0

Pi 3B or 3B+ / Amp2 / pi official 7‘‘touch display

sirvolcano

December 30, 2018 14:25

None

Follow

Hi everybody

Hifiberry amp2 I have a problem…

I have a raspberry pi model 3B and a model 3B+. I want to combine one of them with the hifiberry amp2 and the pi official 7’’ touch display. The combination is powered via Amp2 with the recommended power supply (Mean Well: GST60A18-P1J) which works fine.

On the SD-Card I installed OSMC (Linux osmc 4.14.78-2-osmc) where the media center Kodi 17.6 is running.

Pi 3B/3B+ works fine, as well as the display and amp2 but my problem is the touch function of the display:

The touch function doesn’t work or works intermittently and or delayed - but most of the time it doesn’t work, so at the moment it is not usable for navigation in Kodi. But when I detach amp2 and run the screen with pi only the touch function works fine almost as good as on a smart phone. This problem I have with both models of pi (3B and 3B+).

In some posts I read that the problem might be because of the I2C-Bus which is needed and used by the amp2 and the touch function of the display. Without amp2 I don’t need to connect the additional jumper for SDA and SCL everything works fine. When amp2 is attached, no matter if SDA/SCL is connected or not, either it doesn’t work, or it works very poor as explained above.

So, please can you help me with this issue? Is this hardware combination a total no go combination or is it only a configuration problem?

I tried to give you as much information as possible about my hardware/software setup. If you need more information just go ahead and ask for it. Thanx in advance for helping me!

Best regards

• Facebook
• Twitter
• LinkedIn

11 comments

• 0
  
  sirvolcano January 03, 2019 14:43

  Hi everybody,

  anyone an idea how to solve this problem? Or any hint which direction I should look at?

  Thanx in advance.

  Best regards

• 0
  
  sirvolcano January 04, 2019 16:09

  Hi,

  just an update - I have the same problem not only with OSMC as well the problem is the same in rasbian. Rasbian is installed on a separate SD-Card updated and upgraded, here the information of rasbian version (Linux raspberrypi 4.14.79-v7+ #1159 SMP Sun Nov 4 17:50:20 GMT 2018 armv7l GNU/Linux).

  The problem is identical to that I explained above. Touch function of the display works perfect without Hifiberry amp2.
  As soon as amp2 is attached everything works (sound via amp2 as well) still fine except the touch function, either it is very poor or not existing! It doesn’t matter if I use pi model 3B or 3B+, same behaviour on both pi models.

  Any ideas!? I know from other forums that this combination of hardware works fine!

• 0
  
  sirvolcano January 04, 2019 16:16

  Could be there any amp2 hardware problems - that it might disturb/interfere other peripherials ? (Well the sound via amp2 works very fine when it is attached!)

• 0
  
  HiFiBerry team January 04, 2019 16:17

  The display I2C connection isn't supplies on the GPIO, therefore I don't think there can be any interference. 
  Try powering the display with a separate power supply.

  Best regards,
  Daniel

• 0
  
  sirvolcano January 04, 2019 17:11

  Hi Daniel,

  thank you for the hint. I tried following:

  - Powering pi and amp2 with "Mean Well: GST60A18-P1J" via map2

  - Powering pi original 7'' touch screen directly with a power supply 5V - 2,5A

   

  Result touch function works much, much … better, though not as good as when the amp2 is not attached but really it woks much better.

  Sometimes it doesn't gather the touch command, so that I have to press on the control field several times until touch command is gathered - this happens mostly at the periphery of the screen. 

  But now my questions:

  Do I have a powering problem? Or might it be in combination with something else due to touch function is much better but not 100%!?

  Why it works with an additional power supply and not with powered via amp2. Amp2 is advertised that it is able to supply as well peripherals - That was one of the main reason why I decided to buy the amp2. 

  So do I have here a powering issue? What do I have to do to get this problem under control? I don't want to use additional power supply - I want to run all with only one supply!

  Would the problem be solved if I used a supply with more power and which one would you recommend? 

