CC85XXDK - Bi-Directional Demo Audio Noise; FW1.3

Hi Geoff,

This is due to the fact that you are using the application role "Headset" for the slave. Let me explain:

In the audio device definition file for the aic3101 (aic3101.ppwadd located in the audio_devices folder) we have certain filters. One of these filters is called mic_input, and whenever the application roles Slave/Headset, Slave/Mono headset, Slave/Mono microphone or Slave/Stereo microphone are chosen this filter is enabled meaning that we get different register settings in the aic3101 than for example your master image which is using the application role Headset base station. The reason we have done it this way is because the four application roles mentioned above are all using the microphone channels e.g. Microphone 0 and Microphone 1, and on the PurePath Wireless Audio EB we enable the 3.5mm jack as the input for these application roles. Try your scenario using the 3.5mm jack on the slave, and see if the problem is still present.

If you want to use the RCA connectors on the slave as well please let me know and I can guide you.

Regards

Kristoffer

HI,

we have same problem.

Audio Noise is same of Geoff audio noise.

We have follow configuration : slave with role to master 2 audio channels (Stereo microphone). Master is base for microphone.

Analyze spectrum noise we have a component 300 Hz and secondary harmonic (600Hz, 1200Hz, 2400Hz).

Hi seely,

Could you post your configuration files? Master, slave and project...

-Kristoffer

HI Kristoffer,

in attachment project and noise registration,

Mic is inserted in slave and the noise is received from master.

Thanks

Bye

Hello Kristoffer,

After uninstalling and installing the FW1.3.0.36447 Configurator application, I repeated the tests I reported on May 22, 2012, using unmodified 'Demo Bi-Directional' configuration files seen attached, with no significant difference concerning the reported noise except that it was missing the +20dB amplification I used for the original tests.

See attached sound files 5-M-Sig + MF noise and 6-M-No Sig + MF noise, (corresponding to sound files '1-...' and 2-...' of the previous tests) which were conducted inputing around 1 mV RMS of 1 kHz sine wave signal into Line Inputs of the Slave.

As you suggested in your post of May 22, 2012, I tried input into the 3.5 mm Slave input using a signal generator and also using an electorate mic with voice.  The MF noise out of the slave is identical to the tests using Line Inputs into the Slave.

I am becoming increasingly disturbed by these results and those reported by at least one other Forum contributor as to whether the present firmware can deliver what I am looking for (viz, bi-directional voice transmission with a performance like the existing M to S achieves).

Can you please advise me on what I should do to achieve my aim.

Regards,

Geoff Pickford

Hi all,

Just to make sure we are on the same page. I have attached two zips:

bi_directional_stereo_RCA.zip - This is a project for bi-directional stereo audio streaming (two audio channels in each direction) using PurePath Wireless Audio EBs with CC85XX_CC2590EMs. The RCA connectors should be used on both the master and the slave.

bi_directional_mono_RCA.zip - This is a project for bi-directional mono audio streaming (one audio channel in each direction) using PurePath Wireless Audio EBs with CC85XX_CC2590EMs. The left RCA connector should be used on both the master and the slave.

Fixed offset in the input volume is 20dB on both the master and slave.

Try to connect an MP3/CD player to the input using a 3.5mm jack to RCA cable. The reason I say this is because it's important that the audio source is not sharing the same ground as the Audio EBs. You should for example not use a computer as the audio source when you are powering the Audio EBs with the same computer via the USB connector. This will create ground loops which will introduce noise due to the periodic current consumption of the radio when it goes in and out of TX/RX. Try listen to the audio using headphones in the headphone connector on the Audio EB. You should then experience that no noise is present in either direction, and see that the reported issues in this thread are not due to firmware running on the CC85xx devices.

Please report back when you have tested this, and we can continue the discussion.

Best regards

-Kristoffer

Here's the mono files...

Hi Kristoffer,

we have tried to load your project in headset-sdk,

but master and slave not start! We have tested also to load hex files but

same results.

We can't test your project. Sorry.

We have tested our project (noise) if source audio is a pc o mp3 player the noise is covered by music, but with microphone in a room

the noise is audible,

Bye

Hi Kristoffer,

I am out of Australia for a few weeks from tomorrow and will try your files as soon as I return.

In the meantime, you used the word 'filter' in your post of May 22, and I am puzzled by this term because it does not seem present or relevant.

Could you please explain exactly what you mean by this as I would like to gain a deeper understanding of the hardware and software.

In reference to your post of May 31, my audio source is a battery powered signal generator (or electret microphone), and the EBs are all powered by batteries, with no connection to each other or a computer.  Is your comment therefore still relevant for this situation?

Am I correct in saying that on your boards, you do not experience the noise that I and others are experiencing?

Kind Regards,

Geoff Pickford

Hi Geoff,

Regarding the ‘filter’ I mentioned.. My point is that based on what application role you select, different codec settings are used. You can see this if you open the aic3101.ppwadd. On line 57-59 you find the following:

<cfg_sequence_filter_spec name="mic_input">

    <application_role_list>Slave/Headset,Slave/Mono headset,Slave/Mono microphone,Slave/Stereo microphone</application_role_list>

</cfg_sequence_filter_spec>

So if Headset, Mono headset, Mono mic or Stereo mic is selected we will have what we call mic_input. Further down in the ppwadd file you find the following

-{mic_input

w 30 13 80    # Fully-differential line in L 1, gain 0 db, left ADC unpowered

w 30 16 80    # Fully-differential line in R 1, gain 0 db, right ADC unpowered

-}

+{mic_input

w 30 11 0F    # Single-ended MIC2 in L, gain 0 db

w 30 12 F0    # Single-ended MIC2 in R, gain 0 db

+}

So if mic_input is enabled we enable the single ended MIC2 input (connected to the 3.5 mic connector on the PPW Audio EB) instead of the differential line in input (connected to the left and right RCA connectors on the PPW Audio EB).

