DFE IQ sampling and antenna switching slots confusion

[Pete W]

Hi

I'm confused by the concept of switching slots and sampling slots, or at least the way they are presented as being interleaved. For BLE, I believe that they are required to be equal (1us or 2us). If they are both set to 2us and are interleaved then the antenna switch lines should change every 4us, but when I select '2us spacing' the lines toggle every 2us. This means that, with 2 antennas, the 2us sample slots - which apparently are 4us apart - always land on the same antenna and this makes no sense at all. If my lines are toggling at 2us then a sample must be grabbed every 2us, not every 4. Something is wrong. Moreover, it doesn't seem right for there to be such a 'long' allowance for switching, when the switchover is likely to occur in the ns range. I must be missing something. Surely it's just a matter of defining a switching period and a sampling period, and the sampling period can be shorter if over-sampling is required ?? 

Puzzled.

Pete, UK

[bjorn-spockeli]

Hi Peter, 

Is CTEINLINECONF.CTEINLINECTRLEN set or not? 

Also what is the TSWITCHSPACING and the TSAMPLESPACING set to in the DFECTRL1  register? It is by default set to 2us and 4us respectivly. 

Best regards

Bjørn

[Dmitry]

Hi Pete,

Pete W said:

but when I select '2us spacing' the lines toggle every 2us.

That's right, 1us/1us sample/switch.

Pete W said:

the 2us sample slots - which apparently are 4us apart

A bit confused... with 2us sample slot settings, you will get a sample every 2us.

[Pete W]

Hi Bjørn,

Thanks for the reply.

I should've added that I have CTEINLINECTRLEN set to 0. I hadn't investigated what happens in CTE inline enabled mode as I wasn't intending to use it that way (I thought I'd be stuck with 1 or 2 us timings). I am using AoA mode and I have TSWITCHSPACING set to 2us and TSAMPLESPACING set to 125ns, as I want to oversample for now. I currently don't do any antenna switching, but I am monitoring the control pins to see when they change. I see a continuous sinusoids in my IQ samples (at least after a 1us jump following the reference) so samples are being taken all the time, i.e. through the switching slots.

I'm really struggling with the switching slots. Is the antenna switch-over coincident with the centre of the switching slot, so that the slot's purpose is to allow for slow edges, or is there a switchover at the slot boundaries (as I seem to be seeing)? 

If I enable CTE inline mode, and keep TSWITCHSPACING at 2us, then it seems that the sample timing will use CTEINLINERXMODE1US, and the default for that is 2us (CTEINLINECONF  FFF=010). But the slots alternate! I keep going round in circles with this - nothing seems to add up for me.

Regards

Pete

[Pete W]

Hi Dmitry,

What I meant was that if the antenna switch slots are 2us long, and the sampling slots are also 2us long and fall between the switching slots, as per 5.1 core spec (fig below), then that suggests that sampling is only valid for 2us every 4us, otherwise what is the point of 'switching slots'? It also suggests that only 1 antenna change should occur in each switching slot. It would all be so much easier if the spec just said "switch at rate X, sample at rate Y. Avoid sampling within Z ns of switchover".

Pete

[Dmitry]

I agree, there's some inconsistency between spec and implementation - "Interval between samples" is not the same as "sample slot length". IMHO a "switching slot" just means that we should not use IQ data during transition period, but we can take them though, as nFR52811 does.