Orange Pi One Woes

[rs20]

True.
But it's about "the vibe". You go to the website and most things are 6-9 months old, or simply missing like the OPi One.
Someone could get the mistaken impression that they don't care.

IMHO that is a common misconception. Some software is just finished as it is and doesn't really need any more tinkering.

I think that you raise a good point that blindly expecting firmware/software to be continually updated is unreasonable -- after all, there may be no updates to perform if the device's software stack is simple. However, in this case we're talking about a linux distribution! There are software updates happening continually upstream -- important security updates included. Even if Dave's particular usecase doesn't demand security, it's fairly concerning that all of the bugs disclosed in the past year are still in the "latest" build.

Any company worth its salt should be able to have computer sitting in the corner of the office that's continually updating, building new branches, deploying to a test OPi, testing it to death, and pushing the new image to the website (using a content distribution system that isn't Baidu, for bonus marks).

It makes no difference how right a company is technically, vibes matter.

Vibes matter, security holes kinda matter too 

[ludzinc]

But it's about "the vibe".

[imidis]

Back in my day we compiled our own kernels....

I ordered one, I like the challenge.

[rsjsouza]

Yeah; one of the things you notice eventually is that it's not really about who can build hardware for the lowest price.  It's about who can create hardware with a good support infrastructure for the lowest (or "a reasonable") price.  Arduino did it (in a way that lets a lot of others ride their coattails), Raspberry Pi did it.  MBed is pretty good.  BeagleXXX is ... close.  A *LOT* of others - not so much.

westfw nails it. What would you expect from a board based on a device whose datasheet looks straight from the back of a (used) napkin? The whole solution is far from mature. 

True.
But it's about "the vibe". You go to the website and most things are 6-9 months old, or simply missing like the OPi One.
Someone could get the mistaken impression that they don't care.

IMHO that is a common misconception. Some software is just finished as it is and doesn't really need any more tinkering.

I think that you raise a good point that blindly expecting firmware/software to be continually updated is unreasonable -- after all, there may be no updates to perform if the device's software stack is simple. However, in this case we're talking about a linux distribution! There are software updates happening continually upstream -- important security updates included. Even if Dave's particular usecase doesn't demand security, it's fairly concerning that all of the bugs disclosed in the past year are still in the "latest" build.

Any company worth its salt should be able to have computer sitting in the corner of the office that's continually updating, building new branches, deploying to a test OPi, testing it to death, and pushing the new image to the website (using a content distribution system that isn't Baidu, for bonus marks).

In my experience, keeping HW drivers current with the upstream Linux requires quite an effort; that and the fact no device/board comes out of the fab with fully working/optimized everything. Even in companies worth its salt, there's always peripheral support gaps and optimization left to be implemented after the device/board is released - after all, most of the times the device itself is far from complete when the SW teams start working on the device drivers. 

[macboy]

"It's like the 5$ multimeters sure as hell they are cheap, butbare they safe/accurate??"

Dont know about those 5usd ones. But I got a few 0usd ones and they are all very save and accurate.

Just because you are not dead yet, doesn't mean they are safe.
Oh I see what you did there. You claimed it's "save", and it did save you money, you paid $0 for it. So yeah, "save" but not safe. Accuracy is trivial at 3.5 digits.

[rstofer]

Sometimes trivial accuracy is all it takes.  We just went through a battery replacement followed by an alternator replacement (the real problem) and I thought I would teach my grandson how to tell the difference.  A little 3.5 digit DVM is just the thing and the cheaper the better.  There is a place for $5 multi-meters.  We certainly won't be using my Fluke 189 to work on a car.

[jnz]

Sometimes trivial accuracy is all it takes.  We just went through a battery replacement followed by an alternator replacement (the real problem) [...]  We certainly won't be using my Fluke 189 to work on a car.

Considering your annecdote... I'm wondering if the symptom that you replaced a battery when the alternator was the problem wasn't from the cause you didn't want to use a good tool for the job?


As to the topic at hand... I've in my life seen 3 or 4 "complete" softwares that I don't mind have a date a few years old on. I have to agree with Dave, perception is reality when it comes to updated software.

As to Chinese copy mfg and needing to roll your own software to save a couple of bucks... I wonder where some of you guys make the time to do stuff like this?

[rstofer]

Sometimes trivial accuracy is all it takes.  We just went through a battery replacement followed by an alternator replacement (the real problem) [...]  We certainly won't be using my Fluke 189 to work on a car.

Considering your annecdote... I'm wondering if the symptom that you replaced a battery when the alternator was the problem wasn't from the cause you didn't want to use a good tool for the job?

