S-parameter model for DW1000 diff traces


We are implementing DW1000 on our own PCB. As we chose RO4350 as our PCB core material and core thickness 0.254, the RF layout is different than provided by design files. We managed to take most of the changes into account and achieved pretty decent impedance matching. We work mostly on channel 4, 3.5 to 4.5 GHz and over the range the matching is from -15 dB to -10 dB (measured from balun 50 Ohm side) with the higher frequency range being the worse one. During the benchmarking of our board and tuning the matching circuitry, we noticed that DW1000 diff imput does not seem to be exactly 100 Ohm over the full frequency range. This brings me to question, whether there is some sort of s-parameter model for DW1000 available which could be used in simulation programs such as NI AWR. The model would be of great help when verifying the design and getting the maximum out of the IC. Also, it would considerably lessen the risks when PCB layout (traces lengths and widths) has to be changed due to component changes.

Hi SiimPille,

Are you able to measure receiver sensitivity and how does it compare with the performance listed in the DW1000 datasheet? The datasheet has sensitivity specs for channels 2 and 5. Channel 4 may be similar to Channel 2, though perhaps a bit worse due to the increased noise associated with the larger channel bandwidth.

This would be another way to validate the quality of the matching and possibly easier, compared to measuring with a VNA at frequencies this high.

When you say ‘tuning the matching circuitry’, can you describe, or draw, this matching circuitry? You should just have a pair of series DC-blocking caps on the differential lines between DW1000 and balun.

We have measured the receive sensitivity and it matches the datasheet pretty well, we have no packet loss in LOS conditions until -100 dBm, bigger packet loss starts at -106 dBm and we can receive some packets even at -110 dBm. As we have access to VNA and spectral analyzer measuring at those frequencies is not an issue. However, the purpose of this questions was that we want to achieve as good and as even matching in our RF circuitry as possible. This would allow for most even spectral density and therefore maximum amount of power transmitted within limits. The matching circuitry composes of those same DC block caps and one high Q inductor to equalize the balun and DW1000’s diff inputs impedances. The inductor is needed due to the fact that, as mentioned in my first post, DW1000 differential pair is not exactly 100 Ohm throughout the whole frequency range. As There are a lot of components/parameters (capacitor value, inductor value, trace width/spacing and even balun itself) that can be changed in order to changes the matching near balun and diff traces, it would be easier to have some sort of model to simulate some of the possibilities before changing the hardware. I hope this clears up the need for a model.