0

ADAU1451 EEPROM Memory Failure

Duncan

June 13, 2018 14:34

None

Follow

Hi,

I got my long awaited BeoCreate amp and pi etc, connected everything up to a custom speaker and it has been 'working' relatively well so far.

1. I can get Airplay to work but when i close Spotify then the service needs to be manually restarted through SSH which is kind of annoying.
2. Bluetooth I don't generally use but it works
3. Optical I haven't tried
4. The Spotifyd service hasn't worked at all. I have followed all the forums here and done what I should but to no avail. I assume its a Spotifyd/repository problem as the Hifiberry & Volumio Spotify service also doesn't work.
5. Most concerning issue is that it seems there is a memory failure on the ADAU1451 chip or EEPROM. Details below:

• I downloaded the HifiBerry sample DSP program, made my X-Over settings and download to the Amp. No problems here, I can hear through the speaker my live setting changes etc so I know its communicating fine.
• I go over to write the compilation and all seems to go fine
• Click on the "Check last compilation through DSP" and it come back with a failure message (attached)
  
  
• I can play songs fine, DSP is working, sounds great!
• I issue a sudo reset to the pi and then when the system boots up (selfboot jumper on) I get a horrible white/pink noise coming through the speaker!https://www.dropbox.com/s/dfs382ii0ba8cbt/Speaker%20Noise.m4a?dl=0 
• I can connect to the pi again through SigmaStudio no problem and do the "link compile download" again which promptly stops the white noise and starts playing songs again and sounding good....

I assume this can only be a memory failure on the EEPROM side of things and the board needs to go back to the supplier?

Not sure if there is anything else that I can do or check?

Any assistance would be greatly appreciated!

• Facebook
• Twitter
• LinkedIn

25 comments

• 0
  
  HiFiBerry team June 13, 2018 14:40

  The DSP sometimes behaves a bit problematic with the SELFBOOT jumper set. Sometimes programming works, sometimes it doesn't. The EEPROM itself is fine, otherwise the program that is already programmed into the DSP wouldn't work and you would not hear anything (the DSP does nothing by itself without the EEPROM). Therefore a hardware problem is very unlikely.
  
  Remove the SELFBOOT jumper before writing the program to the DSP. After programming it, set the jumper again and reboot the system.
  The EEPROM verification doesn't seem to work reliably. We've seen errors with this even though the program was written correctly to the EEPROM.

  The spotify daemon had some bugs that should be fixed now. We expect to integrate this into a new Beocreate release soon.

  Best regards
  Daniel

   

• 0
  
  Duncan June 13, 2018 14:42

  I should note that when the system is turned off, speakers removed and the amp boots up, there is audible white/pink sound coming from the amp board itself!?! How is that possible!

  Once I re-download my DSP program all is well and the system behaves.

• 0
  
  Duncan June 13, 2018 14:46

  Hi Daniel,

   

  That was a quick response but not entirely sure what I should do. I have tried exactly as you say:

  1. removing jumper (while powered up)
  2. Compiling DSP
  3. writing DSP
  4. reinstaling jumper
  5. Plays fine at this point still
  6. sudo reboot
  7. White noise starts playing just after the 'ding' indicating the amp has started up
  8. reconnect Sigma & compile DSP again
  9. Plays fine again...
• 0
  
  HiFiBerry team June 13, 2018 14:55

  This is strange. With the jumper removed, programming should work fine. The noise seems to be from an incorrectly programmed DSP. The Beocreate overwrites the program during boot if you haven't selected the customized profile.

   

  Try the following:

  - remove SELFBOOT
  - program DSP
  - set SELFBOOT
  - run the following commands:
  
  gpio mode 0 out
  gpio write 0 1
  sleep 1
  gpio write 0 0
  gpio mode 0 in
  
  What happens then?
  
  Best regards,
  Daniel

• 0
  
  Duncan June 13, 2018 15:23

  In terms of making noises, there are very faint 'popping' noises when something is written (i think this is normal sound, have heard it before)

  

  Other than that noting really happened expect after restarting the airplay service it still wouldn't connect so I cant check if it plays music.

  I issued a sudo reboot, I heard the 'ding-dong', then very loudly I heard the lady speaking the IP adresss but no white noise! Still cant connect via airplay now which is very odd...

