HL7900 – RI not triggered all the time on incoming TCP/MQTT data in eDRX

[htb.tom41]

Hello,

Context

I am currently working on implementing eDRX on the HL7900 / HL7812 modem and I am encountering an issue related to the Ring Indicator (RI).

The device communicates using MQTT over TCP with a broker. My final goal is to run the modem with eDRX enabled, use Lite-Hibernate mode, and rely on the Ring Indicator to wake up the host MCU when a downlink message is received.

However, the RI signal does not seem to trigger consistently. In some situations, when a message is sent to the modem, the application subsystem appears to wake up (view with consumption) and process the incoming data, but the RI signal is not asserted, meaning the host MCU is not notified.

At this point, I don’t know if this is related to:

• a hardware problem on my end,
• how to configure the RI via AT commands,
• how to manage the wake-up phases after RI detection.
• Time between RI detections
• …

I have already run several tests but have not been able to identify the real cause.

Tests performed

TEST 1 – Baseline behavior : RI without eDRX, No low power mode enabled, MQTT connection

Purpose: verify that RI works correctly in a simple scenario.

MQTT connection, subscribe to a topic, then send two ping requests without reading the acknowledgments between them.

TEST 2 – eDRX + Lite-Hibernate : eDRX cycle set to ~20 seconds, Messages sent quickly from the broker

Purpose: observe RI triggering during normal eDRX operation.

Here, the first one works, but the second one causes the HL7900 to crash. The ring indicator does not activate during the second wake-up. I then tried to wake it up (forcing it with wake-up) and send messages with pingreq, but I did not receive any messages from the broker (and the ring indicator no longer works). Looking at the MQTT broker manager, my client is still connected and the broker is receiving the pings sent by the HL.

TEST 3 – Periodic WAKEUP : Same configuration as Test 2, I manually wakes the modem every minute using WAKEUP, I send à PINGREQ and wait the response of the broker.

Purpose : Observe the behavior of the RI between eDRX sleep cycles.

Each time you wake up, here are the commands sent:

>>> AT
OK
>>> AT+KTCPSND=1,2 
>>> C0 00
CONNECT
OK
+KTCP_DATA: 1,2
>>>AT+KTCPSTAT=1
+KTCPSTAT: 3,-1,0,2
OK
>>> AT+KTCPRCV=1,2
CONNECT
[KTCPRCV DATA session 1, 2 byte(s)]
D0 00
--EOF--Pattern--
OK

TEST 4 – Long interval : Same configuration as Test 2, Messages sent from broker at longer intervals

Purpose : Observe whether the timing between eDRX cycles affects RI (waiting for the calibration phase).

Each time you wake up, here are the commands sent:

Hardware and firmware

• Modem: HL7900

• Firmware version: BF_CORE_02_02_00_00_52661

• Hardware platform: * Semtech HL7812 / HL7900 development kit

For the setup:

• External 3.7 V power supply

• Communication via UART1

• AT commands sent from a Python script on PC

Has anyone experienced a similar issue ? Any ideas or suggestions would be greatly appreciated.

Thanks in advance for your help.

[jyijyi]

i saw your +KRIC is not using RI pin…

BTW, is it the same situation for your HL7812?

[htb.tom41]

>>> AT+KRIC?
+KRIC: 64,0,5,2,1

For my test i am using the ‘GPIO2’ pin and not the ‘UART_RI’.
I didn’t mention it here, but I also ran tests with UART_RI and got the same results.
I also tried changing the timing of the pulse or removing it, but it didn’t help.

[jyijyi]

is it the same situation for your HL7812?

[htb.tom41]

Unfortunately, I don’t have an unsoldered HL7812, so I couldn’t perform my tests on the Semtech dev-kit. However, I did the TEST 2 with an HL7812 that was soldered onto a board with a microcontroller, and I obtained approximately the same results.

