audio dropout

Hi Eric, 

Seems from the other post that this is a faulty board issue. 

You should not need to split the available channels for the two masters in this case. As 3 out of the 4 slaves works it can't be related to poor RF environment (both slaves to a given master would be affected, especially at this distance). 

In your application you can't reduce to 2Mbps as this is intended for masters that have maximum 2 channels (you have 4) and that only works with one slave (you have two). 

One recommendation (or at least something to try) is to use timeslot aligned masters if they are always in close proximity to each other. This will limit the affect the two PPW networks have on each other. 

Let us know if you run into other issues. Generally you should experience similar range for the two connected slaves. 

Regards, 
Kjetil

Well my first four person test did not go that great.  People can hear and talk to each other but there is a lot of packet loss so all of a sudden you can't hear one person's mic or you just hear them break through a little.  Pretty frustrating.

What can I do to help get rid of these drop outs?   I would say the long periods of time where you can't hear someone is the biggest issue as that makes the user think it's off or broken.   After that the drop outs are the most frustrating.

I'm open to compromising on sample rates to get improved robustness, or well I guess even if I have to add two more masters or whatever can be done.

Could you give me just a little more information on how to setup time slot aligned masters as you suggest?  I was getting ready to demo to our customer but after this test I've got to improve things first.

Thank you,

-Eric

Ok I think I get it select one as timing master, one as timing slave, run a wire from one gpio on timing master to one gpio on timing slave?

Correct. This will only help if the masters are really close to each other though (a few inches). 

As for the drop-outs, there are some things that influence robustness and co-existence. I'll cut'n'paste from another thread here. 

There are things that can be done to improve the audio link performance, but usually associated with a 'cost':

• Always keep the number of slaves to the lowest possible for a given application. In a multi-cast protocol, time is set aside for each possible slave for potential re-transmissions. If only 1 slave is needed, robustness will hurt if you enable more slaves to join the master
• The lower sample rates allows more re-tranmission attempts for the same latency and thus is more robust to drop-outs
• Latency is a direct trade-off vs. robustness. One are directly trading retransmission attempts for the lower latency. At the lowest settings this goes towards zero retransmission attempts
• Number of audio channel supported. the more channels supported, the more bandwidth is needed over the air and the less robust the link will be. 
• Direction of audio. Typically audio from master to slave is better when having uni-directional audio. For bi-directional links they are equally robust., 
• The various streaming formats requires different band-width. SLAC is more robust than PCM16 format.
• Target output power and CC2590 range extender. Improves range at the cost of power.
• Audio fading can be disabled (under advanced settings) for slightly better re-transmission capabilities.
• RF data-rate. 2 Mbit settings is less vulnerable to RF multi-paths and improves sensitivity. This will have a positive affect on range. 
• Antenna diversity: If implemented on slaves, it will combat multi-path effects and improve the in-range performance.
• Changing or disabling LBT: http://e2e.ti.com/support/low_power_rf/f/382/t/326031.aspx

I did read that post in my hunt to make things better.  I see a lot of things saying audio will be more robust with less slaves/channels which makes me think maybe these don't work too well with anymore than one slave?  Am I just always going to have poor quality audio with this configuration?  If so I'll go use something else, I don't want to fight this only to find out it'll never work well.

Lower sample rate didn't make a difference, throwing out audio muting seems to have helped but it's subjective.  No real change from timing master.   

Will I have a more robust system if I use four masters with three as timing slaves?

In my settings I wasn't able to adjust down to 2Mbit, it said something about it being incompatible with current settings.

Is there a way to go to say 8K sample rate?

Thank you.

Hi Eric, 

All the tips are just things that will improve. It does not do any quantification to increase or improvement - it's also hard to give general guideline as "good enough" will differ from customer to customer, application to application and what kind of environment things are to run in. 

You can easily tweak a configuration to the point where the range/robustness is limited to a in-room experience. On the other side of the scale we have customers and products in the market claiming 200+ meters line of sight. 

There are no other tricks, like 8K sample rate, besides the things I wrote earlier today. 

Regards, 

Kjetil

Ok this is such a frustrating problem, maybe because I'm so close to this working but not sure how to get to root cause.  Here is what I have now.   If I leave the masters running, say for a day or two then I connect all four headsets, then one headset will have audio dropout issues on tx and rx.

This is always one of the audio headsets connected to the master who is providing the mclk to the other master.  All three other radios can communicate with each other flawlessly at this time.   My distance is only about 3 feet, but I've tried longer and standing right on top of the masters.

Now if I power cycle this master and the headsets then everything works fine again.   I think this is why I thought that swapping out the headset with the issues was helping because I was power cycling the master.

So any ideas on that?  I was curious if maybe it picked a new channel, but then I know that this setup was working a day or two ago.   Also it's always one of the slaves connected to this master the slaves on the other master who takes in an external mclk never have an issue in any of my trials.

Thank you,

-Eric