  Best regards

  Edited by sirvolcano January 04, 2019 17:13
• 0
  
  HiFiBerry team January 04, 2019 17:26

  Hi,

  the Meanwell power supply is powerful enough. Unfortunately I can't say what might be the problem in your setup.

  Best regards,
  Daniel

• 0
  
  sirvolcano January 06, 2019 17:18

  Hi Daniel,

  ok, unfortunately I don't have a multimeter but as soon as I get one I will check the power supply at 5V pin at amp2. As well I will check the power supply when the jumper cable are connected to amp2 5V pin, if there is any voltage drop via jumper cable. 

  I am expecting here a constant supply of 5V - but what is the nominal current and max current at the amp2 5V pins?

  I read in the documentation of the official display that it needs about 400mA - is the amp2 able to deliver this at least?

  … as the display consumes approximately 400mA.

  See here, docs: https://www.raspberrypi.org/documentation/hardware/display/ 

  My hardware setup is connected as following:

  • Amp2 is attached to the pi via 40 pin socket
  • Touchscreen is connected via ribbon cable to the pi DSI port
  • Touchscreen is connected via jumper cables to amp2 5V pin and GND pin for power supply
  • No connection to the SCL/SDA pins of GPIO!
  • All are powered via amp2 with the Meanwell power supply

   

  In the same doc (see link above) they are talking about the I2C Bus:

  NB: With the display connected to the GPIO I2C pins, the GPU will assume control of the respective I2C bus. The host operating system should not access this I2C bus, as simultaneous use of the bus by both the GPU and Linux will result in sporadic crashes.

  In my config Linux access is commented out - so only GPU I2C access is ensured:

  #dtparam=i2c_arm=on

  #dtparam=i2s=on

  #dtparam=spi=on

  Or should I change something here?

  Thanx and regards

  Edited by sirvolcano January 06, 2019 17:19
• 0
  
  HiFiBerry team January 06, 2019 17:25

  Yes, the Amp2 can provide the additional 400mA. The configuration look ok. It's important that the display is NOT connected to the Raspberry Pi I2C GPIO.

  Best regards,
  Daniel

• 0
  
  sirvolcano January 12, 2019 16:43

  Hi everybody,

  latest news from my issue with the touchscreen: It is NOT A SOFTWARE ISSUE!!!!

  This post is a bit long and a kind of a report – be patient and read it 😉.

  It is somehow a hardware issue which I can’t explain at the moment. As well it is to mention that I have found the solution for me a little bit coincidently – but still have some question marks in my head! 

  What I have done since my last post:

  I got a multimeter from a friend and measured the touchscreen support directly at the 5V pin to the GND of amp2. (And promptly I destroyed my new – even not one-month old pi 3b+ while measuring because I slipped with the tip of the contact and build a bridge between the 5V and 3V pin, bloody hell ☹ and all this because a very trivial issue but luckily only the pi is destroyed other components are still fine).

  Well before that mishap I measured about 5.06V at the pin. Then I had to change the pi and installed my pi 3b and measured again – very carefully 😉 – and measured again 5.06V, so power supply at the pin is ok. Next step – I connected the jumper cable for the touchscreen support to the 5V and GND pins on amp2 and measured at the end of the cables which will be connected to the display board – here as well 5.06V, so still enough power there. During all this the screen faced down to the table to get good access to the pins. Then I took all the construct with display, pi and amp2 in my hand – this is the coincidently moment – and tried out to navigate with the touch function and Look! It works perfect even the amp2 was attached! Then I played some music and it worked as well and I could even use the touch function at the same time while the music was playing without any problems.

  Maybe one more thing that everybody should know – I assembled some brackets to the chassis of the display at the back to put it straight upward for a comfort use and look. But unfortunately, the free space between the construct and table is a little bit small and at the same time the power supply connector of amp2 faces towards the table and the barrel-connector of the power supply is a bit long, so when it is connected to amp2 the space is even smaller – too small to put it on the bracket’s feet. Therefore, I positioned the whole construct somehow in the box of the display with the foam inside for having as less bending on the supply cable as possible and this is one of the positions when the touch doesn’t work.