I still want you to test the setup I suggested, with a CD/MP3 player instead of your mic; I want to confirm that you get a noise free link from slave to master. If you then replace the CD/MP3 player with your microphone, and the noise returns, we have found the source to the problem.

You are correct, I am not experiencing the noise you and others are experiencing. With RCA connectors it is dead silent, even when adding fixed offset. When using single ended 3.5 jacks, and 15-20 dB added offset I get some noise, but not the same amount as in the audio files presented here in this thread. This is due the nature of a single ended versus differential signal. Noise will simply be added to a single ended signal, and with increased gain this noise will be increased. With a differential signal however the same noise will be added to both the positive and negative signal and therefor eliminated at the receiving end.

Hi See Seely,

The files I posted were configurations for the CC85XXDK and cannot be used together with the CC85XXDK-HEADSET. But can you please compare the following in your own project:

- The noise you experience with your microphone connected.

- The noise you experience with a CD/MP3 player connected (pause the music, so that you can hear the noise, if any).

Best regards

Kristoffer

HI Kristoffer,

I have already posted information for your question:

Kristoffer S said:

- The noise you experience with your microphone connected.

YES only with microphone. Exactly i have mic in slave.

Kristoffer S said:

The noise you experience with a CD/MP3 player connected (pause the music, so that you can hear the noise, if any).

Music cover the noise. The noise is present without other source or if level input source if less or equal to noise level. The input signal level of Cd o mp3 player is more high respect noise input since not ear noise but analyze spectrum of received signal i see presence to noise. I have already posted about spectral component of noise.

We use 3.5mm jack and use slave and master with battery.

We think is a EMI.

Bye

Hi Kristoffer,

I have used the stereo RCA example you provided on May 31, and have done exactly as you suggested.  I am using (and have always used) individual battery power, and MP3 player that have no common earth.   There is no audible noise from M to S, but still the medium frequency (MF) noise from S to M.  This was replicated many times using 2 out of a pool of 3 EVBs.

I have some questions as follows:

SECTION A

Using bi_directional_stereo-RCA example, I took the following actions:

1.  saved the project and configuration files using other names

2.  enabled 'custom setup' for M and S

3.  because 'audio device customisation' was blank, I 'reset all' for M and S

4.  made a small change to sequence, saved, quit, opened and change still present

5.  then 'Reset All', and the change disappeared

6.  QUESTION - Where is 'default' stored, and in what form.  Can this default file be changed and become the new 'default' for configuration files?

SECTION B

Using bi_directional_stereo-RCA example, I took the following actions:

1.  added mic bias and single ended mic to L & R of M and S as per attachment.

2.  M to S has resulted in extremely loud MF noise, even though mic and Line inputs work

3.  S to M has resulted in loud MF noise, even though mic and Line inputs work

4.  Grounds are very sensitive to touch.

5.  QUESTION - what is wrong, and can you please provide an example which uses mic inputs into M and S

SECTION C

I would like to further my understanding of the structure of files as used by the Configurator, and ask if you could correct my following statements:

1.  .PPWPRJ (using Doctype .DTD) calls upon .PPWCFG files which use Doctype DEVCFG.DTD; and calls on <name> of PPWARD file such as 'Master/Stereo Microphone Base Station'.

2.  This PPWARD file calls on PPWADD file such as <name> Audio Device </name>, <value> AIC3101 </value>

3.  QUESTION 1 - Whilst the Configurator HELP files provide some guidance, I cannot find a clear statement of the structure of files and their relationships.  So, tying in with the 'default' question above, which files have priority and where are they stored.  For example, when I attach the project file and configuration files as I have done above, do they call upon external files (not present in the attachment) that may influence their behaviour.  Also, how is the general information in the PPWADD files used in the configuration files.  These may seem basic questions to you, but it is difficult for a non-TI person to use software that has substantial proprietary (and relatively unknown) content that influences user controlled software in an unknown manner.

4.  QUESTION 2 - Even though it appears that it is TI's concept to have an easy to operate Configurator that shortens development time, could TI please consider laying down a generalised design approach that lays down major design steps necessary for a person to develop an application which does not fit the examples already provided.

Many thanks in advance,

Geoff Pickford

Hi Geoff. 

Kristoffer is on vacation so I will try to step in here. 

I'd really like to get to the bottom of the noise issue you have, but lets try to get the new questions out of the way first and then take it from there. 

Section A.
By default, all audio device drivers are stored as .ppwadd files under:

C:\Program Files\Texas Instruments\PurePath Wireless Configurator\audio_devices\

These are .xml files that can be edited directly, but we do recommend that you make a copy of the file and place it under before you edit:
<My Documents>\Texas Instruments\PurePath Wireless Configurator\audio_devices\

This will ensure that you will keep the file if you choose to install a newer version of the configurator at a later stage. 
Inside the file (open in any text editor, but one that understand and can highligh xml syntax might be helpful) you will find a <name> tag. 
At start-up the configurator will search both of the above folders for audio devices. If two files has the same <name> tag it will be listed with #name in the list of available options under the Audio Interface panel inside the configurator, the my documents version takes precedence over the texas instruments provided file. If a <name> tag exist under my documents and not the install folder it will be listed as +name

All this is documented in the configurator help system under 'Audio Device Definition Files'

You can either perform the edit inside the audio device customization panel like you have done already, but resetting will revert all changes. The other option is to do this directly in the .ppwadd file. 

