Hi,
NRF52833 or nRF52840 LLPM 1 master 8 slave
What is the data delay per slave device?
Thank you for all your assistance.
Kind regards,
Peter.Min
Hi,
NRF52833 or nRF52840 LLPM 1 master 8 slave
What is the data delay per slave device?
Thank you for all your assistance.
Kind regards,
Peter.Min
Hi,
This has been tested for a 1 central 1 peripheral scenario using the Bluetooth: LLPM example and 1 ms connection interval. The numbers can be seen in the documentation for that example, typically 1000 to 1400 us. It is labelled as "Transmisison Latency", but in reality it is round-trip time divided by two, so it does account also for some handling in the receiving end.
For longer connection intervals this value will be higher. The achievable connection interval depend on number of devices and the event length for each connection.
The SoftDevice Controller will schedule central events back-to-back for a maximum throughput. Please note that if combining with other BLE roles, such as observer/initiator, you will not necessarily be able to pack connections as tightly, as time is also needed for the other roles. The strategy then is to first connect to all devices, then disable other roles (including the role scanning for new connections), then update connection parameters for minimum connection interval.
Regards,
Terje
Hi tesc,
With 1 master and 8 slave, can each slave achieve a 5ms connection interval?
If it is not possible to achieve a 5ms connection interval for each slave device, what is the minimum connection interval for each slave device using 1 master and 8 slaves?
Thank you for all your assistance.
Kind regards,
Peter.Min
Hi,
Based on the numbers from the SoftDevice Controller Scheduling documentation, I would expect up to 7 connections to peripherals from one central using 2 Mbps PHY and LLPM. The best approach is likely to initiate the connections at a higher connection interval at first, and when all connections are established to decrease it to 5 ms (when the connections don't have to compete for time on the radio peripheral due to concurrent scanning for new connections and establishing those new connections.)
I recommend rading through the scheduling documentation (linked in previous paragraph), as it does give good insight into how the SoftDevice Controller sets up and schedules connections, as well as what is possible.
Regards,
Terje
Hi tesc,
1.LLPM 1 Central device 7 peripheral devices, what is the minimum connection interval for each secondary device?
2.LLPM 1 central device 7 peripheral device, using the development board test, what is the packet loss rate of each slave device within 0.5 meters?
3.LLPM 1 Central device 7 peripheral device. Will bluetooth connection events be affected if flash read/write is enabled for each secondary device?
4. The customer hopes to use Nordic chip to realize 1 master and 8 slave, and the connection interval within 5ms. However, without BLE protocol, can it be realized by using only 2.4g transparent transmission?
5.BLE protocol, non-LLPM, measured 1 master 8 secondary 2.5ms connection interval, 2M PHY 0.5m distance, each external device uploading data to the central device, the measured delay exceeds 20ms, what is the cause of this, why is the average delay time of each device uploading data inconsistent?
Thank you for all your assistance.
Kind regards,
Peter.Min
Hi,
1) After further discussion with the SoftDevice team, it is not clear if the LLPM functionality in the SoftDevice Controller in nRF Connect SDK can handle more than 1 connection per 1 ms connection interval. In other words, that means the limit for 5 ms connection interval may be 5 connections. We do not have numbers for 7 connections, and have not tested if 7 connections is possible within 5 ms interval.
2) Packet loss rates depend on multiple factors, and in particular on the environment. With 0.5 metres distance the packet loss should be minimal.
From the protocol point of view, with no other devices or interference there should not be packet loss in a star network of one central connected to N peripherals, as long as the connection interval is large enough to fit the connection events of all N peripherals.
In practice you may see packet loss due to interference from other devices.
3) Flash write and flash erase can affect BLE performance on the nRF52 series, yes, since the CPU is stalled during flash and erase operations.Flash writing requires 42.5 µs per 32 bit word, and flash erase requires 87.5 ms per page. This will interrupt BLE stack scheduling.
4) 8 connections with connection interval of 5 ms is very tight, and while it may be technically achievable with other technologies it would be a lot of work and I cannot recommend it.
5) With regular BLE, the minimum connection interval is 7.5 ms. The exact nature of the communication is difficult to know without e.g. a sniffer trace using the nRF Sniffer. Can they provide such a log?
PS This thread is public. If you want to share any confidential details, then let me know and we can turn this into a private ticket instead.
Regards,
Terje
