colin
/
piezo-ddr-controller
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

readme: convert relative angle-bracket links to named links so they render

This commit is contained in:
colin 2022-06-30 23:57:49 -07:00
parent a53f273d5b
commit 99dc665ab0
1 changed files with 4 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
README.md
View File

@ -26,9 +26,9 @@ WORK SO FAR
Method 1a is demonstrated. It works on rigid surfaces. See [itm-on-table-drop-30lb-halfway-through.wav](doc/data/itm-on-table-drop-30lb-halfway-through.wav): this was achieved using 181.5 clock single-ended ADC sampling with delay(1) between each sample with no external amplification. The piezo was held in place using an empty water glass.
On pad: <./doc/data/itm-on-pad-heavy-hit-halfway-through.wav> demonstrates the same setup on a Polish pad on carpet but with a full glass of water.
On pad: [itm-on-pad-heavy-hit-halfway-through.wav](doc/data/itm-on-pad-heavy-hit-halfway-through.wav) demonstrates the same setup on a Polish pad on carpet but with a full glass of water.
<./doc/data/itm-on-pad-tape-big-hit-4clock-1024buffer.wav> and <./doc/data/itm-on-pad-tape-big-hit-4clock-1024buffer.png> shows what we get with a BIG stomp if the sensor is taped to a pad
[itm-on-pad-tape-big-hit-4clock-1024buffer.wav](doc/data/itm-on-pad-tape-big-hit-4clock-1024buffer.wav) and [itm-on-pad-tape-big-hit-4clock-1024buffer.png](doc/data/itm-on-pad-tape-big-hit-4clock-1024buffer.png) shows what we get with a BIG stomp if the sensor is taped to a pad
it's really hard to get.
Tried using flat piezos and the L-shaped ones, both behave similarly. Really difficult to get reliable waves.
@ -40,14 +40,14 @@ Try:
- Borrow UW's oscilloscope
- Buy a cheapo scope?
<./doc/data/itm-sandwich-51medianfilt.png> shows results after taping a piezo to a flat surface, loosely setting an acryllic plate on top of it, and then tapping the plate moderately.
[itm-sandwich-51medianfilt.png](doc/data/itm-sandwich-51medianfilt.png) shows results after taping a piezo to a flat surface, loosely setting an acryllic plate on top of it, and then tapping the plate moderately.
There is median filtering on the device side, plus additional filtering on the PC side.
Median filter seems to be key. Even without the plate, the device can _detect_ stomps from 4-5 inches away, though not quite as cleanly.
Placing the piezo such that half of it is in air seems to help detection. I suspect if the whole platform is oscillating, I want the other end of the piezo somewhere that's more fixed, in order to detect the oscillation of the platform itself. Alternatively, I believe a longer piezo element attached all to the surface could help. A smaller element should still receive everything, but the high frequencies are attenuated. Perhaps then, a HPF could be installed to mitigate this attenuation?
Could also try replacing the device-side median filter with a more tuned filter (HPF)
<./doc/data/itm-pad-drop-weight-3in-away-covered-tape.png>: shows clear signal dropping a 30lb weight near the sensor, even when the sensor is fully attached to the pad. Whole signal is about 5 ms.
[itm-pad-drop-weight-3in-away-covered-tape.png](doc/data/itm-pad-drop-weight-3in-away-covered-tape.png): shows clear signal dropping a 30lb weight near the sensor, even when the sensor is fully attached to the pad. Whole signal is about 5 ms.
51-median filt device side
5000-sample DC removal PC side
timescale is accurate