[jyijyi]

I can see the RI signal is working fine with the following test:

+CREG: 1,"C672","0188A9DF",9
AT+KCNXCFG=1,"GPRS","nbiot"
OK
AT+KCNXPROFILE=1
OK
AT+KTCPCFG=1,0,"123.123.123.123",5044
+KTCPCFG: 1

OK
AT+ktcpcnx=1
OK

+KCNX_IND: 1,1,0

+KTCP_IND: 1,1
at+kric=64,0,5,0,1
OK

AT+KTCPSND=1,5
CONNECT

OK

+KTCP_DATA: 1,2
at+ktcprcv=1
+CME ERROR: 3
at+ktcprcv=1,2
CONNECT
aa--EOF--Pattern--
OK

+KTCP_DATA: 1,2
at+ktcprcv=1,2
CONNECT
aa--EOF--Pattern--
OK

+KTCP_DATA: 1,1
at+ktcprcv=1,1
CONNECT
a--EOF--Pattern--
OK
ati3
HL7812.5.7.3.0

OK
at+cedrxs?
OK
AT+CEDRXRDP
+CEDRXRDP: 0

OK

[htb.tom41]

Okay, thank you. In your tests, did you try switching to low power mode (sleep or lite-hibernate) with edrx and waiting for the RI notification to wake up the HL and process the message?

Because what I describe in my tests is mainly during operation with edrx. In normal mode without edrx or low power mode, it works perfectly well and normally.

[jyijyi]

didn’t you mentioned that it has problem in test1?

I have also tried cedrx mode, no problem is found:

at+cedrxs=0
OK
at+cedrxs=1
OK
at+cedrxs?
+CEDRXS: 4,"0101"
+CEDRXS: 5,"0010"

OK
AT+CEDRXRDP
+CEDRXRDP: 5,"0010","0010","0000"

OK
ati3
HL7812.5.7.3.0

OK
at+Cereg?
+CEREG: 5,1,"C672","0188A9DF",9

OK
AT+KCNXCFG=1,"GPRS","nbiot"
OK
AT+KCNXPROFILE=1
OK
AT+KTCPCFG=1,0,"123.123.123.123",5048
+KTCPCFG: 1

OK
AT+ktcpcnx=1
OK

+KCNX_IND: 1,1,0

+KTCP_IND: 1,1
at+kric=64,0,5,0,1
OK
AT+KTCPSND=1,5
CONNECT

OK

+KTCP_DATA: 1,1
aat
OK
at+ktcprcv=1,5
CONNECT
1--EOF--Pattern--
OK

+KTCP_DATA: 1,1
at+ktcprcv=1,5
CONNECT
2--EOF--Pattern--
OK

+KTCP_DATA: 1,2
at+ktcprcv=1,5
CONNECT
34--EOF--Pattern--
OK

+KTCP_DATA: 1,1
at+ktcprcv=1,5
CONNECT
5--EOF--Pattern--
OK
at+ksleep=1,1,10
OK

//here the module goes to sleep mode.
//And I can see the RI pin toggled when receiving TCP data

+KTCP_DATA: 1,7

[htb.tom41]

Okay, thank you for your help. Test #1 just shows that the ring indicator mechanism works normally when the edrx is not implemented.
Regarding your tests, how long after sleep do you send TCP data ? And does it work even if you repeat the sleep cycles after reading the data and restarting the operation ?

Because in my tests, after the first sleep phase, the RI works, and after receiving the first packet, if I go back to sleep, the next ring indicator often has problems, etc.
It’s as if it doesn’t reset correctly.
It only works fairly well when I space out my TCP transmissions by about 2-3 minutes. But sending a message a few tens of seconds after falling asleep doesn’t always work (with a 20-second eDRX cycle in my case).

[jyijyi]

Have you contacted distributor on this issue?

I tried to receive TCP data during sleep mode twice times, no problem is found on RI pin.
i.e.
sleep → server sends TCP data → RI toggles → wake up and read the TCP data → sleep → server sends TCP data → RI toggles

[htb.tom41]

Thanks for your feedback.

Yes, we observe the same behavior as you when using Lite-Hibernate only (without eDRX):
the cycle sleep → TCP data → RI → wake-up → read → sleep works reliably.