Will get back to you with regards to section 3 later, but section 2 needs to wait until I have access to HW to test with and this will not be until early next week. 
Regards,

Kjetil 

Section C.

As you have seen, the configurator is to some extent based on xml files. 
- Application Roles (.ppward)
- Audio Device Drivers (.ppwadd)
- USB Descriptors (.ppwudd)
- In next FW1.4.0 also: HW Platforms (.ppwhpd)

These files makes the basis of a lot of choices within the configurator. They do not call on each other per se, but the configurator looks at the settings within files and can make things available or disable thigs as choices based on the content. If your selected application role contains only input channels (input means only receives wireless audio), the ADC3101 is no longer available as an option in the audio interface panel as it is a pure ADC and this application role requires a DAC or amplifier etc.

All these files can be edited by the user as long as they follow the xml 'syntax' according to the appropriate .dtd file - please note the disclaimer in the top of these files.

Missing a application role? Copy one of the existing one to <my documents>Texas Instruments\PurePath Wireless Configurator\application_roles and change it. 
Need to modify or add a external audio device? Add one to  <my documents>Texas Instruments\PurePath Wireless Configurator\audio_devices 

The other set of .xml files are for project files and configurations. These are in essence snapshot of a configuration at any time, more or less organized as the structure within the configurator (panel for panel). They do not call upon any of the other xml files but rather tells what application role, audio device etc. was selected at the time of the saving. If custom setup is not enabled inside the configurator, then the section within the .cfg file on audio device configuration (I2C sequences) is disregarded and the default ones from the audio device driver for the selected audio device is used instead. 

Question 1.
The files you posted allows me to see all the settings you have done in your configurator. If you had done any changes to your m_dual_headset_base_station.ppward I would not be able to see these. Similar, if you had edited your aic3101.ppwadd, I would get a different I2C sequence if I pressed 'Reset All' in the audio device customization (my aic3101.ppwadd file apply instead).

Question 2.
This is a valuable feedback to us and we're looking to see how we can improve and make things even more easier for our customers. You are not alone with this feedback and this is something we should look to describe better. How and when is to early to tell yet. 

Hope this clarifies things at least a little. 

Regards,
Kjetil
 

Hello Kjetil,

Thanks so much for such a fast response that I and my colleague are now studying.  For your information, I have not modified any of the PPWARD or PPWADD files.

The noise issue has been present for some time despite us spending significant developmental and testing time, and is certainly of great concern to us because we need to show our potential customers that each bi-directional audio transmission (either from M to S or S to M) is high quality AND with no audible noise under quiet listening conditions.

We look forward to your test results and a possible mic input bi-directional example.

Kind Regards,

Geoff Pickford

Hi Geoff. 

Was able to test your cfg files (without any modifications) today and am not able to re-create the noise you have. If the connector is 'floating' the noise is definitely present, but as soon as I attache a microphone or line-in source the noise is gone. Listen to the attached audio file as an example. This is S to M. 

Even though it works just fine, I would not have enabled line in L1 & R1 at the same time as the MIC2. Earlier Kris explained about filter in the ppwadd files. The AIC3101.ppwadd contains a mic_input filter. If you add your chosen application roles to this filter they will come up with Mic2 only:

<cfg_sequence_filter_spec name="mic_input">
<application_role_list>Slave/Headset,Slave/Mono headset,Slave/Mono microphone,Slave/Stereo microphone,Slave/Stereo bidirectional, Master/Dual headset base station</application_role_list>
</cfg_sequence_filter_spec>

My set-up looks like the below image. Please note that I used a battery for the M side as I connected this to my laptop to record the wav and did not want to introduce ground loops (as per Kristoffers earlier post). 

In my audio sample is simply plugged in/out the microphone.

Really not sure where to take this from here - to be able to help I really need to be able to reproduce the noise while the mic or line-in is present.
If very audible noise without a source is a big problem for you guys you could try to see if it can be catched with the new input silence detection we introduced in FW1.4.0 released just this week. Main problem with this is to calibrate a correct and good threshold for the noise that will work when mic is attached and no-one is talking. 
Using FW1.4.0 you can set the audio input silence detection in the Power Management panel. You need to enable this by selecting 'Stop transmitting samples if audio inputs are silent for..' and then set up a appropriate silence threshold.

Let me know if you need any guidance to the latter. The configurator help system do have some more guidelines on how this is done. Tested this quickly now and could find a threshold that will mute the noise but allow my microphone to work OK, but did not test this extensively. 

Regards,
Kjetil 

Hi,

Hello Kjetil,

Still away, but have had a chance to try and implement your suggestion of Jul 4.  I copied the AIC3103.PPWADD file, called it AIC3101_mic.PPWADD, and made the changes you had highlighted in red.

Then changed all references in the M and S .PPWCFG files from AIC3103.PPWADD to AIC3101_mic.PPWADD.

I note that my own PJT, and CFG files reside in My Documents/Texas Instruments/Projects, and that the AIC files are in Program Files/Texas Instruments/Pure Path Wireless Configurator/Audio Devices.  Presume this is correct.

When I restarted the PPWPRJ file in Configurator (FW1.3), the M and S showed major and non-correctable errors in Audio Interface, Audio Streaming, and Audio Customisation.

Would you mind sending me copies of your project, config, PPWADD files.

Thanks,

Geoff Pickford

Hi Geoff.

The audio posted earlier was recorded with FW1.3.0 and your configuration files as is without any modifications to any files. 
The problem you faced with the critical errors are because you now have two AIC3101 files in the same folder with the same <name> tag.
Copy your AIC3101_mic.PPWADD to your <My Documents>Texas Instruments\PurePath Wireless Configurator\audio_devices\ and it should work just fine.