  Bluetooth works fine but the DSP I assume is VERY wrong as it sounds terrible with lots of distortion.

  I unplugged to avoid damaging the speakers, rebooted and the audio seems better when announcing IP. Still no Airplay...

  Bluetooth connects now fine, audio sounds ok.

  Edited by Duncan June 13, 2018 15:24
• 0
  
  Duncan June 13, 2018 15:56

  Oh, I remember now.

  When bluetooth is connected it seems to kill the wifi connection. When bluetooth is disconnected then I can connect again via SSH & SigmaStudio.

  Unfortunately whatever has happened now has killed the ability to airplay. Even resetting the iPhone hasn't worked ;)

  I have resorted now to adding a white noise source in SigmaStudio which works to test audio.

  I issued a sudo shutdown, unplugged, waited a few minutes and then plugged back in.

  Reprogrammed the DSP according to required x-over etc (with selfboot jumper off) and then replaced selfboot & issued an SSH reset.

  Seems to be working however I'm having problems with airplay & bluetooth is very soft. Will fiddle a bit more and see what happens. Very glad that it doesnt seem to be a hardware issue!

• 0
  
  Duncan June 14, 2018 10:43

  Hi Daniel,

  1. Can you explain what the commands were that you gave above? I assume some kind of hardware reset?
  2. Any idea when a new software revision is coming out that will hopefully get rid of the airplay/spotify/dropped network issues?

  Thanks

• 0
  
  HiFiBerry team June 15, 2018 04:46

  Hi,

  the commands above just send a reset to the chip making it boot from the EEPROM-

  A new image will be released soon, but I can't commit a specific date. Expect it in 1-2 weeks.

  Best regards,
  Daniel

• 0
  
  Duncan June 17, 2018 09:33

  Hi,

  I am at a loss now. Nothing that I do results in a clean boot with the ability to play music without a login to the DSP and a reprogram. The moment I turn the amp off and on (using sudo shutdown or plug out of amp or plug off at wall) with the selfboot jumper on...

  The DSP program is the stock unedited downloaded from HifiBerry & just trying to download that to the DSP.

  The reprogramming process is as per above and on other posts.websites + a combination thereof and nothing seems to work

  There is a significant speaker pop when I plug in the power to the amp, normal?

  Tried a clean install & doesnt solve the programming issues...

  Any last ideas before I send the board back to the supplier?

  Thanks, Duncan

  Duncan

   

• 0
  
  HiFiBerry team June 17, 2018 09:45

  Are you using the Beocreate image? If yes, you need to select the option to use an individual profile. Otherwise it will always overwrite the program that you programmed into the DSP. 

  When reseting the DSP with the commands above, the DSP program was ok - correct? In this case, the hardware works perfectly fine. 

  Best regards,
  Daniel

• 0
  
  Duncan June 17, 2018 11:09

  Hi,

  Yes its set to "basic full-range" in the web interface prior to doing any dsp programming. I have reset this again now and it seems to startup and restart ok for now (selfboot jumper was on BTW). I will refrain from doing programming of the DSP for a few days and see if anything changes. It seems something odd is happening during DSP programming that causes failure/corruption in my case.

  Yes, the GPIO commands worked.

  Is the stock free-range speaker profile available to download as it sounds quite different to the stock HifiBerry download for sigmastudio?

   

  Thank you

  Duncan

• 0
  
  Duncan July 17, 2018 19:11

  Hi Daniel,

  I managed to get what I thought was a good DSP running, i followed all the programming steps at outlined above including the reset steps while the amp was powered up using the GPIO commands. All seemed fine, and its been working relatively well up to now.

  Our power failed recently for the first time in ages and about 10 seconds after power returned the speakers let out an almighty buzzing sound which I assume is the DSP being corrupt or something.

  So it seems the DSP keeps its program in tact when there is power on the PI/AMP and the moment the power falls away the DSP loses its program or becomes corrupt.

  Is this normal? Is it something that might be fixed in the new TCP/Sigma programmer? I'm stuck, any things to check?

   

  Duncan

   

• 0
  
  HiFiBerry team July 17, 2018 19:31

  Hi,

  the current TCPDaemon has some limitations with longer DSP programs. We are working on a new version that will be integrated in a new release in the future. This will also have a feature to check that the EEPROM has been programmed correctly. 