I wasn't involved in the troubleshooting until the new battery failed the day after it was replaced.  That's the whole point of the educational opportunity.  I KNOW how to troubleshoot electrical systems, my grandson does not.  It would be nice to transfer some of my experiences while I am still around.

I only need a couple of numbers off of the meter.  Somewhere around 12.6V when I walk up to the car and the battery is charged.  If it is much less, there's an issue.  Somewhere around 13.8V when the alternator is turning (battery charged) and it probably works.  If I only get 12.6V or less when the engine is running, the alternator is toast.  Turn on the headlights and watch the battery voltage for a few minutes...

So, I just walked out to my truck and the battery measured 12.2V.  Fine!  It hasn't been driven in a week, I wouldn't expect it to be fully charged.  I started the engine and got 15.1V.  Again, fine!  It is charging the battery that we already know is a bit low. Done!

One other useful measurement that is easier done with an analog meter is to watch the battery voltage when the engine is cranking.  If it drops much below 10V, there is some kind of high internal resistance.  I have seen batteries that appear to be perfectly charged be totally incapable of cranking an engine.

None of these measurements require even 3.5 digits.  A $6 Harbor Freight meter is entirely adequate.  I cheated and used my spiffy new EEVBlog meter but it certainly wasn't necessary.

[stmdude]

I started the engine and got 15.1V.  Again, fine!

I think I've been around cars too much...
15.1V is not fine. Indicative of a faulty charge regulator.  But, it's integrated into the alternator on almost all modern cars, so swapping the alternator is the fix.

13.8V is where it's supposed to be when the motor is running.

[rstofer]

Back to the main topic:  I'm interested in Dave's "Super Computer".  And I have no idea why...

What kind of problems can it be used for, how are problems factored to allow for parallelism and, given the limited inter-processor bandwidth, what kind of performance can be achieved?  I use Xilinx Vivado and I understand it runs well on a 4 Xeon 32 thread machine but that's a bit out of my league.  Is it possible for the Linux process scheduler to run threads on different boards?  I know there are a number of Linux Cluster projects out there but I haven't been following along.  But with $10 boards, these things become a lot more interesting.

I could see buying a bunch of $10 boards and tying them together somehow (Ethernet, USB, <something else>) but I'm not seeing the big picture.

There has to be more to it than just creating a bunch of SETI nodes.

[jnz]

I think I've been around cars too much...
15.1V is not fine. Indicative of a faulty charge regulator.  But, it's integrated into the alternator on almost all modern cars, so swapping the alternator is the fix.

15V on a new vehicle is fine. 14.8 is the typical "target voltage" for the platforms I've tinkered with.


As to the topic at hand... I just saw the video and I'm stunned if anything still thinks the cost difference in the Rasp vs Orange aren't instantly made up in software support. More so, that it makes the RaspPi look like the deal it is.

[rstofer]

I started the engine and got 15.1V.  Again, fine!

I think I've been around cars too much...
15.1V is not fine. Indicative of a faulty charge regulator.  But, it's integrated into the alternator on almost all modern cars, so swapping the alternator is the fix.

13.8V is where it's supposed to be when the motor is running.

13.8 is the right number when the battery is fully charged.  The 15.1V will very probably drop down in a minute or so.  I just didn't let it run more than the time it took to measure the voltage.  If I drove it around the block, I would expect that 13.8V number.

[rstofer]

I think I've been around cars too much...
15.1V is not fine. Indicative of a faulty charge regulator.  But, it's integrated into the alternator on almost all modern cars, so swapping the alternator is the fix.

15V on a new vehicle is fine. 14.8 is the typical "target voltage" for the platforms I've tinkered with.


As to the topic at hand... I just saw the video and I'm stunned if anything still thinks the cost difference in the Rasp vs Orange aren't instantly made up in software support. More so, that it makes the RaspPi look like the deal it is.

Sure, but how many iterations of Linux have there been to get it to the state it's in?  And let's not count Ubuntu with the Unity desktop.  Software is an iterative process and it will take some time for the Orange PIs to catch up.  First they have to get <something> working, then they need to sell enough boards to create a critical mass and then the whole world will start working on drivers.  It wasn't so many years ago that getting WiFi working on Linux was a magic act.  It took several years to get it straightened out.  Now, nobody remembers...  Yes, it was primarily due to the dongle manufacturers not releasing information but, still, I remember spending hours getting WiFi to work.

[XFDDesign]

A bit of thread necromancy, but does anyone know if the lubuntu for OrangePiPlus will work on the OrangePiLite? I can't imagine why not, but the link for the lite is DOA.

