Serious issues V2

[Rob F]

One other thing, if the display isn't connected to the controller it will stay on the splash screen with tank and version 3.0. Now I think after all this that's what's happening. You unplug from controller, take to computer, it programs and restarts but stays on the splash screen that is normal. It never receives a response from the controller to it doesn't finish starting up.

[Arne]

Hi Rob,

Thanks!

First I have to set something straight: the missing line, SUCCESS - Found homeIcon.raw file. didn't show up because I was using the first software version while testing.
So in V2.90 this code is not there:

Code:

if (!sd.exists("homeIcon.raw")) {
  Serial.println("ERROR - Can't find homeIcon.raw file!");
} else Serial.println("SUCCESS - Found homeIcon.raw file.");

Well I cleaned up, removed all older versions so I could not mix up again...

Unziped the latest version V2.93 and started again:

Slow formated the display memory card (FAT32) and carefully, to be 100% sure, placed all files on both mem cards.
Placed the Libraries files in the correct location and connected the internet cable to router.

Flashed the controller then flashed the display with the Reset_Controller file and then flashed the display file.
I checked if the mods you gave me were in place and whether the display was correctly chosen in line 24 or so.

This is wat happens:

"Thank you for choosing Robo-Tank!"
OK
"Setting up controller"
"An update has been installed!"
OK
"Updating controller..."

And then it hangs at the first screen.

I opened up the serial monitor which makes the display reset, is it reset?
This is what the message in the serial monitor says:

SUCCESS - sd card initialized.
SUCCESS - Found homeIcon.raw file.
LOADING 1: 1423
LOADING 2: 1425

And it just hangs... sorry...

I tried this a couple of times, same result.

[niksunen]

For some reason controller not send (or display cannot read) command to "start display and load homescreen. So im think, that your problem lies to controller, not display. Example if controller cannot read sd card, or find it there "index.htm" file, it hang and boot "order" don't go further. And it need to get all the way, before controller send command to display start, and draw / load homescreen.

You can next use that serial monitor and open it to contoller. And then reboot system, And if it fails to find sd card or says it corrupted. It helps when you take card out, and clean it well :D And put it back in again. (i have to do this once, and then serial monitor show that it cannot open sd card)

[Arne]

Thanks Niksun,

I connected to the controller and started serial monitor:

SUCCESS - sd card initialized.
SUCCESS - Found index.htm file.

---

Sorry Rob, I missed your last reply.

Normaly I keep the display and controller connected, als when flashing software.
On the other hand, RT was powered and hung at the splash screen.
I disconnected the cable I normaly used and connected a short patch cable.
It directly went to the home screen and is running fine now.

Serial monitor says:

SUCCESS - sd card initialized.
SUCCESS - Found index.htm file.
Estimated pre-scale: 12.56
Final pre-scale: 13


I will do some testing to see if I can simmulate this so we can be sure the cable is bad.

[niksunen]

Thanks Niksun,

I connected to the controller and started serial monitor:

SUCCESS - sd card initialized.
SUCCESS - Found index.htm file.

---

Sorry Rob, I missed your last reply.

Normaly I keep the display and controller connected, als when flashing software.
On the other hand, RT was powered and hung at the splash screen.
I disconnected the cable I normaly used and connected a short patch cable.
It directly went to the home screen and is running fine now.

Serial monitor says:

SUCCESS - sd card initialized.
SUCCESS - Found index.htm file.
Estimated pre-scale: 12.56
Final pre-scale: 13


I will do some testing to see if I can simmulate this so we can be sure the cable is bad.

Good that you got it working :) 


Offtopic: Those two lines "Estimated pre-scale: 12.56, Final pre-scale: 13" are coming to Adafruit_PWMServoDriver library, and that output has nothing to do with robotank. And they can take off, by editing this  Adafruit_PWMServoDriver.cpp file and find there line 28 

Code:

// Set to true to print some debug messages, or false to disable them.
#define ENABLE_DEBUG_OUTPUT true

And change it to 

Code:

// Set to true to print some debug messages, or false to disable them.
#define ENABLE_DEBUG_OUTPUT false

I do this because don't like anything "extra" to coming serial monitor, just what i want :D

[Arne]

Thanks Niksun,

I have the usb connected to the controller and the serial monitor window open, RT runs at splashscreen with the cable I normaly use and have used before.
When I disconnect and reconnect the RJ cable on the display side, the display resets > splash screen.
But... the time it normaly needs to go to the home page, after this time the splash screen stays on but the green connection icon appears...
Like I said, it takes about the same time I think.
Looks like it's only showing a part...

Next I keep power on and connect a very short patch cable in stead of the "old" one, IT RUNS... ;-)
This is what the serial monitor says:

Estimated pre-scale: 12.56
Final pre-scale: 13
Re-connecting: 88725

I honestly think it has to do with the cable.
Unbelievable, were talking about new cables...

