Expected throughput on nrf52840

[Gary]

Hi,

I have two nrf52840 boards and I am using the ble_app_att_mtu_throughput test app. The tests seem to be working OK, but I am not getting the throughput I was expecting. The boards are next to each other.

With BLE_GAP_PHY_2MBPS I get 1046.48 Kbps, which I guess it in the right ball park, but for BLE_GAP_PHY_1MBPS I am only getting 262.27 Kbps.

Also for BLE_GAP_PHY_CODED with the boards next to each other I am only getting 26.20 Kbps.

I was expecting more for 1MBPS and CODED. I have tried with S140_nrf52840_5.0.0-2.alpha and S140_nrf52840_5.0.0-3.alpha and the numbers seem to be around the same.

Any thoughts, comment suggestions?

Thanks, Gary.

[bjorn-spockeli]

Using the same method that the Bluetooth SIG to calculate the theoretical throughput, see this blog, I arrive the following calculations.

Time to send 265byte PDU @ 125kbps: 265*8bits/125kbps = 16.96ms

Time to receive 10byte PDU @ 125kbps: 10*8bits/125kbps = 640us

The Inter frame space, T_IFS ,the time interval between two consecutive packets = 150us

The usable payload that of a single ATT packet that fits inside the max PDU size is 247bytes, thus the theoretical throughput should be

247*8bits/(640us + 150us + 16960us + 150us) =~ 109kbps

If you have set the connection interval to 400ms and have Connection Event Extension enabled then you should be able to fit 400ms/16.96ms =~23 packets in one connection event, but it sounds like you're getting around half of that. I will look into this and update my answer.

[Gary]

According to this blog post devzone.nordicsemi.com/.../ it seems that Ole Bauck was getting 21.3 Kbps for BLE 5 Long Range.