When you open up the configurator again it should show a # sign next to the AIC3101 device name in the selection list under the External Audio Device. 
Alternatively you can change the <name> tag to AIC3101_mic or something similar. 

Attached a zip containing a modified AIC3101 file (changed names and added the changes i highlighted in red above). The project files and cfg files is for the new audio device definition with 'reset all' performed under the audio device customization. These files are also based on FW1.3.0 (believe there might have been minor modifications to the AIC3101.ppwadd file between FW1.3.0 and FW1.4.0) and with microphone connected I am not able to hear any noise.

Regards, 

Kjetil 

Hi Kjetil,

Without change, I used your files (post of Jul 9) with FW 1.3 and FW 1.4, with identical results.

See attachment, which includes 3 folders, one for a sine wave; one for FW 1.3 and one for FW 1.4.  

Sine Wave folder contains a Sinewave around 1 kHz, at -42 dBm (i.e. approx 10 mV RMS).  The original sound file (mp3), and a TIFF showing the spectrum (1024 bits, Fanning, Log frequency axis).  These data are provided as a reference signal for the noise tests.

FW 1.3 contains a sound file and spectrum of the noise at the S without any input into the M

FW 1.4 contains a sound file and spectrum of the noise at the S without any input into the M; and the same for noise at the M without any input into the S.

Several observations:-

1.  The subjective noise in each of these cases is unacceptably high

2.  You can see the components of noise, e.g. S is 364, 723, 1092, 1376, 1814, 25177, 2540 and 2907 Hz; with M as 720, 1089, 1449, 1813, 2179, 2906, 3265 Hz.  A number of these frequencies are the same for M and S.

3.  When the S is off, the M noise stops, and when the M is off, the S noise stops.

4.  When a 5 ohm resistor is inserted in series with the power supply to each board, and resulting CRO display shows the repeating sawtooth peaks approx 1.4 ms period, with every second peak around double the size (presumably one peak representing the Tx and one the Rx periods).  These periods correspond with the lowest frequencies for the S.  The peaks are modulated with approximately a rough sine wave with a period of many hundreds of ms.

These results are replicated for any pair of mother boards out of a total of 4 (all combinations), and also replicated wether the inverted F or the SMA whip antenna are used.  On all cases, the boards were powered with batteries of voltage 5 to 7 volts, and thus fully floating.  I measured the voltage marked 3.3 in several places and found it varied from 3.0 to 3.3 V depending upon the board.

The obvious question is - why are we both getting totally different results using the same FW and software, but different boards.  Also, why are several other forum members experiencing the same.

One way of assisting this problem's solution would be to post me out (on loan) a complete CC85XXDK, already programmed and known to perform properly.  Something radical has to be done because after almost a year, my Company is almost to the end of persevering with this project for bidirectional voice.  To that end, the address is Pickford Resources Pty Ltd, PO Box 618, Wollongong, NSW 2520, Australia.

Kind Regards,

Geoff Pickford

Hi Geoff, 

Have ordered a kit (am all out of extras) and will ship it to you pre-configured if we can not get to the bottom of this by the. I expect the boards to arrive here early next week. 

Did some new measurements today using FW1.4.0 with the attached configuration files. 
Recorded the headphone output on the master side (had similar result for slave) with no input at slave side (nothing connected to the DK).

Important to note that I turned the laptops microphone input gain to maximum here.

The first file is with the radio output power set to -10dBm on both sides of the link

I redid the measurement with the standard +5 dBm settings

For the attached configuration files the timeslot for the wireless protocol is 3000us. This means that every 3000us there is a power-pull due to the increase in CC85xx+CC2590 power consumption (turning the radio on). How much of a pull depends on the output power of both radios. The latest image shows the power-pulling at 1/3000=333 Hz and artifacts of this (within a timeslot the radio switches between Rx and Tx and have periods with radio off). 

With fixed 20dB input volume offset on the microphones, the noise floor will off course increase. 

How much different in levels are the noise vs. the wanted signals and have your tried the input silence detection to see if this can be useful for you?

Best regards, 
Kjetil 

Hello Kjetil,

Thank you for responding so fast and professionally. 

I forgot to mention that I am using a signal generator calibrated as dBu (i.e. 0 dBu = 0.775 V rms), and I think that Audacity uses dBFS, but not sure (even though we are somewhat lucky that 0 dBFS to roughly equal 0 dBu, which is the usual scale for audio levels).    I found my computer input volume was not quite at maximum, but have proof that in sum total the spectra I sent you (and the following figures) are within several dB of true dBu.

When summarising both of our latest results using FW 1.4 and the noise spectra:-

Source        100   300    800    3,000    10,000 Hz

Kjetil            -52     -54     -49       -72           -63   dBu

GP M          -69      -49     -46      -69           -87

GP S          -78      -75     -68      -84           -90

Whilst my M is considerably louder than the S, your M and my M are not all that different.  Please note that I cannot use an output power of -10 dBm, and must use either 0 or +5 dBm.

Your question re signal level is important, but not critical in terms of audible electronic noise.  I will be using inexpensive headphones or ear-buds of Sensitivity (S) between 90 and 100 dB SPL at 1 mW rms into a 32 ohm load.  The unit I am using for my tests has Sensitivity = 98 dB (which is fairly high), which means that a 300 mV P-P signal (i.e. 106 mV rms == -18 dBu) represents a power of 0.35 mW rms and outputs an SPL = 94 dB.