[CJay]

I think I've been around cars too much...
15.1V is not fine. Indicative of a faulty charge regulator.  But, it's integrated into the alternator on almost all modern cars, so swapping the alternator is the fix.

15V on a new vehicle is fine. 14.8 is the typical "target voltage" for the platforms I've tinkered with.


As to the topic at hand... I just saw the video and I'm stunned if anything still thinks the cost difference in the Rasp vs Orange aren't instantly made up in software support. More so, that it makes the RaspPi look like the deal it is.

Sure, but how many iterations of Linux have there been to get it to the state it's in?  And let's not count Ubuntu with the Unity desktop.  Software is an iterative process and it will take some time for the Orange PIs to catch up.  First they have to get <something> working, then they need to sell enough boards to create a critical mass and then the whole world will start working on drivers.  It wasn't so many years ago that getting WiFi working on Linux was a magic act.  It took several years to get it straightened out.  Now, nobody remembers...  Yes, it was primarily due to the dongle manufacturers not releasing information but, still, I remember spending hours getting WiFi to work.

I have also spent a lot of time fiddling around trying to make WiFi work on Linux, it was, to be polite, awkward, and resulted in a lot of foul language as well as a few Windows installs.
 
I managed to kill a fresh Ubuntu (10.04 I *think* but it could have been earlier) install to the point where it kernel panicked on boot by just editing the WiFi files. I have no idea how, I explained what I *think* I did to several people and they were baffled.

That one got a format too.

[Monkeh]

I managed to kill a fresh Ubuntu (10.04 I *think* but it could have been earlier) install to the point where it kernel panicked on boot by just editing the WiFi files. I have no idea how, I explained what I *think* I did to several people and they were baffled.

I'm baffled without you explaining anything. But I think you'll find poking blindly will do much the same to Windows..

[CJay]

I managed to kill a fresh Ubuntu (10.04 I *think* but it could have been earlier) install to the point where it kernel panicked on boot by just editing the WiFi files. I have no idea how, I explained what I *think* I did to several people and they were baffled.

I'm baffled without you explaining anything. But I think you'll find poking blindly will do much the same to Windows..

I'm just as baffled but it was too long ago to find or remember full details and as it was a clean install, it was quicker to start afresh than spend time trying to fix it.

It was back in the day when you needed to mess about with supplicant files and download 'stuff' to get it working.

I am 99.9% certain I changed nothing other than files named in a guide for configuring WiFi but I suppose it could have been a nasty coincidence involving a bad sector or an incomplete write somewhere.

FWIW, I like both Windows and Linux but my Linux skills are lagging a good way behind my Windows knowledge as I make a decent living from poking around in the guts of Windows from NT to 10 on desktops and servers, I merely offer the anecdote as amusing and indicative of my luck/lack of linux skill.

[eneuro]

Back to the main topic:  I'm interested in Dave's "Super Computer".  And I have no idea why...

I've bought Nvidia 750 Ti with 640 CUDA cores and 4GB graphics card memory-I wonder how many Orange One Pi needed to beat this 60W graphics card in searching for aliens? 

Anyway, thank sto @EEVblog post I've in my hands Orange Pi One with 16GB Armbian SD card, however I need power this thing not from wall socket but 12V lead acid battery, so I've DC-DC  but unsure how Orange Pi One PCb looks like under solder mask in the area of this nasty small power plug jack not compatible with many common power supplies.

How to remove this power plug and do not short circuit PCB trackc in the area of this power plug? 
It coule be nice see Orange Pi One PCB layout, while destructive method by removeing solder mask is not an option-I'd like to have not damages solder mask and ensure that soldering DC-DC 5V custom power wires do not lands too close to PCB tracks...

Thanks in advance for someone who already removed solder mask in this part of Orange Pi One and could help see how PCB layout looks like close to this power supply jack?

Update: Never mind-on this Orange Pi One bottom photo it looks like we can use those unused holes under jack to put own power supply wires inside and quite easy solder and glue own after removing this bloody useless jack power connector   

[eneuro]

That is interesting-after desoldering this useless jack power connector there is a lot of space on top of Orange Pi One to solder positive (red) terminal and easy to solder negative symetric on the bottom side below  by reusing those two ground jack connectors



Now powered from 12V car battery using $1 SMPS to step down to 5V .



It has installed armbian server version upgraded with www & mysql & PHP - reday to communicate via web browsers with connected hardware using two I2C interfaces available on Orange Pi One GPI/O connector 

Does anyone knows howto check CPU temperature within Armian Linux OS using script or a few lines of C code ?
Temperature is shown after login and at ~20*C abmient during  setup of WWW & Mysql & PHP login script wshowed not more than 45*C -most time ~40*C without any heatsink ...