[niksunen]

Is your old cable straight or crossover ? I don't know if this really effect, because i don't know how Rob is taking care both situations. Because there is two types of rj-45 cables. And i dont ever see anything mention, that you should use only some of them in robotank.

There is picture of their diffrenses. I think that power go through pin 1 and 2 because they are same on both cable types. But i dont know how rest goes, and if it really nothing matter :)  And maybe serial 3 is using pins 4&5 because they are same also in both situations, but then serial 1 & 2 swap places depending wich cable type you use. And i dont know how this is handled because, i think that serial connections are "hard coded / wired", unless ic chips somewho change them according wich cable type is used. But if not, i don't know if system will work both of them ? Because then serial 1 and 2 can mixed up and this should effect controller somehow.

   

[Arne]

Thanks Niksun,

Well, I also thought about that but this cable "worked" in the old situation.
I have to order a new one so maybe Rob can shine a light on the cable subject and let us know what it should be, crossed or straight.

[niksunen]

Thanks Niksun,

Well, I also thought about that but this cable "worked" in the old situation.
I have to order a new one so maybe Rob can shine a light on the cable subject and let us know what it should be, crossed or straight.

Yep Rob can tell us more :) 

But if im not remember wrong, Rob change this v2 / 2 gen how informations are carried between display and controller. Example if i use modified version of robotank 2.3 (this version is designed to first gen, and only slight modified to work with new gen) i get green connection logo, because in that version it uses serial 2 to carry this information. And now with current versions i dont get it anymore, because my ic is faulty and serial 3 is "broken" and this is what carry that connection "information"

But like i said (and you :D ) only Rob can tell us how its designed and will cable type effect anything

[Arne]

You probably recognize this Niksun and I'm sure Rob will too, it's just a $1.00 dollar cable and I have been sitting on this for a couple of days now... hell... ;-))

[Rob F]

Glad to hear you're getting closer. I've come across a couple cables that the meter said should work but they didn't, its rare but can happen. The Ethernet patch cable should be straight through, not a crossover. If a crossover cable is used it won't work, nothing will communicate, shouldn't damage anything but it won't work. In the original version 6 pins were being used new version all 8 are being used.

[Arne]

Yeah, I know Rob, the cables I have overhere are straight also.

Just like you said, I also tested both cables beeping them through, both seems to be fine.
One works and it looks like the other is causing problems...

The shorter one is a round patch cable, the longer one (2m) is flat, both practicaly new.

Rob, with my RT first gen. I noticed that the light temp now and then said N/A eventhough there was a sensor connected.
I didn't mind because I knew it had to be changed anyway.
Could this also be caused by this faulty cable?

[Rob F]

Yeah goes to show not all cables are the same. On the old version you shouldn't have gotten that with light temp, could have been the cable. There's still a couple bugs in the new software with the sensor data but it's almost sorted out. Should have another patch out this weekend.

[Arne]

Oh, but that's not a problem Rob.
Now I know it's almost definitely the cable I can start rebuilding everything so I have lights above the tank again.
The light temp looked fine with gen1, so it will be fine now too... ;-)

I have to sort out another thing, when I press feed, I get to the dosing and the other way around...
It looks like the touch pads are mirrored around the vertical centre in the middle of the screen... ;-)
Guess it has to do with the cal. file eventhough I'm quite sure I picked the right one, but you never know.
What's the easiest way Rob?

---

Replaced the cal file in libraries and flashed the display again, worked... ;-)

[niksunen]

Oh, but that's not a problem Rob.
Now I know it's almost definitely the cable I can start rebuilding everything so I have lights above the tank again.
The light temp looked fine with gen1, so it will be fine now too... ;-)

I have to sort out another thing, when I press feed, I get to the dosing and the other way around...
It looks like the touch pads are mirrored around the vertical centre in the middle of the screen... ;-)
Guess it has to do with the cal. file eventhough I'm quite sure I picked the right one, but you never know.
What's the easiest way Rob?

I personaly calibrated my display by myself, so i dont use any "ready" values. I used UTouch_Calibration.ino (you can find it Utouch library example files. Edit right display type, change right touch pins, and add code

Code:

 pinMode(9, OUTPUT);
 analogWrite(9, 255);

to right place in sketch, because you dont get backlight without it :) 
And then upload sketch to display, and do the calibration. Then you can read display calibration values, and write them your own calibration file, then it is most accuracy :)

These are my display calibration values (in calibration it shows 8 points around the display, and you have to click them all, and then program calculate accurate calibration values, just for your screen) 

   

[Arne]

It works now but thanks for the tip Niksun.
You're never too old to learn...

[niksunen]

It works now but thanks for the tip Niksun.
You're never too old to learn...

Yep, they work most cases with those calibration files, but i wanted it to be accurate as possible, so i did it myself. And like you said, it always good learn things. And now i know how i can calibrate screen if ever needed somewhere else :D

[Arne]

You've been a great help to me Niksun, thanks !
What should we do without you and Rob... ;-))