ANT Forum

How many ANT nodes should then be able to exist in close proximity if using 8Hz and 12Hz.

I have a issue where 12 close nodes work fine @ 8Hz but fail at 12Hz.
Is the number of nodes close to each other proportional.

(Sorry I prviously logged this as a Guest).

This thread has been moved to
http://www.thisisant.com/component/option,com_fireboard/Itemid,146/func,view/catid,25/id,1550/      

What is your network topology? Are all of your nodes independent, like in the example? The topology plays a part in your level of performance. What I am getting at here is that it may not actually be RF interference that is causing your problems, but limitations within your ANT devices.

As for how many independent 8 Hz or 12 Hz nodes that are acceptable, this is not something we have specifically tested.      

I dont know which example topology you are refering too.

However my topology is basically a head-to-tail linear type.
Each node with the exeption of the GW connectes to another node.

GW<->n1<->n2<->n3<->n4<->...<->n12<-

This forms the equivalent of a bi-directional point-to-point data link.
All nodes use the same channel definitions and frequency. Each node runs one master and one slave channel. The slave channels are configured as shared slave channels. Each Slave will connect to the first availible Master using the pairing bit. For this test each master will only allow 1 slave to connect thereby forcing a head-to-tail type topology.

During testing with all devices within each others RF range I have noticed that at 8Mhz I can have 12 nodes running very stable. At 12 Hz everything starts going wrong at about 8 nodes. Where ther are sync misses and the nodes restart so as to search again.

I use our own module with an atmga644pa @ 8Mhz and a Nodric nRf24AP2 ANT chip.

I am thinking that it may be due to my applications low loss tolerance as I restart the node after only 4 consecutive ANT sync misses. I will try using a higher value.      

I made a typo error above 8Mhz should be 8Hz for the ant period.

I increased my applications Ant sync loss tolerance and now noticed that the slave never recovers once a sync error is reported.

My assumption then with regards my observations and test is that the increased activity due to the number of nodes at a higher rate increases the chance of there being some on-the-air collision after which the slave losses sync. This is ok butthe slave in my test only recovers by searching again.

I ruled out RF range. Each device was range tested to well over 30m open air. Each modules antennae centre frequency has been verified. The crystals are within spec.
I ruled out interferance as I monitored the air with a spectrum analyzer during the testing.

What could be causing the slave to lose sync without recovering?

Any suggestions would be welcome.      

Hi,

I qoute from
http://www.thisisant.com/component/option,com_fireboard/Itemid,146/func,view/catid,25/id,976/

"At the application layer, you could specify to close the channel after a certain number of Rx fails, but this is not built into ANT."

Why then is there a network event EVENT_RX_FAIL_GO_TO_SEARCH documented?

In my observations I have not yet seen this code being generated by Ant. I am using shared slave channels if that makes any difference.

In my investigations it seems as though once a sync error is reported the slave never recovers and never drops back to search. My application has to reset the Nordic and restart.
I have several questions with regards this.

1. If Ant does indeed drop back to searching then from what I read elsewhere it can happen after a second or two of sync misses depending on the Ant message rate - how many misses would be reported by a slave working at 12Hz before it would drop back to search?

2. If this happens, where a slave drops back to search, will Ant search for the previous master it was connected to auotmatically or will it search again from scratch based on its previous search configuration?

3. If it does indeed connect to the previous master will it still use its previous shared slave channel slave id?

The reason for all these questions is for me to try to understand why my slaves sometimes lose sync for no apparent reason.

Tks      

Please refer to AN11 - ANT Channel Search for more details on the the search mechanism. Note that the Event EVENT_RX_FAIL_GO_TO_SEARCH is not supported by all ANT parts - refer to the ANT Message Protocol and Usage document for details.

Is your channel period the same on both the master and slave on your channel configuration? Why are you using a shared channel? From your other post, it seems like your connections are one master to one slave.

Have you tried reproducing your setup with ANTware?      

Hi,

Thanks, I re-read the application note and conclude then that
24 misses (2 seconds at 12Hz) should be reported before dropping back to searching as per AN-11.

I am using the Nordic nRF24AP2 which should support the dropping back to search. However mine dont do that.I am probably doing something wrong but cant see it.

My channel periods are the same on both master and the slave.

In my other post I decribed my test scenario. In reality each master is configurable to support up to 4 slaves. The design is such that a basic mesh network is created. For my test the parameters are purposely set to allow only one slave per master.

Unfortunately I dont have ANTware.
Thanks.      

You can find ANTware on the Developer Zone on the website. We highly recommend using ANTware to evaluate the functionality of the protocol while you are working on the design on your system. It would also allow you to post logs with the full configuration sequence for your channels, and all messages when you see the problem.

Using ANTware with a setup of one master and one shared slave, you'll be able to confirm that the EVENT_RX_FAIL_GO_TO_SEARCH does appear in the AP2. Also note that unless you explicitly disabled search by setting the search timeouts to zero, it is not possible for the slave not to search. If you do not see this event, it may be because of issues in your serial driver implementation.      

I have resolved my issue and can confirm that the AP2 does indeed drop back to search after 2 seconds of consecutive sync misses as documented.
I have now furthered my testing and for others benefit I can answer my orginal questions.

1. If Ant does indeed drop back to searching then from what I read elsewhere it can happen after a second or two of sync misses depending on the Ant message rate - how many misses would be reported by a slave working at 12Hz before it would drop back to search?
ANS: 24 if at 12 Hz or 2 seconds worth of syncs for any rate > 2HZ

2. If this happens, where a slave drops back to search, will Ant search for the previous master it was connected to auotmatically or will it search again from scratch based on its previous search configuration?
Ans: The search after dropping due to excessive sync misses will be for the slaves current paired master. This a great feature to facilitate recovery of a link. I stand to be corrected but I could not find this documented.

3. If it does indeed connect to the previous master will it still use its previous shared slave channel slave id?
ANS: Yes. Also great feature. The Master must be transmitting/polling on the shared slave ID though else the search fails.

Now I will see how many devices I can connect at 12 Hz reliably.