It works  !

[arekm]

cat /sys/class/thermal/thermal_zone1/temp

Here 40C, too but with heatsink and inside http://www.thingiverse.com/thing:1447933 case.

[eneuro]

Thx it works 

Quick native ARM development in C gives us those CPU temperatures in 1 second intervals:

$  ./opotemps
45 47 46 47 46 47 44 47 46 46 46 46 45 47 45 45 47 45 45 45 45 46 44 47 45 44 45 46 46 44 45 45 46 46 46 46 45 45 47 45 46 47 45 47 45 47 45 47 45 46 45 45 44 45 46 45 45 44 45 45 44 46 46 45 45 46 45 44 44 45 45 44 45 44 45 45 45 45 45 44 46 45 47 47 47 47 47 46 47 45 46 48 45 46 47 46 48 46 47 46 49 47 44 45 49 46 45 45 44 44 44 45 44 45 44 44 45 45 45 45 44 45 43 44 43 45 44 45 47 46 45 45 44 44 46 43

This is during instalation of Qt SDk for ARM.

We have almost everything to start more serious native application development on this device.

I'm interested in I2C (TWII) GPIO access from Qt application in C/C++ so quick listing of available I2C's on Orange Pi one leads to this:

$ ls -l /dev/i2c-?
crw------- 1 root root 89, 0 Jul 21 22:34 /dev/i2c-0
crw------- 1 root root 89, 1 Jul 21 22:34 /dev/i2c-1

However, who knows whether those two I2C CPU pins vs GPIO assigment is correct for Orange Pi one ? 



It looks like we have 3.3V I2C (TWII) if this pins schematics is correct, so we need to do some volatge level switching to connect many eg. 5Vcc powered 16MHz ATTiny85 MPUs as I2C slaves doing other hard job of gathering interesting information...

What about those GPIO 3.3Vcc pins current capabilities to power up directly some I2C 3.3V capable devices?

[martinayotte]

Yes, those pins are correct PA11/PA12 for I2C-0 and PA18/PA19 for I2C-1.

If you need voltage level-shifter, simply use MOSFET/PullUps like in this :


I'm powering a MCP23017 directly from 3.3V found on the header pin1, in such case, no needs for level-shifter.

[eneuro]

Yes, those pins are correct PA11/PA12 for I2C-0 and PA18/PA19 for I2C-1.

Thx, hopefully pin 1 is marked and it has 3.27V.
However, measured voltages on I2C PA11/PA12  and PA18/PA19  are also 3.27V when powered on Armbian jessie server -does it mean we do not need any pull up resistors for I2C in Orange Pi One?

I have more time today play with I2C on this device-didn't check so far if there is any resistance between those I2C pins and 3.3Vcc close to them 

Update: Yep, powered off and disconnected from battery it looks like there are 2kOhm pullups to 3.3V there... if I'm right?

Nice, no worries: I've wrote a few lines of C code to try access ADXL345 with its ADo pin grounded (i2C address 0x53) connected without any pullups to Orange Pi One i2C bus 1 detected with this I2C Linux utility:

Code: [Select]

# i2cdetect -r 1
WARNING! This program can confuse your I2C bus, cause data loss and worse!
I will probe file /dev/i2c-1 using read byte commands.
I will probe address range 0x03-0x77.
Continue? [Y/n]
     0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f
00:          -- -- -- -- -- -- -- -- -- -- -- -- --
10: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
20: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
30: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
40: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
50: -- -- -- 53 -- -- -- -- -- -- -- -- -- -- -- --
60: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
70: -- -- -- -- -- -- -- --
and than run simple ARM code compiled with gcc:

Code: [Select]

root@orangepione:/tmp# gcc -Wall  adxl345i2c.c  -o adxl345i2c
Custom software  reads from ADXL345 register 0x00 what datasheet says: 0xE5 , so we were able easy connect to I2C devices from Orange Pi One powered by Armbian Linux 

Code: [Select]

root@orangepione:/tmp# ./adxl345i2c
Device 0x53 (ADXL345 with ADO->GND) register: 0x00 data: 0xE5

[lemi1]

hello.I am new here and hawe bad english.But if you can help me.

Poslano z mojega ONE A2003 z uporabo Tapatalk

[lemi1]

I want to run on orange pi pc plus h3 1gb ddr ,print server,nas or media server  and conect 1tb ex disk on it to run torrent box all on one board.I biy 2 but get one.Waitin gor second.
Is able to someo e upload som img.reddy for do that.Cous I hawe problem to install all that.Try many of image but alwais hawe new priblem.Pls help.Thank you.

Poslano z mojega ONE A2003 z uporabo Tapatalk