It also means that a 30 mV P-P signal (i.e. 10.6 mV rms == -38 dBu) is a power of 0.0035 mW rms and SPL = 74 dB.

So, I am needing a range of headphone volumes between approximately 74 dBA (SPL) when the background noise is around 40 to 70 dBA, and 94 dBA (SPL) when the background noise is up to around 90 dBA.  Whilst it is obvious that the higher background noise  situation requires an input signal of -18 dBu (which is comfortably louder than the electronic noise), our product is going to be also used in quiet conditions ( i.e. 40 to 70 dBA) where the electronic noise is obviously audible and completely unacceptable.  

Rather than all the previous measurements, a simple test can be used where if you can hear audible electronic noise (especially one that has pure tone elements and/or odd harmonics) using medium quality headphones  when testing in a typical office or laboratory environment, this is an unsatisfactory result.

Where should we go from here please.

Regards,

Geoff Pickford

Hi Kjetil,

I presume that my questions are difficult, but ask another in an attempt to solve my problems.

I have been trying to track down the noise, and believe that it is unlikely to be on any of the digital sides of the signal path, largely due to the nature of digital transmission, and also due to the low frequency set of harmonics that appear to be linked to Tx/Rx timing.

Therefore, it appears that it could be in any of these locations:

1.   codec input circuitry including mic bias for mic input, ground loops, and Tx/Rx current demands reflecting back onto the input

2.  codec internal problems up to point of ADC conversion

3.  codec output circuitry

When M is off, then S noise is not present, and vice versa.  This indicates that the noise is likely to be created in areas 1 or 2 above.

Could you please, as a matter of urgency, assist in tracking down this problem.  It may be that at the present state of firmware (or hardware) development, the CC85XX system is more suited for you original 'audio streaming of CD quality audio' - i.e. in one direction, and not bidirectional (albeit it only up to 8 kHz audio) as my project demands.

Could you also let me know if the kit you ordered for delivery to yourself early this week has arrived.  Maybe, if you are experiencing the same noise, it is not productive to send this to me?

By the way, the FW 1.4 option of using the input volume silence detection is not an option for me because the period of detection, there will be unacceptable noise present.  We really need to stop the noise at its source.

Kind Regards,

Geoff Pickford

Hi Geoff,

I have attached two pictures of a test I want you to perform, just to be 100% certain that we are on the same page. I find it really strange that such a basic test as the one we are talking about has different outcome. I don't think we should dive deep into other potential issues that may or may not be a problem before we have this basic test figured out.

In the attached puictures I have programmed the devices with the project I posted May 31st: "bi_directional_stereo_RCA.zip". One board is programmed as master and the other board is programmed as slave.

In the first picture I have plugged a mobile phone (Samsung galaxy S2) to the LINE IN RCA on the master, and I listen to the audio received by the slave by connecting a headphone to the headphone connector. With the mobile phone I play a song, and I turn the volume on the phone all the way down to zero. This way I transfer silence (makes it easier to hear any noise), and since the song is playing I know that the phone is properly grounded. If I pause the song it takes 3 seconds and a humming noise will be heard in the headphones. This is probably due to some timer in the Andriod OS that leaves the signals in the 3.5mm jack connector floating. But if the song is playing I hear no noise in the headphones.

I then test the other direction by connecting the phone to the slave's LINE IN RCA, and connect the headphones to the master's headphone connector. No noise is audible here either. As I have understood you will hear noise in the last situation, but not in the first. Is this correct?

Hi Kristoffer and Kjetil,

PART A

I have been able to eliminate the noise problem, even if I have not been able to find the root cause – it is highly likely it was due to a 'corrupted' aic3101.ppwadd file that my project and configuration files kept calling on.  

Frustrated with endless fruitless tests on various combinations of 4 mother boards and 8 RF boards, and despite re-installing the Configurator a number of times, I uninstalled the Configurator, AND deleted all TI files that were still present (an important lesson here because the Windows software removal tool does not remove all files), and installed FW 1.4 again.

I also loaded Kristoffer's May 31 "bi_directional_stereo_RCA.zip" file, unmodified, using battery power supplies, floating signal generator, and floating computer using Audacity as before.

Applying a 1 kHz test signal, either at -60 to -40 dBu produced excellent performance when using the RCA inputs of either M or S, and receiving at the S or M respectively, both using headphone outputs.  See typical spectrum below for M to S, at -60 dBu:

As you can see, the peak is -41.7 dB at 1,000 Hz, which lines up with the fact that the signal into M was at -60 dBu, and was amplified by +20 dB, which should have been -40 dB.  My signal is 1.7 dB lower because of signal generator setup.  More importantly, there is no evidence of any noise above -90 dBu - a superb result.

Similar results are for other volume levels except at -20 dBu (in reality around 0 dBu because of the 20 dB amplification) which shows some distortion and harmonics due to the stage being overdriven (see attachment) - not unexpected or unreasonable.

PART B

I then tried to take Kritsoffer's May 31 "bi_directional_stereo_RCA.zip" file, and see what the codec settings were, but could not because the Audio Interface Custom setup was disabled.  I found it not possible to use "reset', because this reset the configuration file to standard aic3101.ppwadd settings which did not work properly when programmed.

PART C

I then tried to start fro fresh by creating a new Project File, two new configuration files based on 'Bidirectional Audio', 'stereo headset base station' for M and 'Bidirectional Audio', 'Stereo Headset' for S.

When programmed, I was back to unacceptable noise, which was made worse when I added 20 dB to the Volume Control of M and S.

PART D

I then tried the 'Wireless Headset' example program, and found that the RCA inputs to M and mic input to S both produced superb results, e.g.:

Note the absence of noise, with a -40 dBu input signal (remember, no added gain).

