Documentation

We found a bug in the demodulator circuit. The 1.0uF capacitor was too big, which meant that reader commands were decoded incorrectly, particularly when decoding long commands (such as READ) or many commands in sequence (such as the Gen 2 singulation process). Changing the capacitor (C2 in the PCB layout) from 1.0uF to 0.1uF fixes this problem.

demod_1.0uFcap.pdf illustrates the problem with the large cap; bit_line drops below bit_pwr. It shows the WISP trying to decode reader commands, including a long series of QReps, most of which are missed.

demod_0.1uFcap.pdf shows the same experiment with the new cap.

newcapread.pdf shows the impact the change has when using the READ command to return sensor values; using the old cap collapses at a certain point, where a WISP with the new cap degrades gracefully.

The CAD files have been update to reflect this change (version 8).

2. LED Orientation

Our board assembly house likes to stuff LED backwards. The dot in the LED should be nearest the demodulator

3. Seiko S1000 miss labeled footprint

The schematic specifies the S1000C20-14T1 (which is incorrect), however the BOM is correct.

The correct part number is the S-1000C20-N4T1 which has the SC-82AB footprint.

The CAD files have been update to reflect this change (version 8).

Comparison of voltage supervisors

In response to the low availability of S1000 devices, here is a list of some 2.0V voltage supervisors that would be suitable for the WISP. Please note that not all devices listed are directly compatible with the footprint used on WISP. See the package compatibility notes.

Firmware Snapshots and Applicable Reader

Intended for Impinj Speedway Reader, may also work with ThingMagic Development kit reader (using Intel reader chip).