How to achieve low radio latency and USB MIDI on Nordic nRF52840?

Official Hardware Nano Family Nano 33 BLE
1

Hi everyone.

I'm considering Nano 33 BLE to replace my current prototype with a pair of Micro Pro + NRF24L01.

As Bluetooth can be too slow for my requirements (1mbps with latency below 2ms, which I have working now with NRF24L01), I would like to make the BLE module work similarly to the old NRF24L01.

From quick glance in the specsheet of nRF52840 https://content.arduino.cc/assets/Nano_BLE_MCU-nRF52840_PS_v1.1.pdf, it seems that MODE Nrf_1Mbit or Nrf_2Mbit should work the same way as with NRF24L01. Please correct me if I'm wrong.

Now the question is how to switch nRF52840 to this mode? Has anyone tried it and can confirm that communication between two Nano33 BLEs can be set up to work the same "raw" way as with NRF24L01, without all the BLE stuff? Is it possible without dealing with JLink firmware and whatnot?

I'm using nRF24/RF24 library for my current prototype. What should I use with Nano33 BLE to make it work the same way?

Also, I will need one Nano33 BLE to emulate a USB MIDI device, and, as I understand, Nano33 supports mbed, so PluggableUSB libraries should work, right?

2

The Mbed cores use a different API for their PluggableUSB libraries, last time I checked, the MIDIUSB library was not supported.

I do provide a USB MIDI implementation for the Nano 33 BLE as part of my Control Surface library: GitHub - tttapa/Control-Surface: Arduino library for creating MIDI controllers and other MIDI devices.

However, while implementing that, I encountered issues with the underlying USB stack that still have not been addressed by the Arduino team: Nano 33 BLE unreliable USB communication · Issue #129 · arduino/ArduinoCore-mbed · GitHub
I don't know if they have been fixed in upstream Mbed OS.

3

Coincidentally I'm using Control Surface library (it's great, thank you for your work, Peter) in my project.

I was glad to see Nano 33 BLE mentioned here Control Surface: MIDI over USB but, as you explained, my excitement was premature.

This means another board I was looking at (Seeed XIAO nRF52840) will suffer from the same issue. So, I'll have to live with my current setup and postpone the miniaturization idea for a while. Unless there are other small Arduino-like boards with native USB + a low latency wireless solution.

4

@PieterP I'm wondering if there are any workarounds to make your PluggableUSBMIDI work with nRF52840, even if it means sacrificing some performance?

In that GitHub thread, I see they at least offered a patch to fix the bytes being dropped out in the middle of the payload. Is the patch already included in the main branch? I see they merged something on September 18, but not sure what changes were included there.

If I understand correctly, if they have merged that patch, then there's only the problem that you cannot send more than 64 bytes at once. Could it be worked around with some ugly data splitting and timing adjustments (send 64, wait for ... not sure what and how long... send the next 64 bytes until you are done)? That might hurt performance, but as long as we can get reasonably close to the old MIDI 1 spec bitrate, it would be acceptable for many projects, with a large warning that you should not attempt to process large sysex data blocks on nRF52840.