  Do you feel comfortable to do some Linux command line stuff? In this case, you could replace the old server by the new one, but it will need editing some files and installing some software.

  Best regards,
  Daniel 

• 0
  
  Duncan July 18, 2018 05:34

  Hi,

  I don't understand why the DSP would fail, because of being a long program, after a power cycle.... Surely if it writes and saves the program and can survive a reboot it should survive a power cycle??? Something during the power cycle wipes/corrupts the eeprom memory?

  No problem to run some CLI stuff but will probably need some assistance here and there if its complex. As long as it doesn't slow you down to get the new version out ;)

   

• 0
  
  HiFiBerry team July 18, 2018 19:05

  My guess it that it doesn't write it correctly to the EEPROM. In this case, the program would run fine as long as the board still has power.

  Give me a bit time to prepare a new Raspbian with the new server. We're working on some cool new things, but this means that the code is changing quite a lot at the moment.

  Best regards,
  Daniel

• 0
  
  Gilbert July 17, 2020 20:38

  I am facing the same problem with the EEPROM. Everything works fine until I power off the Beocreate. I tried it with the selfboot jumper attached while programming, without it and attaching it after powering off, it makes no difference. I'm using the Beocreate Universal v10 project with a crossover and hardware volume control, but even with the standard Beocreate Universal v10 it doesn't store the program.

  The problem is discussed by another user in the analog devices Q&A and also there was no solution found:

  https://ez.analog.com/dsp/sigmadsp/f/q-a/163998/adau1452-selfboot-problem

• 0
  
  HiFiBerry team July 18, 2020 05:30

  How to you program the DSP? Using dsptoolkit or via SigmaStudio?

• 0
  
  Gilbert July 18, 2020 05:49

  With Sigma Studio. I tried also to use the standard Beocreate Universal v10, which was then recognized by the HifiberryOS, but after shutdown it was also gone.

• 0
  
  Gilbert July 18, 2020 18:43

  I just tried to upload the Beocreate Universal v10 program and verify it with the "Check last configuration through DSP" function. SigmaStudio gave me following exception:

  

  Hope this might help in solving the problem.

• 0
  
  HiFiBerry team July 21, 2020 09:15

  The "check" stuff in SigmaStudio isn't really working. Does deploying a pre-configured problem from command line using DSPToolkit work?

• 0
  
  Gilbert July 21, 2020 20:15

  Tried today, doesn't work either.

• 0
  
  Gilbert July 22, 2020 20:56

  So today I tried a new SD card with the latest HifiberryOS and another RPi 4. I unplugged every accessory (Rotary Controller and RPi official touch display). HifiberryOS didn't recognize the Beocreate, therefore I tried it with the /etc/hifiberry/has_dsp hack. Then it said "There is a problem accessing the sound processor". When installing the Beocreate Universal Program in HifiberryOS I just received a failure message. Then I checked the status of the core in SigmaStudio and following failure occured, similar to that in the inital post of this thread.

  Is there something wrong with my beocreate?

• 0
  
  HiFiBerry team July 23, 2020 06:13

  It might be a hardware problem. I'll open a ticket for this. 

• 0
  
  Peter August 03, 2020 03:08

  I have a very similar problem to this. I am trying to use the spdif/Toslink as the main input. I am using RPi3 with Raspberry Pi OS and communicating to it over TCP/IP from Sigmastudio 4.5. The self boot jumper has been removed. I have tried to run a very simple program of my own to play the unfiltered Toslink signal (16bit, 44.1kHz) to one of the onboard amps. I have also tried the ReCreate-Basic-042019.dspproj file from GitHub. Both result in the same core_control error flag that Gilbert is seeing. I don't yet understand the implication of this error. The signal path from Toslink to TPA3128D is working sometimes, but not always reliably. Also, Sigmastudio seems to get disconnected from the target quite often although the network link between the PC and Raspberry PI is stable and I can keep a solid link between them via ssh.

   

• 0
  
  HiFiBerry team August 03, 2020 05:39

  You can ignore the flags. Reading back data to SigmaStudio doesn't work realiable. It can even create problems as some registers might change without you noticiing it. Therefore, we recommend NOT to read back data from the DSP.