Small outdoor vault


I made a small outdoor vault (still in progress). The concrete pad surface is 14 inches across, and is 28 inches below grade. The pad is mostly 2 inches thick but the center part has a 6 inch cylinder going about a foot down. All told it used about 80 pounds of cement. Underneath the corners of the pad are 3.5 inch boreholes going 2 feet further down filled with sand, to help drainage. The sidewalls of the enclosure are made from 8x16x2 inch blocks and are just outside the pad, so they sit on surrounding sand and not directly on the pad. This was actually intended for a different device, but while I’m still getting that ready, I put a R-Shake AM.R79D5.00.EHZ into it to see what I could see. It uses a Pi Zero because I was originally thinking of putting it in a small diameter borehole and wanted the smallest possible package. Then I found out at my location even a 40 foot borehole wasn’t much lower-noise than the surface, and had the disadvantage of being underwater.

The vault is right next to a road, only about 5 feet away from the road edge. Not an ideal location, but the easiest place to build this, and during the night there isn’t much traffic. I knew I would see cars and trucks, but I was surprised to find that it easily picks up both pedestrians (each footfall makes a clear spike) as well as bicycles, due I guess to the texture of the road surface. Even though it’s not terribly deep, I notice the reported CPU temperature only varies +/- 1 degree C from day to night. I have a second RShake outdoors above ground in a small enclosure and it varies over 10 deg. C.


here is a closer view. Because I used a Pi Zero I also needed an external USB-Ethernet adaptor, and I am using an external PoE -> 5V converter as well.

Avoid or filter out occasional spike or glitch?

Here is an example of a pedestrian on the road. This one is me but everyone I’ve seen has a similar walking pace, about 2 steps per second. I’ve noticed people walking together often walk in step.


How does the spectra look?



Here is an example: two people walking, a car, and then a jogger. All the foot traffic was along the far edge of the road, so about 40 feet away at the closest point of approach to the R-Shake, which sits underground on an 80 pound concrete block. It is impressive to me how sensitive a detector this is.


If you meant background noise, here’s an example around 3 am local time with no cars passing immediately nearby (but maybe some farther away). The two constant tones at 19 Hz and its 2nd harmonic are likely from some HVAC fan nearby. In the bottom plot of the spectrum there is a peak between 0.2 and 0.3 Hz which may be the background microseism from offshore wave action.

1 Like

Last night at 11:48 pm (PDT) there was a small M2.0 quake centered only 3 km away from me, but 45 km deep. The R-Shake in the vault, R79D5 was able to detect it, as did my other nearby device on the surface RF7DC.

Note: USGS says this happened at 6:48:22 UTC. I wrote the wrong number on the graph, on closer examination my first peak is at 6:48:26.932 UTC on RF7DC and 6:48:26.922 UTC on R79D5 so 45 km in 4.9 seconds means a velocity of 9.2 km/sec (which still seems a bit high).
Time: 2019-05-05 06:48:22 (UTC)
Location: 45.443°N 122.639°W
Depth: 45.0 km

1 Like
How accurate should the timing be?

Hi @jbeale,

You’re right, 9.2 km/sec is at least a couple km/s too high. The residual error on the location and depth solution is on the order of ±10 km, which probably accounts for a good deal of the problem. This is likely because there’s not enough broadband station density to give a good depth solution. You can see how many stations are used for this solution here. As you can see on the Phases page of the quake summary, they weight picks from stations as far away as Hood (!!) for this quake pretty highly. This is because, although Portland is instrumented with quite a large number of strong motion sensors, the broadband density is still dismal.

This is a good demonstration of why network density matters. Your Shake was sitting pretty close to directly above the hypocenter, and thus your picks are probably more accurate than what PNSN or USGS can come up with in a solution that’s weighted so heavily on stations ~80 km away. Cheers!

If you want to calculate a more accurate hypocenter solution (for example in SWARM) using the Shakes in the Portland area, it’s probably safe to use a P velocity closer to 5 km/s as detailed here.


1 Like

Hi Ian, thank you for this very informative response and links! Yes, I see my station is quite a bit closer than any they used for their solution. I was not imagining that I would have any better information than the professional network, but it seems that location does count for something.


Another observation: my underground vault is at the front edge of the yard, which is 35 feet from the nearest point on the foundation, and about 50 feet away from the center of the house. When I simply walk around inside the house, even on the 2nd floor, the sensor still picks up each footstep. Quietly standing up from a sitting position on a chair is still visible. From one point of view, the Shake is impressively sensitive. On the other hand I had hoped the outdoor location would have better isolation from what I think of as fairly low-energy indoor noise.

tinfoil hat tip: if you’re worried the neighbors may have a R-Shake and are snooping on your movements, just run the clothes dryer- in my case, that causes enough noise to drown all this out!

1 Like