This post is older than 2 years and might not be relevant anymore
More Info: Consider searching for newer posts

Is measuring RSSI enough for mass production?

Hi,

In production line, I was supposed to do BLE test using DTM with UART interface and MT8852B from Anritsu.

However, the antenna is specifically tuned for plastic enclosure assembly.

After assembly, UART interface is no longer exposed so I think I cannot conduct BLE test using DTM.

I can just do BLE test before assembly but I doubt that whether I can get meaningful test results without the enclosure.

So I think the only way to test BLE after enclosure assembly is measuring RSSI and pairing test.

I would like to ask that it's enough and safe for mass production instead of using DTM.

Thanks.

  • RSSI is normally a composite number for all BLE channels. Doesn't really represent the variations you will likely see across the whole ISM band. You can use RSSI for power output if you do custom code and measure it while exclusively on a channel. Your test jig will have a big affect on the numbers.  Measuring near field power can be plagued with inaccuracy.  Small changes in the position of the EUT or someone just placing their hand in wrong place can make the numbers go up or down. Spend time on making a nice reliable test jig.

    Also, RSSI says nothing about channel accuracy and receiver functionality.

    A minimum test plan would be power output and channel accuracy on 3 ism channels (low, mid, high) and a similar receiver test with current monitoring so you can verify the board draws the correct current during these tests. Then you just need some sort of functional test where some packets go back and forth from test station to EUT.

  • Do you know any low-cost equipment to test output power and channel accuracy on advertising channels?

    I think your idea is good, which enables BLE test even when the PCBA is assembled with the enclosure.

    But I need to know which equipment can do the minimum test plan without any "wired" interface.

  • The absolute cheapest approach would be to code a standard DK as your master.  Then buy a spectrum analyzer for channel power measurement and frequency accuracy.  You can write the whole test plan in LabView and most good spectrum analyzers are supported in LabView.

    You will have to invest a lot of effort into setting up your test solution. Normally I would dedicate an ATE Engineer to the task and even then it would take 3 months or more to build it.

    A reasonable quality spectrum analyzer can be bought for under $10,000 USD. You will need a GPS 10MHz reference for the frequency data to have any meaning.  Then you will have to buy directional couplers and combiners to have the spec analyzer and the master DK on the same signal source.

    I think LabView is about $5,000 a seat. If you are really good in windows you could probably write the thing in visual basic  or python instead.

  • I suppose the absolute cheapest approach is to use the DK to measure RSSI.  But as I said it is only relevant and a single channel at a time.  You don't want a composite number.  And this method does nothing for measuring frequency accuracy.

    However, RSSI measurements vary by hardware.  Everyone implements differently and rarely are they based on absolute power level.  Instead RSSI is normal based on relative noise floor and received signal to strength ratio.

    The spec analyzer solution is best/cheapest.  But really cheap would be rssi on dk. Depends on your threshold for pain if your customers start getting marginal hardware.

  • Remember that our chips are tested and screened in our factory before they are shipped to distributors/customers. So as mentioned above a minimal test level would be to use rssi or an attenuator to check that the tx power level is higher than a given threshold. (chek the level of a good product and use that as a reference when setting the limited for screening)

    You could also use a proprietary input to change channels/mode during dtm testing as explained in this app note from anritsu.

    There are also companies like Litepoint that can measure tx power and rx sensitivity +++ based on advertisement and scan response packets.

Related