What are the constraints for multiple (that is, maximum), simultaneous audio streams using the CC85XX?

Hi David, 

First of all, compliments to you for a thorough job reading and understand all the documents. It seems that you have gained a good understanding of our offering. 

The following from the UG is an error we will update in the next revision:  "The protocol supports a star network topology with a single protocol master node and up to seven protocol slave nodes.". It should really read four protocol slave nodes as it does on page 8.

So to your questions:

1. Everything that is actively transmitting in the 2.4 GHz band can in effect cause impact on the PurePath Wireless audio link. To what extent and how serious this depends really on the type of interferer. Basically you can divide interferers into two groups, static and frequency hopping. For the static interferers like WiFi and uOwens etc. the built in adaptive frequency hopping algorithm of PurePath works rather well and, depending on the transmit duty-cycle of the static interferer, we can detect these and move away in frequency. For frequency hopping systems (not talking about other PurePath Wireless networks) we have no chance determining the jumping pattern and thus it is little for us to do.

From your description it seems that you are worried about three interferers. a low-duty / idle cycled WiFi, Bluetooth and other PurePath Wireless networks. 

• Idle WiFi - if WiFi is idle during PurePath operation it should have little to no impact. With our adaptive frequency hopping we spread our 4 active channels as much as possible in the 2.4 GHz band and the trial channel iterates over the remaining 14 (out of 18 total) channels. Even with some in-frequent WiFi communication between node and access point, the chances of obstructing all 4 active channels is low. Built in re-transmission of our system ensures that we can survive loosing some packets (how many is as you say dependent on the configuration of the network).
• BT - Bluetooth is operating at 79 channels roughly with roughly 1 MHz occupied bandwidth for ~650us at each channel. PurePath wireless uses 18 channels (more active than the 14 other) with 3.8 MHz occupied bandwidth at ~2-4000us at each channel. Interference (both ways) is then just statistically chances of being at the same frequency at the same time. For a mute event on PurePath Wireless to occur you need multiple timeslots (at varoius frequencies) in a row to crash to mute which is even harder statistically.
• Other PPW - We are able to detect other PPW networks to some extent and have mechanisms to ensure that two independent networks won't align in time/frequency. 

Best regards, 

Kjetil

Kjetil:

Thanks much for the quick response and information.

Some followup on questions 2 and 3, with #3 first:

3. On the manufacturer's ID: if I understood your answer (and after a re-read of 2.2.1.1) I simply pick the device ID of any of the devices I purchased (really, TI's ID to mark that specific device), then based on the fact that it is globally unique, I can simply use it as my own manufacturer's ID. Is this correct?

2. On multiple PMs: the video was useful, though the system in the video was much smaller in scope than what I anticipate (with only a single stream/PM). Since I control the design of my "big" system, I can physically spread out the PMs as needed. But, this raises a couple of questions:

2a. If the links are designed to use the CC2590 (and this seems to be a good idea, given the fact that it is intended for use within a theaters/stages with a lot of steel in the buildings), how would any distances greater than the wavelength (about 4.1") really make any difference? Ideally, I'd like to have the 3 or 4 PMs in the same console (assume 6" apart), but I can place them, say, 1m apart. The video did not really address the idea of distance between PMs, so any additional info/hints/tips would be appreciated. Also in the video (since I don't have that eval board), what antenna were being used?

2b. On AFH: from my read of the docs, it appears that the CC85XX chooses from among the 18 channels dynamically. If I know, in any specific system with multiple PMs, that certain frequencies (say, 12 of the 18) work well, is there any way to constrain an individual PM to use a specific subset of the channels? In this way, multiple PMs could coexist without knowledge of (or interference from) each other. I didn't see anything in the command set that would imply that this is possible, but I would offer this as a suggestion for the next firmware revision. I think it could work like this: under the control of an external system, a PM is set into "scan for best channels" mode and paired with a PS. The PS is then moved to the individual locations where the performers would typically be placed, and the PM scans through the channels. (We assume that most performers are not going to move very much, and some - keyboardists and drummers - can really move their position at all.) The quality metric for each channel would be downloaded from the PM for each PS/performer/instrument location. The sound engineer for the theater/stage system would then pick the four best (and, ideally) non-overlapping channels for each PM, based on the physical location of the PSs to which it will be paired. Then, as the last step, those hand-picked channels are uploaded to each PM, and those are the only channels it would use.

I'll be anxious to hear some feedback on the possibility of this idea (if it already works that way, I've obviously missed it in the docs!) in a future firmware update. (And, this does sound to me like something that is entirely a firmware update.)

Again, thank much for the information. The design is proceeding.

Dave

Kjetil:

Thanks very much, this is the information I needed (especially on 2b - will be looking forward to this).

One last followup and I am probably complete with this thread: I have tested the CC85XX eval boards and am very pleased with their operation. As a digital designer, I'm perfectly comfortable with the design and implementation of the circuit. However, I have no RF experience and as I think about the packaging of these devices for performers (mic, instrument, etc.), is there a preferred mechanical orientation of the master relative to the slave? That is, what it the 3 dimensional "shape" of the signal from the CC85XX eval board? Is it a sphere, disk/donut, cylinder, conical beam, etc? In the case of the disk/donut, cylinder and beam, how is the axis of each oriented relative to the PC board itself? Is it normal to the board, or is the axis parallel to the board? If parallel, which direction? Or, is the signal dispersion broad enough that it really doesn't matter?

Thanks.