Connection speed

[yuval]

Hi,

My project is based on NRF52832 with SDK11.

I got 2 working units, 1 is advertiser and 1 central.

The advertiser is advertising every 2 second.

The central unit is trying to connect to the advertiser every few hours for few seconds to pass some data.

I found out that when the 2 units are near (up to 1m distance) the central finds the advertiser much quicker (about some few  seconds) than when the units are 3-4m or more distance away.

In that case it can take the central about 8-15 seconds to find the unit (in scanning).

I set the advertiser to +4 power.

Is there something in my software which can help the central unit finds the advertiser (in scanning) more quickly?

Also, Can you recommend about other possible solutions?

B.r

Yuval.

[Turbo J]

yuval said:

The advertiser is advertising every 2 second.

Waaay too slow if your intended average connecting (scanning) time was supposed to be in single digit seconds. You are off by about an order of magnitude, try something in the 200-300ms range.

There is a list of recommended BTLE advertising periods (measured in ms) from Apple. Try one of those first.

It is possible that your lab environment is noisy enough that some advertisements fail to work at longer distances. Being closer than 4m means that no other signals on the 2.4GHz band (like WLAN) could corrupt the BTLE data.

[Edvin]

Can you please try to capture a sniffer trace using the nRF Sniffer for Bluetooth LE? And keep the sniffer device close to the central that is trying to connect. Do one trace when it is very close (1m) and one when it is 3-4m or more, so that it takes 8-15 seconds. Can you save those traces as .pcapng and upload them here please?

Best regards,

Edvin

[yuval]

LOCK_Few_meters_away.pcapngLOCK_10cm.pcapng

Hi,

Thanks for your answer.

I shutdown the central unit and operate only the sniffer.

The advertiser is advertising every 2s some bursts - (It's a mechanism we built that every 2s it awakes the unit and advertises for some few 10 ms)

As you can see when the advertiser is close to the sniffer then the sniffer sees much more bursts......

I think this is the problem when the central is active, sometimes it connect fast and some times it take much more seconds to connect....(It sees much less advertise packets and for some seconds even none !)

Is there something i can do in my software to fix that? (i already transmit in +4). Are there any hardware recommendation to help the receiver sees more packets ?!

In my application the advertiser is located at one side of an iron door and the central is located on the other side so maybe that's explain the reduce in the RSSI.... and the missing advertiser packets.

[Edvin]

yuval said:

In my application the advertiser is located at one side of an iron door and the central is located on the other side so maybe that's explain the reduce in the RSSI.... and the missing advertiser packets.

I guess that is not ideal. 

Is this a new product under development? Or are you patching up some old device? The reason I ask is that the SDK version is very old. You should consider updating to 17.1.0 if you are developing something new. 

I see that the RSSI numbers (Received Signal Strength Indicator). When you took these measurements, was that also through the metal door? If it was, can you try to place the sniffer on the same side of the door as the advertiser (or open the door)?

The reason you are seeing that it takes longer to connect when you are further away is that the number of packets that it sees is a lot lower. 

If the sniffer trace that you uploaded is not through the iron door, then I would say that you should do a layout review. Perhaps there are some issues with the layout. If you'd like that, then open a new ticket and upload the schematics and PCB files (you may want to use a private ticket for that). 

If the RSSI is a lot better without the door blocking it, then the HW may be good, and you may want to consider moving the devices and see if the performance is better a bit further away from the door. I don't know anything about the location. But perhaps the signal is better through the walls than directly through the metal door.

Best regards,

Edvin