you are correct, its the best solution I've played around with so far, by far!

I have my edison blinking actual lights today, becoming more comfortable with the system and the pinouts of the mini breakout board. I have attached a hi-res pic of the edison blinking alternate LEDs on pin(s) J19-10 (13) and J20-9 (12). One of the 'fits' with the board (aside from its 1.8v logic, 3 mA pin limit, 130 mA Vcc limit) is that the pinouts are mapped to several designations depending on which development system you're using, and which 'breakout board' you have. By the way, don't bother with the Arduino breakout board... its twice the price of the Edison in the first place, and its larger than the Arduino Due board! I am using the development module (the Edison itself) and the Intel mini-breakout board. Also, for the moment, I'm using the Arduino IDE to build my sketches. The Edison supports Elipse and Python (which may come in VERY handy for the calculator main firmware, we'll see). Hi-res pic (click on it) then explanation:



Vcc is (blue) on the back rail, Vss is (green) on the front rail. The edison has been inverted and plugged into the solderless bread board; its 1.8v logic is being 'shifted' to 8.2v using our friend the LM339. The two resistors nearest the edison are a voltage divider providing the 1.5v ref for the voltage shifter; the other two resistors are the 'source' current limiting (17mA) resistors for the LEDs. The white line output from the edison (J19-10 (13)) is Arduino pin13; the black line output from the edison (J20-9 (12) is Arduino pin12. The sketch code is following:

Code:


/*
  Blink
  Turns on an LED on for one second, then off for one second, repeatedly.

  Most Arduinos have an on-board LED you can control. On the Uno and
  Leonardo, it is attached to digital pin 13. If you're unsure what
  pin the on-board LED is connected to on your Arduino model, check
  the documentation at http://arduino.cc

  This example code is in the public domain.

  modified 8 March 2015
  by Mark H Harris
 */
int t_delay = 700;

// the setup function runs once when you press reset or power the board
void setup() {
  // initialize digital pin 13 as an output.
  pinMode(13, OUTPUT);
  pinMode(12, OUTPUT);
}

// the loop function runs over and over again forever
void loop() {
  digitalWrite(13, HIGH);   // turn the LED on (HIGH is the voltage level)
  digitalWrite(12, LOW);
  delay(t_delay);              // wait a while
  digitalWrite(13, LOW);    // turn the LED off by making the voltage LOW
  digitalWrite(12,HIGH);
  delay(t_delay);              // wait a while
}

-

I like this pic, because the camera caught the sketch 'between' the (13,HIGH) and (12,LOW) instruction(s)! The pin13 is coming HIGH, and the pin12 has not gone out, yet.

This post is demonstrating what I mean by, 'grown-up board'. This is a lot of work and 'understanding' to blink a couple of LEDs. In the voltage shifting, if you make a mistake, it will probably be fatal for the edison. Something I should note is that my 'probes' are 'resting' in the output holes of the edison (which would normally short them out against the 'shield foil' covering the processor underneath! I have placed a small strip of note card as an insulator between the breakout holes and the processor shield.

I have supported the edison with a cylinder of rolled note card just to take pressure (not much) off the power pins.

That's all for now.

Cheers,
marcus

[matthiaspaul]

The Edison is a nice little module and is certainly very interesting for many embedded applications. I specifically like the fact that it is x86-compatible. However, I don't think it is particularly well suited for a calculator, even a high-tech one. While multi-threading is nice and some algorithms can profit from it, it is hardly essential in a calculator to require hardware support for it. The Edison provides a compound processor with three cores, two Atoms running at 500 MHz and one Quark running at 100 MHz. One of them would be enough to accomplish the task. Also, we don't need 1 GB DRAM in a calculator. 32 MB would be more than enough.
Of course, noone would complain about it, if it could be had for "free", but there's always a trade-off. I see a general tendency to prioritize higher speed over lower power consumption, but I don't consider it particularly wise. There are applications, where speed is necessary, but in a calculator? Also, the most common reason for lack of speed today is not missing processor power but inefficient coding.
I for one would hate to change or recharge batteries every couple of days, as much as I hate to recharge my phone every couple of days. In fact, from a tool like a calculator I expect that it will be available without taking care of it all the time, that it will effectively run for at least half a year on a single charge (ideally much longer) under normal use. Therefore, in my opinion, a load current of about 100 mA at 8V is unacceptable for this kind of application.

Therefore, the Intel Quark X1000 series appears to be a better, although still not ideal choice for a calculator. Running at 400 MHz, they have 512 KB SRAM onboard. Unfortunately, the Intel Galileo board, which features such a processor, is too large and contains alot of other stuff not needed in a calculator, thereby increasing the power consumption again, but the processor is also available as a stand-alone chip in low quantities, so it could be used in own "Edison-like" designs (although not everybody will be able to cope with the BGA package).

Intel even announced another Quark variant for the forthcoming Intel Curie module recently - this SoC will come with only 80 KB SRAM and 384 KB flash, but yet lower power consumption. The Curie would be suitable for something like a direct 34S successor, but may be too small for a 43S already.

However, given that Intel is just starting to enter this market segment, it appears likely that they will bring out more Quark variants in the next couple of years... It's certainly worth keeping an eye on this.

