nRF52840 maximum data throughput

RE: nRF52840 maximum data throughput

endnode — Sun, 05 Mar 2017 10:57:01 GMT

Hi Roger, I totally agree, going with something like ESP8266 or ESP32 based low-power WiFi should be much easier then fiddling with multiple BLE chips. I cannot exclude that by teaming 3-5 nRF52840 you couldn't achieve same bandwidth as on ESP with still lower power consumption but development cost (on both HW and SW) should prevent you from going that way...

RE: nRF52840 maximum data throughput

Roger Clark — Sat, 04 Mar 2017 23:00:33 GMT

The speed is impressive, but I'm not sure whether it would be worth running multiple devices at full speed to get what you can easily get over a WiFi link.

I know that Wifi devices can take a lot of current, but I wonder if its really that much more than multple nR52's running at max speed, vs some power optimised Wifi device

And the hardware and software effort required to build a system using nRF52's, may make it uneconomic. Plus the risk of attempting to build a system with multiple devices all transmitting and receiving in very close proximity to each other could easily cause problems

Personally, not a road I would go down, unless someone was willing to pay for all the R&D knowing that the project had significant risk factors (IMHO)

RE: nRF52840 maximum data throughput

endnode — Sat, 04 Mar 2017 21:39:41 GMT

Well if you team the devices then you should be able to get there. I'm really impressed by BT5.0 demo by Nordic but I'm struggling to decode if this is throughput on APP layer (on top of GATT) or on link layer. Should not do big difference with all the extensions (where PDU and ATT_MTU sizes are much larger then overhead) but still. I will make separate question.

RE: nRF52840 maximum data throughput

Roger Clark — Sat, 04 Mar 2017 20:22:57 GMT

@endnote

Thanks. Thats higher than I thought it was. But its still not 2Mbps ;-)

RE: nRF52840 maximum data throughput

endnode — Sat, 04 Mar 2017 19:49:26 GMT

OK, so you would like to use BLE. The channel hopping in BLE is exactly for the purpose of minimizing collisions but because BLE is very much time-dependent and these separate devices (running separate BLE stacks) would need to be in microsecond sync to be able to prevent these and have 100% of theoretical bandwidth. There is one way how to bypass the hopping and that's channel map which can be set for each device and you can limit the band down to single channel. Of course you will destroy original properties limiting collisions but if you are able to coordinate this mapping and at the same time ensure that none of assigned channels is noisy it could work. So theoretically you could get 37*1.36Mbps if you team 37 devices on each side and perfectly distribute available bandwidth...

RE: nRF52840 maximum data throughput

DJ — Sat, 04 Mar 2017 18:47:05 GMT

I have worked with nRF51822 and nRF24L01+ in the past and have written proprietary protocols to get 2 Mbps. With BLE 5.0, if running several chips chips on both sides and parallelize the stream, what is the "available bandwidth in the radio"? There are 37 data channels and 3 advertisement channels in the new BLE stack, but with hopping, it is unclear to me how many independent channels you can fit there.

RE: nRF52840 maximum data throughput

endnode — Sat, 04 Mar 2017 12:35:19 GMT

Btw. for BLE here is maximum throughput demo with BT5.0 2Mbps feature and all possible PDU, ATT_MTU and Connection interval extensions. This gives you 1.3~1.4Mbps. Note that this is most probably on the link layer and that this is one way unacknowledged (and most probably unprocessed) streaming. If you want to do any meaningful with the data it will drop, but how much is very application specific.

RE: nRF52840 maximum data throughput

endnode — Sat, 04 Mar 2017 11:22:11 GMT

Hi again, first you need to be very specific about what networking protocol you are talking about. 2Mbps is some physical bitrate modulated on top of one channel inside 2.4GHz ISM band. That has very little to do with throughput which is your concern. So are you taking about theoretical (proprietary) data transfer over nRF52840 radio or some specific protocol stack like BLE or IEEE 802.15.4 or ANT or what? Then you are talking about running it simultaneously. Do you mean having several chips on both sides and parallelize the stream? As most of the protocols are peer to peer you can of course run like that until you have available bandwidth in the radio. And this is again determined by the protocol stack you want to use because they differ in terms of number of channels, channel allocation (or hopping), link reliability mechanisms, overhead in lower layers etc.

RE: nRF52840 maximum data throughput

Roger Clark — Sat, 04 Mar 2017 06:56:25 GMT

I recommend you use WiFi as IMHO BLE is not designed for the sort of application you have outlined

(Yes. WiFi may take more power, but you don't get bandwidth without taking more power)

RE: nRF52840 maximum data throughput

DJ — Sat, 04 Mar 2017 00:28:38 GMT

Hi Jan,

Thank you for your reply. The application of interest indeed requires >2 Mbps and I'm exploring options to do so in the lowest power possible. I fully agree with you regarding caveats in getting 2 Mbps from each of this module (e.g., overhead, reliability, signal integrity, etc.). However, assuming we can do that for the sake of argument, my question was how many of these nRF52840 chips can be used, independently (or simultaneously), to increase the overall throughput?

For instance, perhaps you can de-tune the center frequency or leverage the fact that there are separate data / advertisement channels to get higher total bandwidth (independent data stream). As an example, if I were to use, say 10 of these nRF52840 (configured as TX), each streaming at 2 Mbps to a receiver that can handle this load (less power constraint here), could I get an effective bandwidth of 20 Mbps? **

Hope this clarifies my original question.

Thanks, DJ

** It's possible the overall power consumption of 10 of these chips, fully streaming, might consume more overall power than other solutions.

RE: nRF52840 maximum data throughput

endnode — Sat, 04 Mar 2017 00:07:30 GMT

Hi DJ,

What do you mean by simultaneously? This is so vague that I even cannot give you one from all the myriads of answers. On the second question: in theory you might get continuous 2Mbps with some proprietary radio protocol but I'm not sure that chip would handle it for long time. In addition it's practically useless unless you are in some kind of lab environment such (e.g. shielded chamber). Your questions indicate that your application is most probably not suitable for this range of embedded low energy ICs, are you sure you don't want to use some other off the shelf technology like WiFi?

Cheers Jan