Helping increasing throughput, Weird deadtime behaviour

[lagrangepoint1]

I am investigating increasing throughput in our application. Upon investigation both with our software and the ble_app_att_mtu_throughput example code. I found that there is approximately 5ms dead time after the last packet. This is effectively reduce the number of packets we can send per connection internval. 

I have validated this by using different connection intervals and event lengths. The two images attached (caught using a BLE sniffer) show connection intervals of 7.5ms and 50ms. You will see that there always a 5ms dead time between the last packet and the start of the connection interval. This reduce the number of packets we can send per connection internval and thus reduce throughput.

Is there anyway to stop this from happening?

Softdevice used is S140 on the nRF52840

[Jimmy Wong]

Hi,

What is the NRF_SDH_BLE_GAP_EVENT_LENGTH in your setting?

In the softdevice S140 specification v2.1, it shows the throughput as below.

I did the example code on throughput with packet error rate (https://github.com/jimmywong2003/nrf5-packet-error-rate-measurement-on-ble-connection).

 

[Edvin]

Hello,

What sort of throughput are you looking at? Have you enabled DLE (Data length extension)?

[lagrangepoint1]

The NRF_SDH_BLE_GAP_EVENT_LENGTH was set to 244. I found that as I increase the NRF_SDH_BLE_GAP_EVENT_LENGTH the time between the last packet and the next connection interval increased. 

Is this suppose to be happening?

[Edvin]

If you look in the ble_app_uart_example, this one has about the maximum throughput you can get, except using 2MBPS, I believe. What you want to look for is the NRF_SDH_BLE_GAP_DATA_LENGTH. Try setting this one equal to what it is in the ble_app_uart example (251).

What sort of throughput are you looking at? It is difficult to say whether you can improve it further without knowing what you already have. You can't use the radio 100% of the time. It has to follow the BLE specification.

BR,

Edvin