Zigbee brightness cluster attribute and cluster

[JorgenM]

Hello,

Using v1.9.1 nrfConnect and the Coordinator example on a nrf52840 -DK.

From Nordic docs:


I think I may be miss interpreting Zigbee Cluster spec...

Below is example of an RGB light bulb I'm trying to change brightness on but can't get it to change.
(I tried many different ways too long to list)

#1  Based on below cluster info etc. can you give example of ZCL commend format to set brightness to 50%?
We can use "move to level" to 50%. I believe range is 0x000 - 0xfe if I read it right?

I can turn same bulb on/off by sending on/off command. 
zcl cmd 0xcf3e 01 0x0006 0x01 --- ON, as example.

But I can't read the attribute showing on/off status? Am i missing some setting in Kconfig?
I tried including profile id 0x0104 also since my default is HA.

#2  Can you me show ZCL command example of how to read this?

From Nordic doc: 
zcl attr read h:dst_addr d:ep h:cluster [-c] h:profile h:attr_id

PS: I looked at other tickets but looks like my issue is much simpler.

Thank you, Jorgen

[AHaug]

Hi Jorgen,

Is the light bulb application based on a sample application or is the firmware written from scratch? And is the RGB bulb a specific device, or just a bulb that is ZigBee-compatible?

#1  Based on below cluster info etc. can you give example of ZCL commend format to set brightness to 50%?

If you have not seen the zigbee samples already I believe the light bulb sample is a dimmable ZigBee device which might contain the functionality you are looking for. And I will take a look and see if I have a sample available tomorrow morning

#2  Can you me show ZCL command example of how to read this?

I believe I also have a sample for this, but I will need to find it first

Kind regards,
Andreas

[JorgenM]

Hi Andreas, It's just a generic Zigbee HA1.2 RGB light bulb with dimming and color control. 
The Nordic zigbee light bulb controller example is a eval board "light switch" so it doesn't show actual codes that are sent. 
At least I could not find it...?
I think I'm not sending the correct attr_id 
In fact I'm not sure where I can find the att_id , for "level" (dimming) as example.

I assume it's not the same as "level" cluster which is 0x0800 as per below.

An example of the just the command would be great.

Thanks!
Jorgen

[JorgenM]

Thanks Andreas, 

I was able to step up and down now!

If I step too many times down or up and device is all the way to it's upper or lower limit already I get error msg saying="request pool is full, wait a little".

Can you tell me what causes this error?

I should not have to keep track of where I am on the step scale for a device, right?
It should just reject or do nothing if I keep stepping beyond full/lowest brightness, please confirm.

The "request pool..." error does not clear unless I push reset on eval brd and then I have to restart stack.
So it's probably related to Nordic FW.
Is there a way to clear this message with command?

Thanks again!
Jorgen

[AHaug]

Hi Jorgen,

I will see if I can dig up an explaination to this, however it might take some time (again due to vacation).

My immediate thought is that the request pool only either has room for a given amount of orders (fixed bit number of orders can be held at a given time), or that the limit has been overstepped. Does the error clear if you step down after reaching the upper limit?

Kind regards,
Andreas

[JorgenM]

Andreas,
No problem we have other things to work on .
I believe once first error is generated it does not matter if I step down when at high limit or if I'm at low limit and step up. I will confirm this.
( "waiting" does not help) 
Thx, Jorgen

[AHaug]

Hi Jorgen, 

Apologies for the long response time. Have you checked if the error with the request pool is present in NCS v2.0.0 as well? 

Kind regards,
Andreas

[JorgenM]

Andreas,
No I have not. Can you confirm that was a change in v2.0 ?

I don't mind moving to a newer version NCS once in a while, however doing this we have to regression test everything we know works in v1.9 since behavior is always a bit different.

Thanks,
Jorgen