  Nevertheless, I measured the 5V pin in this position as well and power supply was ok – 5.06V but touch function poor, so it can’t be a power supply issue. With this finding I took the construct in the hand again, hold it at the bracket and bended the power supply in many directions – small bends, big bends of the cable, jiggled and shifted directly the barrel-connector – and tried at the same time to navigate with the touch and there was no problem with the touch function, it worked perfectly. Then I tried jacking up the brackets at its feet to get more space between the construct and the table, so that the cable can hang little bit more free and straight. Subsequently I put the construction on the table and not in the box again. Result – bad touch reaction. In this position I measured the power again, still 5.06V. …???... What is the problem? I don’t know! Doesn’t matter in which position and constellation only the cables and power supply are connected in a right way and firmly and there is no other contact of the components with anything else so that we can exclude any possible short circuits or something similar.

  Someone told me to shield the ribbon cable between screen board and pi by wrapping the ribbon cable with some aluminium foil, which I did already before (and which didn’t help). Based on that ideas I put the supply cable and loudspeaker cable and with other parts and cables closer together to see if there are any EMC issues – still good touch reaction when I hold it in my hand, so that can’t be the problem either!

  When there is the problem with the bad or no touch reaction – sometimes while trying the mouse arrow appears in the upper left corner of the screen but doesn’t move. So, I guess sometimes the touch reaction is been gathered by the system but not used.

  But here is another phenomenon – physically I keep the construct in that position in my hand in which it works perfectly as described above and don’t change. Then I change only the config.txt to rotate the display 180° and reboot. After reboot I see everything upside-down in the display and try to navigate with touch – nothing happens, so no touch function even the supply cable is in that position in that it worked before rotating the display via change of config!

  So, hopefully now everybody can understand why I still have question marks in my head and can’t explain the behaviour! But fortunately, and/or hopefully I found now my solution – after I will be done all my implementation – I would install everything in that way as I would hold it in my hand and hope that it works!

  But this is pitiful and pelting and keeps still some doubts that it will work in the end when everything is ready, and I put it in a nice box. I would like to exclude the doubts and be sure that it will work fine when I am ready with everything!

  So, if anyone have an idea what the problem might cause – go ahead and tell me! 😉

  Best regards

• 0
  
  sirvolcano January 12, 2019 16:52

  If someone wants, I could make some Fotos of the whole construct and post it here!

  Edited by sirvolcano January 12, 2019 16:53
• 0
  
  sirvolcano February 09, 2019 18:22

  Hello Hifiberry-Community,

  Reason of my problem is found and solved!!!

  Due to my holidays were end at the beginning of the year, I couldn’t work that much at my project, but nevertheless now I found the reason of my hardware problem:

  IT IS THE POWER SUPPLY AND IT IS A GROUND PROBLEM!!!

  Already I reported that it is not a software issue as I thought at the beginning and I started to search on the hardware. As I realized that the touch function is perfect when I touch the chassis of the display at the same time, I come up with an idea that it might be a grounding problem.

  Then I started to measure from the pins over the board up to the power supply plug barrel. And look – there is the reason of the problem – the power supply. The power supply is grounded at the input but the output is not!!!

  The voltage measurement between the minus pole of the output to the GND shows a voltage of ≈115VAC. Then I heard and read that it is common that the output of such a power supply is not grounded and such a high voltage is allowed as well, as long as the current doesn’t exceed 0,5mA. But this is a problem for capacitive touch displays.

  Then I checked the power supply of my notebook (which has a similar output: 19V/3,24A) and realized that the output of it is grounded! Then I connected it to my project and look – it works perfectly!!!

  I bought the power supply (Mean Well: GST60A18-P1J) as recommended from Hifiberry and expected that this kind of issues are already checked, therefore I didn’t make any mind about a problem like this and this cost me a brand-new pi 3b+ and many … many lost times.

  So dear Hifiberry-Team, you might or should change your recommendation on your website. In general, the recommended power supply works but not in the combination with a capacitive touch display – I had no problems with the sound or with pi. Well at least there should be a hint that this recommended power supply causes problems when it is used with a capacitive touch display! This would help many other people who have in mind to start a similar project with a similar combination!

  Best regards