PART E

I want to now use microphone inputs for both M and S, and tried to see what codec settings were present in the example configuration files, but ran into the same problem as experienced in PART B.

CONCLUSION

With the project and configuration files that you have provided, as well as the example files, the noise performance of the CC85XX DK is exceptional, but it is difficult to proceed further because the I am having difficulty determining the codec settings used in these cases.

QUESTIONS

1.  Can you please supply the key to determining the codec settings of the above files when they do not appear in the 'audio device customisation' when not already enabled?

2.  Also, to assist, would you please provide your May 31 "bi_directional_stereo_RCA.zip" file, with the 'audio device customisation' already enabled?

3.  Most importantly, would you please provide an example with bidirectional capability, with M and S both with microphone inputs.  At the same time provide codec settings as requested above?

4.  If easy to do, can you also please provide an example with bidirectional microphone inputs for a M and two S's, obviously only mono microphone to each of the slaves?

Many thanks again for providing this forum, and for your valued participation.

Kind Regards,

Geoff Pickford

Hi Geoff,

PART B

The audio device configurations used is indeed the standard ones. To see the settings simply enable the custom setup and press reset all.

PART C

"Stereo headset base station" for M will enable the RCA inputs as inputs. "Stereo headset" for S will enable the microphone connector (J8) as input. If you don't connect a properly grounded audio source to J8 on the slave side you will hear noise in the headphones on the master side.

PART D

This example should be similar to the one you try to create in C, so It's strange that you see different performance in C.

QUESTIONS

1. These following two scenarious will be 100% equivalent:

-disable custom setup of the audio device

-enable custom setup of the audio device and press reset all

2. I will add this as a zip in a later forum post. The name of the zip will be:

bi_directional_stereo_RCA_customSetupEnabled.zip

3. I will add this as a zip in a later forum post. The name of the zip will be:

bi_directional_mono_micInputs.zip

4. I will add this as a zip in a later forum post. The name of the zip will be:

bi_directional_2slaves_micInputs.zip

Since we have solved the basic issues you were facing I don't see the point of sending you the kit as we talked about.

Hope this clarifies things :-)

Regards

Kristoffer

Attached: bi_directional_stereo_RCA_customSetupEnabled.zip

Attached: bi_directional_mono_micInputs.zip

Attached: bi_directional_2slaves_micInputs.zip

Hi Kristoffer.

Have tested the three configurations, and all come out really good, very similar to the test results I posted yesterday.  We will now study them to make sure we can create other successful configurations.

I tried reseting the 'audio device customisation' on your 3 examples, and found two of them to do exactly as you said in your last post.

However, your 'bi_directional_2slaves_micInputs' did not.  The two lists below show only those registers changed:

Before Reset:

OFF to SR-SWITCH
w 30 2A 60 # Pop reduction, output drivers delayed 100 ms
w 30 11 0F # Single-ended MIC2 in L, gain 0 db
w 30 12 F0 # Single-ended MIC2 in R, gain 0 db

INACTIVE to LOW-POWER
w 30 19 80 # Enable mic bias 2.5V
w 30 13 FC # Left ADC powered
w 30 16 FC # Right ADC powered

LOW-POWER to ACTIVE
OK

ACTIVE to LOW-POWER
OK

LOW-POWER to INACTIVE
w 30 0F 80 # Mute left ADC PGA
w 30 10 80 # Mute right ADC PGA
w 30 19 00 # Disable mic bias
w 30 13 F8 # Left ADC unpowered
w 30 16 F8 # Right ADC unpowered

After Reset:

OFF to SR-SWITCH
w 30 13 80 # Fully-differential line in L 1, gain 0 db, left ADC unpowered
w 30 16 80 # Fully-differential line in R 1, gain 0 db, right ADC unpowered
w 30 2A 60 # Pop reduction, output drivers delayed 100 ms

INACTIVE to LOW-POWER
w 30 00 00 # Select register page 0
w 30 13 84 # Fully-differential line in L 1, gain 0 db, left ADC powered
w 30 16 84 # Fully-differential line in R 1, gain 0 db, left ADC powered

LOW-POWER to ACTIVE
OK

ACTIVE to LOW-POWER
OK

LOW-POWER to INACTIVE
w 30 13 80 # Fully-differential line in L 1, gain 0 db, left ADC unpowered
w 30 16 80 # Fully-differential line in R 1, gain 0 db, right ADC unpowered

In other words, resetting changed the registers from mic inputs to RCA inputs, despite the fact that I had previously completely deleted all TI files and Configurator, and re-installed.

I tried another computer (both running VMWare Fusion on Macs running OS 10.7.4), and the same configuration did the same thing.  Further, 'bi_directional_mono_micInputs' changed, but only the order of several of the registers as they appeared on the page, but not the contents.

I deleted all TI files and Configurator on the second Mac, and it then behaved like the first Mac.

QUESTION: Could you please tell me how this may have occurred, because to me it is an unknown factor that is likely to cause problems in the future?

Regards,

Geoff Pickford

Hi Geoff,

Great that the configurations worked well!

The reason that the codec settings for "bi_directional_2slaves_micInputs" changes when you press "Reset all" is because I have modified the sequences inside the Audio device customization panel to match your requirements. Let me explain:

When custom setup of the audio device is disabled the settings in the ppwadd file is used. When custom setup is enabled the settings inside the Audio device customization panel have presidence over the ppwadd file. If you press "Reset all" the configurator will read from the ppwadd file and copy these settings into the sequences in the Audio device customization panel.

