The CIR results read from ACC_MEM I found so far and also our own tests lead to CIR data with a first peak around sample #750. Before the first arriving path, there is just noise at the channel, which means there is no useful information at all. Therefore, I wonder why there is such a long noisy period in ACC_MEM. What is the trigger to fill ACC_MEM? It can’t be related to the distance, because the receiver has no knowledge about at the time of reception of these samples. Is there a way to move the first path to maybe sample #100, which would enable a longer part of the CIR in ACC_MEM?
Hello @AxelS,
The ACC_MEM window is not positioned based on the actual propagation distance. It is filled from the receiver’s internal preamble accumulation/CIR estimation process after preamble acquisition and SFD processing. The receiver does not know the true range at that point, so the first path is intentionally located with a large pre-trigger/noise region to allow the CIA to search for early paths and maintain timestamp margin.
Seeing the first path around sample ~750 is therefore expected behavior, not an indication that the receiver waited for that distance before filling ACC_MEM.
There is no supported register setting to arbitrarily shift the first path to sample ~100 in ACC_MEM. The useful approaches are to read the region of interest around the reported first-path index, use the CIA diagnostics/first-path index registers to locate it, and process only that portion on the host.
The way I’ve always viewed this is that the system has to estimate where to start the CIR accumulation buffer based on an initial guess. Since that guess is by definition made before correlation has been done it’ll be triggered mostly by raw signal strength. This means it could have triggered on a reflection and so be late. It’s very unlikely this initial guess is going to be significantly early.
If you know your initial guess is more likely to be late than it is to be early it makes sense to align your buffer to allow more time before the guess than after.