CC2530: Sometimes the coordinator is failed to send out the packets

Yonghua Pan

Intellectual 1900 points

Part Number: CC2530
Other Parts Discussed in Thread: Z-STACK, CC2591, CC2592

Hi,

According to a customer's field testing, they found sometimes CC2530 is failed to send out packets.

According to the packet sniffer, I found there was no packet from the coordinator during this issue happening.

I've recommended them to try the following 3 options, but no improvement. Would you let me know any further debugging methods on such issue.

1. #define CCA_THR 0xF7

#define CCA_THR_HGM 0xFC

2. #define CCA_THR 0xF2

#define CCA_THR_HGM 0xF7

3. #define CCA_THR 0xF7

#define CCA_THR_HGM 0xFC

In function macMcuInit()：

Change MDMCTRL0 = 0x85;to be MDMCTRL0 = 0xA5

Change MDMCTRL1 = CORR_THR; to be MDMCTRL1 = 0x38

Regards,

Yonghua

over 6 years ago

0 YiKai Chen over 6 years ago

Guru 735695 points

When you said " they found sometimes CC2530 is failed to send out packets", what API and codes do they use in application codes?

0 Yonghua Pan over 6 years ago in reply to YiKai Chen

TI__Intellectual 1900 points

SDK ver: Z-Stack Home 1.2.2a.44539，
Example：SampleSwitch，
API：zcl_SendCommand

BTW, this issue cannot be reproduced in clean environment.
Looking forward to get some constructive comments.
Thanks!

Regards,
Yonghua

0 YiKai Chen over 6 years ago in reply to Yonghua Pan

Guru 735695 points

Does SampleSwitch act as end device? By the way, what do you mean " this issue cannot be reproduced in clean environment."? Is the environment noisy or is there a lot of 2.4 G interference?

0 Yonghua Pan over 6 years ago in reply to YiKai Chen

TI__Intellectual 1900 points

No, it's coordinator. Yes, the environment is noisy since it's end customer's showroom and there are some 2.4Ghz devices around.

0 YiKai Chen over 6 years ago in reply to Yonghua Pan

Guru 735695 points

If SampleSwitch is coordinator, how about zcl_SendCommand destination? Is it end device or router?

0 Yonghua Pan over 6 years ago in reply to YiKai Chen

TI__Intellectual 1900 points

Hi, Yikai, Destination is router.

0 YiKai Chen over 6 years ago in reply to Yonghua Pan

Guru 735695 points

When SampleSwitch cannot send zcl_SendCommand, what is the return status of zcl_SendCommand?

0 Yonghua Pan over 6 years ago in reply to YiKai Chen

TI__Intellectual 1900 points

It returns SUCCESS while this issue happening.

0 YiKai Chen over 6 years ago in reply to Yonghua Pan

Guru 735695 points

Do you mean zcl_SendCommand returns SUCCESS but actually you don't see anything send out in sniffer?

0 Yonghua Pan over 6 years ago in reply to YiKai Chen

TI__Intellectual 1900 points

Yes, Yikai.

0 YiKai Chen over 6 years ago in reply to Yonghua Pan

Guru 735695 points

I would suggest if SampleSwitch sends router request instead of real zcl_SendCommand because the routing path is lost.

0 Toby over 6 years ago

TI__Mastermind 35070 points

Hi,

One other suggestion is to add more channels to the DEFAULT_CHANLIST so that when the ZC forms the network, it chooses the best channel.

Regards,
Toby

0 Ryan Brown1 over 6 years ago

TI__Guru**** 219527 points

Hi Yonghua,

If possible, please have the customer provide sniffer logs, more detailed ZC behavior (does it still send out link status messages regularly, require a restart after the issue occurs, send any other data/commands, etc), and network topology (neighbors directly connected, total devices, etc). Try increase maxCsmaBackoffs to 5 and have them increase their maximum network buffer sizes: processors.wiki.ti.com/.../Zigbee_Known_Issues_and_Proposed_Fixes

http://www.ti.com/lit/wp/swra635/swra635.pdf

Regards,
Ryan

0 Yonghua Pan over 6 years ago in reply to Ryan Brown1

TI__Intellectual 1900 points

Hi, Ryan,

Please see attached sniffer log in which there are no packets to devices CA8F and 1065.

Thanks!

Regards,

Yonghuahttps://e2e.ti.com/cfs-file/__key/communityserver-discussions-components-files/158/No_5F00_packet_5F00_to_5F00_Device_5F00_CA8F_5F00_and_5F00_1065.7z

0 Ryan Brown1 over 6 years ago in reply to Yonghua Pan

TI__Guru**** 219527 points

Hi Yonghua,

Everything sent out from 0xCA8F are broadcasts and retransmissions, therefore the coordinator has no need to reply with a MAC ACK. 0x1065 does send Route Reply (260) unicast directly to the coordinator for which it receives a MAC ACK. The coordinator itself sends an On/Off: Read Attributes (273), APS ACK (286), and On/Off: Default Response (288) to 0xC847 as routed through 0x1065 and with the appropriate MAC ACKs.

Also, the link status list shows that these devices are very close to or past the default MAX_NEIGHBOR_ENTRIES value but show a great connection to the coordinator (incoming/outgoing costs of 0x1 are optimal). The coordinator itself has no link status message but the log is only 10 seconds long, shorter than the 15-second default NWK_LINK_STATUS_PERIOD. I see very few retries other than an IAS Zone: Default Response from 0xAB64 but this has little to do with the routers in question. There is some prevalent route request storms which can be addressed with DEFAULT_ROUTE_REQUEST_RADIUS and ROUTE_DISCOVERY_TIME.

I cannot infer anything from this sniffer log, are there any specific packets I should be looking at or other behavior to notice?

Also, in your original post you state CCA_TRH_HGM values of 0xF7/0xFC but the default is 0x06. Is this a typo or do you intend to lower it by 10 dBm (signed char values)? Keep in mind that values greater than the threshold indicate that the receiver holds the CCA as busy, therefore you want to raise the threshold (increase CCA_THR) to gain leniency. However, you shouldn't change these values too much due to IEEE 802.15.4 Specifications.

Regards,
Ryan

0 Yonghua Pan over 6 years ago in reply to Ryan Brown1

TI__Intellectual 1900 points

Hi, Ryan,

I took the following code as reference and want to change the threashold to be -80dBm

#define CCA_THR 0xFC /* -4-76=-80dBm when CC2530 operated alone or with CC2591 in LGM */
#define CCA_THR_HGM 0x06 /* 6-76=-70dBm when CC2530 operated with CC2591 in HGM */

Thanks for your comment, i'll take care of these parameters.

We're trying to convince customer to test in field with your recommendation from WiKi above. Will provide more sniffer logs after the testing done.

Regards,

Yonghua

0 Ryan Brown1 over 6 years ago in reply to Yonghua Pan

TI__Guru**** 219527 points

Hi Yonghua,

Is the customer using an external PA/LNA like the CC2591/CC2592? Otherwise it is not necessary to change the CCA_THR_HGM.

As mentioned earlier, a threshold value of -80 dBm is more strict than -70 dBm in the context of allowing CCA operation, also 0xF7 = -9 dBm and 0xF2 = -14 in the context of signed bytes.

The customer could reprogram the ZC without deleting the NV pages of flash or clone the ZC with the modified Z-Stack changes in order to avoid recommissioning the network.

Regards,
Ryan

Zigbee & Thread

Zigbee & Thread forum

CC2530: Sometimes the coordinator is failed to send out the packets