A rather confusing thing (which will be fixed in the next configurator release) is the fact that many of the example projects have old (non-ideal) saved settings that will apply if custom setup of the audio device is enabled. For example if you open the example project "CC85XXDK - Wireless Headset" you will see that custom setup of the audio device is disabled for both M and S. So-far everything is ok, and this example works as it should. The settings in the ppwadd file is now used. But if you enable custom setup for the master, and then click the Audio device customization panel you will see old saved settings. These saved settings are part of the device configuration ("cc85xxdk_demo_analog_input_bi_directional_master.ppwexcfg") and has nothing to do with the ppwadd file. They have now presidence over the ppwadd file. If you now press "Reset all" you will see that the settings change, because "Reset all" reads the ppwadd file and copy these settings into the Audio device customization panel.

In the following projects bi_directional_stereo_RCA_customSetupEnabled and bi_directional_mono_micInputs I have enabled custom setup and pressed "Reset all" (because you requested me to) so that you could see the configurations. I could have just disabled the custom setup and it would have been equivalent. So why you see changes in Mac OS is very strange!! Are you certain that you used the same configurator versions on your windows machine and your Mac?

Regards

Kristoffer

Hello Kristoffer,

Thanks for your helpful description in your Aug 2 Post.  This and other information TI has provided has assisted me in creating successful Projects, albeit a bit laborious, and I would like to list the detailed description of this creation to see if it can be improved;

1.  Create new project - File, New Project, enter File Name, save (eg Bidir-M&2xS)
2. Create 1st Device Configuration - Create New.
3. Enter Configuration Name, (eg M); Set, enter File Name (eg M), save
4. Network Role - Select Master
5. Application Role - Select Stereo Bi-Directional Base Station; Select - Create
6. Audio Streaming, Master to Slave - Change Streaming Format to SLAC
7. Audio Streaming, Slave to Master - Change Streaming Format to SLAC; Change Audio IO to AD1
8. Create Device Identification - Create New, Enter Manufacturer ID (eg 11111111);  Enter Manufacturer Name (eg PR); Create Manufacture ID
9. Use stored Name; Enter Product ID; Enter Product Name; add to Product List
10. IO Mapping - Choose Audio Device Reset Control (eg GIO2); Choose Network Pairing Button (e.g. GIO3), Choose Network Status LED (e.g. GIO3)
11. save
12. Create 2nd Device Configuration - Create New
13. Enter Configuration Name, (eg S1); Set, enter File Name (eg S1), save 
14. Network Role - Select Slave
15. Application Role - Select Mono Headset; Select - Create
16. Audio Streaming, Master to Slave - Change Audio IO to AD1
17. Audio Streaming, Slave to Master -  Change to Microphone0
18. Create Device Identification - use ID from M
19. IO Mapping - Choose Audio Device Reset Control (eg GIO2); Choose Network Pairing Button (e.g. GIO3), Choose Network Status LED (e.g. GIO3)
20. save
21. Repeat for S2

Hi Geoff,

Question 1: Your process looks fine.

Question 2: You are right. If you want to make changes to an allready existing audio device definition, and you want this change to apply every time you press the reset all button, you must do this change within the audio device definition file (.ppwadd). However, instead of making changes directly in the files that comes together with the download of the configurator, we recommend to make a copy of the device definition file you want to change and save this copy in the following folder:

C:\Documents and Settings\<user>\My Documents\Texas Instruments\PurePath Wireless Configurator\audio_devices\

If the name tag (inside the definition) is identical to a name tag in an allready existing audio device definition the user generated audio device definition overrides the installed (original) one, and this will be marked with a # in the audio device list. For more info see the" Audio Device Definition Files" chapter in the configurator help system under the sub chapter "File Handling".

Two features in the configurator that might be useful to you:

"Copy existing..." - This is a button in the Project panel that is handy if you want to create a device configuration that will be only slightly different than an allready exixting device configuration.

"Import settings from other device configuration" - This button is located next to the "Save" button. Pressing this button will copy the current sub panel of an allready existing device configurations.

Best regards

Kristoffer

Hi Kristoffer.

Will do all you have suggested.

However, what if you want different Codec settings for different configurations in the one project, eg Line Input for M and Microphone inputs for S, or even more so, very different audio paths through the Codec for each of the 3 configurations.  It appears to me that the .ppwadd changes are global.  Is it possible to define this in the ,ppward file - which to me seems magically to call on specific Application Roles, each of which can be different, but not sure how to use it for my question.

Also, I would value comments on Question 3 (re noise) of my post of Aug 5.

Finally, do you have any PCB layouts of the CC85XX using a chip antenna, or even third party sources of these?  if not, having no experience with RF design, is it possible to use the existing PCB layout (used in the CC85XX DK) and simply replace the inverted F antenna with a chip - both seem to need a 50 ohm feed.

Many thanks,

Geoff Pickford

Geoff, 

My comments to item 3 - there will be power-pulling with frequency = 1/timeslot. The higher RF output power on the board the bigger the pull. This has to do with the fact that a CC85xx will draw ╠~10mA with the radio off and ~30mA with the radio on (standalong at ~+5 dBm outputpower). With the CC2590 the difference between off and on can be twice as much. Proper care with regards to layout have to be taken to prevent this from being picked up by the audio paths. For your setup that has a ADC at the slave side and DAC at the master side (or visa versa) this pulling can be picked up at both sides. 

We spent a significant amount of effort on this on both our standard DK and the DK-HEADSET to minimize the effect, but amplify the signal enough (input volume offset for the slave/ADC and PC/Recorded input gain for the master/DAC) and you will be able to find traces of this. Remember that to be able to see this for the plot i posted I had to turn the input gain in my laptop to maximum. Without doing this I was not able to detect anything. 

