We have been successfully sending UDP messages at a sustained rate of 100 Hz (100 messages/second) from an arduino RP2040 Nano over a wifi connection. The UDP message payload is 33 bytes, so including 8 bytes of UDP header and 20 bytes of IP header our outbound data rate is:
(20 bytes IP hdr + 8 bytes UDP hdr + 33 bytes UDP payload) * 100 msg/sec = 6100 bytes/sec, or roughly 60,000 bits/sec (60 kbits/sec).
We are now trying to do the same using a W5500 Ethernet Shield connected to the RP2040 Nano but we are only realizing a message rate of about 16-17 Hz, which is about 1000 bytes/sec or roughly 10 kbits/sec. Our messages include sequence numbers so we can identify dropped messages, but our received message sequences have no gaps so we know the outgoing message rate is indeed just 17 Hz with ethernet vs. 100 Hz over wifi.
The codebase is identical except for the initialization of the network connection – we use #define ETHERNET and #ifdef/#else/#endif conditional compilation to include either Ethernet.h or WiFiNINA.h and to configure the connection for each, but otherwise the code is the same. We are using version 2.0.2 of the Ethernet library.
From all we have read online, our target of 60 kbits/sec should be achievable with the W5500, so we are wondering what errors or omissions we are making that are limiting our throughput to 10 kbits/sec. We have tried explicitly configuring the SPI using settings like:
SPI.beginTransaction(SPISettings(30000000, MSBFIRST, SPI_MODE0));
as well as other speeds like 10, 20, 40, and 80 MHz but none of it makes any difference… we still only get 16-17 messages/second.
We have completely eliminated all references to Serial, and we do not initialize it, but the speed remains the same. We are not aware of using any other component making use of SPI (we are using the accelerometers on the RP2040 Nano but those use an I2C connection), so we don’t think there’s anything that should be competing with or affecting our Ethernet traffic.
We will be grateful for any information, suggestions, pointers, links, etc that help us either to achieve the data rates we are hoping for or to understand why that’s not possible with this configuration (in which case suggestions of alternative hardware and software configurations that might work for us are appreciated).
Thanks,
Chris
