Forum - Recent Threads

L2CAP with BLE and STM32

yosoufe — Tue, 05 Jan 2021 19:01:01 GMT

It has been few weeks that I am trying to learn about Bluetooth low energy (BLE). This is the first time I code for bluetooth and I spent a lot on learning the basics. I managed to get my python script to communicate with the Nucleo board via the BLE expansion board. It seems the BLE is not designed for relatively large data. Every packet is like around 20 bytes of data. I read about L2CAP which seems it can increase the throughput. But I do not find an example or any code to use the BLE and L2CAP. L2CAP is marked as supported in the BLE exansion board but I do not know how to activate it and how it works.

My purpose is to be able to send the ADC measurements over BLE to the PC and do the rest of calculation on PC. It has been impossible to have a good speed of data transfer for this purpose. For example when I send 250 bytes from nucleo board to PC without L2CAP, it takes more than 1 seconds. The alternative is to do calculations on the nucleo board and use the BLE just as a notification.

If you know about the L2CAP and BLE please help me. I do not find anyway to send data using L2CAP with the bluenrg library as a server. bluenrg library has few functions to receive L2CAP stuff or sending as a client, but I could not find anything to initiate the communication over the L2CAP as server.

Thanks

Yousof

RE: L2CAP with BLE and STM32

yosoufe — Fri, 08 Jan 2021 06:19:00 GMT

As I understood, the BLE sends in packets of 20 bytes, So the 250 bytes will be sent in 13 packets.

Do you think I have something wrong or inefficient that takes so much ling? I will double check that then, Thanks.

RE: L2CAP with BLE and STM32

yosoufe — Fri, 08 Jan 2021 06:16:46 GMT

Hi @shabaz, Thanks

Correrct. The bluenrg library that is provided by ST does not seem to have L2CAP-suppported high level APIs (or I could not find). Maybe because cortex M7 is not powerful enough? Or it is not provided yet? I am not sure.

I was not also thinking that I could write them myself

RE: L2CAP with BLE and STM32

shabaz — Wed, 06 Jan 2021 13:56:49 GMT

By the way, here is a very simplistic diagram.

If you're experiencing 1-second delays to send information, it could be that the higher-level API calls you're using, are causing advertisements and characteristics discovery procedures each time (which all filter down the stack from the location where they are initiated in each state machine in the stack), prior to sending your desired application data. All of those procedures take the times specified in the BLE spec.

You may need to change what API calls you are using, or pass particular handles for subsequent requests to send your application data, or some other parameter. But you'll still have to work at the high level API, it's normally not possible to take a short-cut and try to send function calls mid-way in the stack directly to the L2CAP layer, that will break things.

RE: L2CAP with BLE and STM32

shabaz — Wed, 06 Jan 2021 13:41:34 GMT

Hi Yousof,

I have not used the particular hardware or software that you mention, however generally it is not expected to be necessary to code at the L2CAP level. The software stack for BLE for your processor/environment, or firmware inside the RF function inside the chip, will do all that. You normally only have access at a higher level API, which will interact at GAP and GATT level, if not an even higher level friendlier abstraction.

When you make the API calls that will cause interaction at these higher levels, and the BLE software stack will automatically convert the high level instructions into the low level L2CAP layer.

It is easiest to imagine it like I2C communication where you normally send bytes to your i2c_write function call, and do not need to toggle the SDA/SCL lines individually.

The high level API will translate down internally in the stack to the low layer, and will then be handled accordingly over the air.

If you're actually not interested in using the supplied BLE stack and just want to transmit raw bytes and handle your own error detection, retransmits, your own channel selection etc., then the chip may have a 'raw' mode, but then that is not BLE, the stack is too complex to code, and such a raw mode will not work with a PC expecting BLE communications.

Anyway to un-publish the blog?

yosoufe — Fri, 25 Dec 2020 20:58:18 GMT

Hi,

Is there any way to un-publish a blog? I am writing a blog and save it as a draft. And sometimes I click on "publish" instead of "save as draft" by mistake. Is there anyway to undo it?

It is not possible to easily delete the blog and recreate it, because I need to upload the images and videos again. Otherwise clicking on the copied image, results in "image not found" error message.

Even a pop question, asking for confirmation like "You are going to publish the blog. Are you sure? Continue?" or something like that would be useful to avoid publishing by mistake.

Thanks

RE: Anyway to un-publish the blog?

yosoufe — Sat, 26 Dec 2020 22:35:06 GMT

Thanks. I did not know that. That is useful.

RE: Anyway to un-publish the blog?

Andrew J — Sat, 26 Dec 2020 20:07:02 GMT

Tthere isn’t a way to unpublish and this does catch people out - Jan’s suggestion is the best for when you do it. The workaround we’ve settled on is to set a publish date as soon as you create the blog post, before typing anything. Click on advanced at the bottom, then set the publish date a long way into the future. That way, even if you do press Publish instead of Save as Draft, nothing will appear until the date. When you are ready to actually publish it, then just remove the date and publish.

RE: Anyway to un-publish the blog?

yosoufe — Sat, 26 Dec 2020 07:26:35 GMT

Thanks,

I tried delete. But delete, means uploading all the images again and also there is a chace to click the publish again by mistake.

I am going with the "DRAFT" option.