Greetings,

Matthias

[MarkHaysHarris777]

(03-23-2015 08:29 PM)matthiaspaul Wrote:  The Edison is a nice little module . . . [but] I don't think it is particularly well suited for a calculator, even a high-tech one.

I'm sorry Matthias, but you have GOT to be kidding?

(03-23-2015 08:29 PM)matthiaspaul Wrote:  While multi-threading is nice and some algorithms can profit from it, it is hardly essential in a calculator to require hardware support for it.

Nope, you're not kidding. OK, well, what is wrong with normal calculators?

Answer... they don't collaborate well, its difficult to print from them (in normal ways, to normal devices) and they almost never interface well with the outside world, generally speaking.

This Intel Edison will provide the platform for a full featured fully collaborative hand-held calculation platform which is also a mobile print server, mobile web server, mobile database, mobile universal plotter (science lab, you fill in the blank). Not to mention that it is fully multiuser, multitasking, modern operating kernel which is open free and universally standard!

(03-23-2015 08:29 PM)matthiaspaul Wrote:  The Edison provides a compound processor with three cores . . . There are applications, where speed is necessary, but in a calculator?

Of course, in a calculator! What was it that the BORG said to Data, "you think so three dimensionally..."

Picture a scientific calculator that not only does arbitrary precision mathematics, but is also collaborative. As you calculate it is assembling a spread sheet and putting it on the web so that your collaborative team can see it as soon as you calculate it... or picture a calculator that can print to *any* of the office printers, or picture a scientific calculator that has full (science lab) connectivity to *any* sensors you want to throw at it, or can easily be connected to a 2D motorized plotter? Not to mention, you can logon to your new calculator from your PC, or notebook, or phone, or tablet, and talk to it from across the room... I can think of a zillion ways to use this device from presentations to industrial control... all from a scientific calculator!

Here another point... people are going to want this thing, just to play with it! Its gonna be more fun, and more productive, you wait and see...

... or, keep telling me why its a dumb idea and will never work. :-p

Cheers,
marcus

[BarryMead]

(03-25-2015 01:59 AM)MarkHaysHarris777 Wrote:  

(03-23-2015 08:29 PM)matthiaspaul Wrote:  The Edison is a nice little module . . . [but] I don't think it is particularly well suited for a calculator, even a high-tech one.

I'm sorry Matthias, but you have GOT to be kidding?

(03-23-2015 08:29 PM)matthiaspaul Wrote:  While multi-threading is nice and some algorithms can profit from it, it is hardly essential in a calculator to require hardware support for it.

Nope, you're not kidding. OK, well, what is wrong with normal calculators?

Answer... they don't collaborate well, its difficult to print from them (in normal ways, to normal devices) and they almost never interface well with the outside world, generally speaking.

This Intel Edison will provide the platform for a full featured fully collaborative hand-held calculation platform which is also a mobile print server, mobile web server, mobile database, mobile universal plotter (science lab, you fill in the blank). Not to mention that it is fully multiuser, multitasking, modern operating kernel which is open free and universally standard!

(03-23-2015 08:29 PM)matthiaspaul Wrote:  The Edison provides a compound processor with three cores . . . There are applications, where speed is necessary, but in a calculator?

Of course, in a calculator! What was it that the BORG said to Data, "you think so three dimensionally..."

Picture a scientific calculator that not only does arbitrary precision mathematics, but is also collaborative. As you calculate it is assembling a spread sheet and putting it on the web so that your collaborative team can see it as soon as you calculate it... or picture a calculator that can print to *any* of the office printers, or picture a scientific calculator that has full (science lab) connectivity to *any* sensors you want to throw at it, or can easily be connected to a 2D motorized plotter? Not to mention, you can logon to your new calculator from your PC, or notebook, or phone, or tablet, and talk to it from across the room... I can think of a zillion ways to use this device from presentations to industrial control... all from a scientific calculator!

Here another point... people are going to want this thing, just to play with it! Its gonna be more fun, and more productive, you wait and see...

... or, keep telling me why its a dumb idea and will never work. :-p

Cheers,
marcus

Mark: Matthias wasn't trying to discourage you from using the Intel Edison, merely suggesting that it is significantly more powerful than typical calculator processors. I agree that having more power gives you greater capability to grow the feature set of the device, but with growth in features comes growth in software development and if the calculator were to do all of the things you described including interfacing with spreadsheets, printers, plotters, other collaborating scientists or students, etc, this would require a whole team of software developers and the budget of a large corporation to pull it off. From my reading of his posting he was attempting to match the REQUIRED FEATURES of the WP-43S with the CAPABILITIES of a suitable microcontroller. Everyone knows that more processing power can always be UNUSED, but the cost of keeping it around when you don't need it is power consumption. The way I understood his feedback, he was trying to be helpful in matching the OPTIMUM power consumption (MINIMUM REQUIRED CPU FEATURES) with the workload that would be required of the WP-43S. I know and I think Matthias knows that you can make the calculator work with a faster more capable CPU, but the batteries will not last as long as if you selected a less powerful CPU. There is a "Sweet Spot" with maximum battery life and just enough processor features. I don't think he intended to suggest that it was a "DUMB IDEA" or that it "WON'T WORK". I am only saying how I read his feedback, he may have a different opinion.
Respectfully, Barry