However, the issue appears specifically when eDRX is enabled together with Lite-Hibernate (or other LPM).

In that case:

• the first downlink after sleep usually triggers RI correctly,
• but after going back to sleep, RI does not reliably trigger anymore,
• sometimes the modem wakes up (visible on current), but no RI is generated,
• and in some cases, it never triggers again until we manually wake up the modem.

Also, when the issue occurs, we can still send data (e.g. ping), and the broker responds,
but no downlink notification is received on the modem side (no RI / no TCP indication).

[jyijyi]

my test above has enabled at+cedrxs=1

[htb.tom41]

Okay, are you using the Semtech HL7900 dev kit?
For the low-power mode, are you using Lite-Hibernate? If so, with what delay ?

Would it be possible for you to send me the AT commands you’re using and the order in which you use them ?

Do you think the MQTT broker affects the use of the ring indicator?

[jyijyi]

I was using the HL78 red dev kit.
Do you mean you want to know the +KSLEEP setting?

[htb.tom41]

Yes, please, and if possible, the AT commands you used and the order in which you used them.

[jyijyi]

at+ksleep?
+KSLEEP: 2

OK

Do you see such issue if using +KSLEEP in 2?
Have you contacted distributor on this issue?

[htb.tom41]

Maybe we misunderstood each other, but from what I’ve seen, AT+KSLEEP 2 corresponds to an awake mode.

That said, I just tested it with eDRX mode while still using KSLEEP 2, and I’m encountering the same problem.

[jyijyi]

@htb.tom41

That said, I just tested it with eDRX mode while still using KSLEEP 2, and I’m encountering the same problem.

then does this mean i cannot reproduce your problem?

BTW, my test below has also put the module to lite-hibernate mode by “at+ksleep=1,1,10”:

HL7900 – RI not triggered all the time on incoming TCP/MQTT data in eDRX

didn’t you mentioned that it has problem in test1? I have also tried cedrx mode, no problem is found: at+cedrxs=0 OK at+cedrxs=1 OK at+cedrxs? +CEDRXS: 4,"0101" +CEDRXS: 5,"0010" OK AT+CEDRXRDP +CEDRXRDP: 5,"0010","0010","0000" OK ati3 HL7812.5.7.3.0 OK at+Cereg? +CEREG: 5,1,"C672","0188A9DF",9 OK AT+KCNXCFG=1,"GPRS","nbiot" OK AT+KCNXPROFILE=1 OK AT+KTCPCFG=1,0,"123.123.123.123",5048 +KTCPCFG: 1 OK AT+ktcpcnx=1 OK +KCNX_IND: 1,1,0 +KTCP_IND: 1,1 at+kric=64,0,5,0,1 OK AT+KTCPSND=1,5 CO…

[htb.tom41]

It looks like you’ve run the same test. I just ran another test here with ksleep = 2, but I’m still having the same issue.

Here are the AT commands I’m using (in order):

- AT+CFUN=4

- AT+CEDRXS=1,4,2

- AT+CPSMS=0

- AT+KSLEEP=2

- AT+CFUN=1,1

- AT+CGDCONT=1,“IPV4V6”,“***.***.***”

- AT+KCNXCFG=1,“GPRS”,“***.***.***”

- AT+KTCPCFG=1,0,“****”,****

- AT+KTCPCNX=1

- AT+KTCPSND=1,73 [MQTT CONNECT]

CONNECT

OK

+KTCP_DATA: 1,4

- AT+KTCPRCV=1,4

- AT+KTCPSND=1,39 [MQTT SUBSCRIBE]

CONNECT

OK

+KTCP_DATA: 1,5

- AT+KTCPRCV=1,5

...

Here is the test graph showing the initialization phase, connection, and then the eDRX cycles with message waiting (in my case, I send a single message). And as you can see in red, sometimes it wakes up but doesn’t activate RI even though the message is intended for it.

Can you tell me if you’re seeing the same situation?

[jyijyi]

a bit confused…

is now issue related to RI pin or eDRX?

Have you contacted distributor on this issue?