RE: Anyway to un-publish the blog?

kmikemoo — Sat, 26 Dec 2020 01:16:12 GMT

yosoufe I haven't used it myself, but have you tried the Delete button on the right side of the screen?

RE: Anyway to un-publish the blog?

Jan Cumps — Fri, 25 Dec 2020 22:07:16 GMT

You can't unpublish. When it happens to me, I put a big DRAFT message on the first line in red font.

RE: How are you going to connect the sensor?

dwinhold — Fri, 27 Nov 2020 19:32:10 GMT

In the instruction paper that came with the sensor, it shows an alternative solution for connection.

Dale Winhold

RE: How are you going to connect the sensor?

ralphjy — Fri, 27 Nov 2020 16:20:04 GMT

Thanks. I'm afraid I'm guilty of a case of not RTFM sufficiently. I had downloaded that document and used parts of it already. Appreciate that you pointed out that it had the frequency curve included. I also missed the part about the resonant frequency being >30kHz. I think I need new glasses .

Ralph

RE: How are you going to connect the sensor?

michaelkellett — Fri, 27 Nov 2020 15:08:11 GMT

I'm one of the judges so I have to be careful what I say but I think it's OK to point out that you can find a spec with frequency response here:

https://www.kemet.com/en/us/sensors.html?t=931

RE: How are you going to connect the sensor?

neuromodulator — Fri, 27 Nov 2020 01:41:02 GMT

That's good, I suspect the STM32 is one of the best alternatives, because of the number of bits and speed. I haven't done much work with the STM32, but maybe I'll give it a try to see if I can get it to do what I need without having to spend too much effort on it

How are you going to connect the sensor?

neuromodulator — Thu, 26 Nov 2020 22:53:35 GMT

The only requirement for the "experimental design challenge" is to use the vibration sensor. How are you going to use it? I suppose most (if not everyone) will sample the analog signal, but what are you going to use to sample it? Are you going to use the provided STM32 MCU, a different MCU. a DAQ, or something else?

As of me I still haven't decided yet. I got a NI DAQ, a DMM with sampling capabilities and of course a bunch of MCUs and FPGAs. the NI DAQ is probably the easiest to use, its 200 ksps @ 16 bits. Then my DMM could sample up to 1msps @ 16 bits, but not continuously, or at much higher resolution but also a much slower speed. Then the provided STM32 can sample at up to 3.6 Msps @ 16 bits, (and eventually at higher speed through interleaving) but requires much more effort as it needs to be programmed.

RE: How are you going to connect the sensor?

ralphjy — Fri, 27 Nov 2020 01:32:32 GMT

Hi Doug,

If you look on page 4 of the manual - the suggested connectors are SM03B-SRSS-TB or BM03B-SRSS-TB, these are surface mount PCB connectors. I think even if you found the cable connectors they would be too hard to put together unless you had a crimp tool that could handle the small pins. I wasn't able to find any pre-built cable assemblies so I'm going to try the PCB connectors to see if I can build an adapter. Probably simplest to just cut it off. It almost looks like they expect you to do it as there are instructions in the manual on page 5.

Ralph

RE: How are you going to connect the sensor?

neuromodulator — Fri, 27 Nov 2020 01:22:12 GMT

I also noticed its way too tiny, so I'll eventually cut it off and use something more convenient

RE: How are you going to connect the sensor?

neuromodulator — Fri, 27 Nov 2020 01:19:51 GMT

Yes, but through oversampling you can reduce noise, so its not a bad idea to go for high sampling speed

RE: How are you going to connect the sensor?

zst123 — Fri, 27 Nov 2020 01:15:19 GMT

I will say go for it and use the NI DAQ if that is what you are most familiar with. Personally I am going to use the STM32 ADC sampling with DMA as it is something I have experience with. After this competition it will be nice to learn from each other's blogs on how a variety of tools can be used.

RE: How are you going to connect the sensor?

dougw — Fri, 27 Nov 2020 01:13:37 GMT

I will hook it up this weekend and see what I can do with it. The first issue for me will be connecting to the tiny connector - I don't think I have a mating connector yet. It looks to be a 1mm pitch. I will probably blog as I go in case it helps anyone else.

RE: How are you going to connect the sensor?

ralphjy — Fri, 27 Nov 2020 00:32:39 GMT

The sensor spec indicates the frequency range is 15KHz and the measuring range is about 1V, so your NI DAQ has more than enough capability. I wish the spec had a frequency response curve but I guess that's something to be determined. There's also no power supply sensitivity curve although the operating range is specified as 3-5.5V (I plan to use it at 3.3V).

How does the sensor communicate?

parasquid — Tue, 06 Oct 2020 20:16:10 GMT

Is it through voltage fluctuations?

I can see that there is only one wire as the signal. Is it modulated somehow, or is there a carrier wave?

I wasn't able to find the information in the datasheet linked to the article, but maybe someone can point me in the right direction?

RE: How does the sensor communicate?

parasquid — Wed, 07 Oct 2020 04:56:51 GMT

Thanks, that's great to know!

RE: How does the sensor communicate?

genebren — Tue, 06 Oct 2020 22:29:55 GMT

This is from the KEMET catalog and shows the frequency and amplitude range (along side a range of example applications). Ideally, you would sample the sensor at a high rate and perform a FFT to identify the amplitude at different frequencies.

I hope that this helps.