Layout techniques to minimize power pulling includes:
- splitting of ground planes (Audio and RF)
- star-routing of power (GND and VCC)
- separate power regulators (Audio and RF) with good power supply rejection ratio
- good decoupling placed as close to circuits as possible

Best regards,
Kjetil 

Kjetil,

It certainly appears that your team achieved a really good result in respect to noise in terms of circuit and OCB design.

As indicated in my previous post, my colleague and I spent a fair part of today trying to put Kristoffer's suggestions into practice - without success due to two main problems.  We had previously duplicated the aic3101.ppwadd file and a stereo and mono version of the .ppward files, and stored them as instructed inside the appropriated sub-directories in the 'My Documents' Directory.  These files were modified by changing the internal names etc so as to be consistent with the .ppwcfg files.

It did not matter what we did, but the Configurator did not recognise the new .ppward files, because they did not cause the .ppwcfg files to even specify that they were M or S files etc.

We then changed the duplicate aic3101.ppwadd file and found this also was not effective in passing on its information to the .ppwdcfg files.  Even small 'comments' changes were not passed on.

Worse than this, when we Reset All, the original version of the aic3101 file showed in the Customization panel, EVEN WHEN we deleted both the original and modified files.  In other words, there was no aic3101.ppwadd file in any directory.

QUESTION 1.  Any thoughts on getting the project to recognise modified .ppward files? For example, can you please list essential parts of these files that have to be consistent with other files such as .ppwcfg and .ppward files?

QUESTION 2.  Any ideas where the Configurator is finding the deleted aic3101.ppwadd files?  Same thing happens when we insert Default.  It almost appears as if there is another aic3101.ppwadd file somewhere embedded in the Configurator.

QUESTION 3.  (already posted) Do you have any PCB layouts of the CC85XX using a chip antenna, or even third party sources of these?  if not, having no experience with RF design, is it possible to use the existing PCB layout (used in the CC85XX DK) and simply replace the inverted F antenna with a chip - both seem to need a 50 ohm feed.

Thanks again for being so patient - I am hoping the frequency of my questions will soon drop to just about nothing.

Geoff Pickford

Hi Geoff,

QUESTION1:

I'm not sure if I understand your question or problem. Could you please elaborate on what you are trying to do? If it helps I have attached a modified ppward file that should be recognized by the configurator. Place this in the "application_roles" folder under My Documents, and you should see it under "MASTER - Upstream audio" in the configurator. The name will be "MODIFIED Stereo output base station". An important part of the <name> tag is the Slave/ or Master/ prefix in the start of the name.

<name>Master/MODIFIED Stereo output base station</name>

QUESTION 2:

The only place the configurator looks for .ppwadd-files is in the "audio_devices" folder in the install folder and the "audio_devices" folder in My Documents. I tried to remove the aic3101.ppwadd file from the install folder (I don't have any in my My documents folder), and as a result the example project "CC85XXDK - Preloaded Demo" gives an error message in the Audio Interface panel because no audio device is selected. The drop down list doesn't have an option for selecting aic3101. The only possible explanations I can come up with to explain why you still see aic3101 in the list are:

- Maybe you have (by a mistake) modified the name tag inside some of the other ppwadd-files. For example if you open your aic3204.ppwadd file in an editor you could maybe see that the name tag is aic3101.

- Maybe you are looking too fast in the drop down list and that you are mistaking adc3101 to be aic3101. I have done this a few times...

- Maybe you have deleted your aic3101 files while the configurator is running and expecting to see the impact of this delete immediately. You have to close and reopen the configurator in order to see the effect of deleting ppwadd-files.

The magic you are referring to is something we have talked about earlier. There are several filters in the ppwadd-files in order to set up different i2c sequences depending on the application role selected. For examplein the aic3101.ppwadd you find:

On line 57:

    <cfg_sequence_filter_spec name="mic_input">
        <application_role_list>Master/Mono headset,Slave/Mono headset,Slave/Stereo headset,Slave/Mono microphone,Slave/Stereo microphone</application_role_list>
    </cfg_sequence_filter_spec>

These lines have the following meaning: If some of the listed application roles are selected we have what we have called "mic_input". Further down in the file on line 82 we see:

        -{mic_input
        w 30 13 80    # Fully-differential line in L 1, gain 0 db, left ADC unpowered
        w 30 16 80    # Fully-differential line in R 1, gain 0 db, right ADC unpowered
        -}
        +{mic_input
        w 30 11 0F    # Single-ended MIC2 in L, gain 0 db
        w 30 12 F0    # Single-ended MIC2 in R, gain 0 db
        +}

This means that if the application role selected is one of the following:

Master/Mono headset,  Slave/Mono headset,  Slave/Stereo headset,  Slave/Mono microphone,  Slave/Stereo microphone

we have mic_input and the i2c sequence that will be applied is

w 30 11 0F    # Single-ended MIC2 in L, gain 0 db
w 30 12 F0    # Single-ended MIC2 in R, gain 0 db

If any other appliaction role is selected we don't have mic_input and the i2c sequence that will be applied is

w 30 13 80    # Fully-differential line in L 1, gain 0 db, left ADC unpowered
w 30 16 80    # Fully-differential line in R 1, gain 0 db, right ADC unpowered

QUESTION 3:

Unfortunatelly we don't have any ref designs using chip antennas. But simply copy TI's ref design all the way to the PCB antenna (you can skip the zero ohm and the placeholder for the shunt component), and continue with the ref design for the chip antenna (they often come with some recommended external components). Just make sure the antenna is a 50 ohm antenna and designed to operate in the 2.4 GHz band.