Data Tx Noise Spikes at 1 Second Intervals

[John Stuart]

I have identified some +500 count noise spikes on a PoE (Power over Ethernet) powered RShake.

Now that the Pacific Ocean wave microseisms have died down, my nighttime noise levels are well below 100 cts RMS,  I am now seeing periodic noise spikes on my RShake data stream.

When I activate the Swarm "Wave to Monitor" function, the RShake sends data bursts at 1 second intervals, and the 500 count noise spikes appear every 1 second.

Other periodic spikes are also recorded, probably related to the normal data transmissions.

This RShake is powered over about 100 ft of Cat5 cable with a PoE splitter (DC/DC converter) adjacent to the RShake.  It also has its geophone buried 4 ft underground in a 1" PVC pipe.

So the data related noise spikes could be caused by coupling between the Cat5 and geophone cable (Belden shielded twisted pair. shield connected to RPi's USB jack housing.), or

the spikes may be related to a momentary drop in 5VDC power voltage into the RPi when data transmission starts, or

caused by something else??  (I have never used WiFi on this RPi / RShake, and it does have the Ver.2 auto update.)

Any ideas what may be causing these spikes?  I will continue to investigate, , , 

John Stuart

Lafayette, CA,

 

[Branden Christensen]

Hi John:

My hunch is that your POE is not providing enough juice. I saw some similar spikes during development when we were using 2 amp 5 V supplies. Moving to 2.5 Amp supplies solved the problem. If you were a magnitude 9 backer, then I would try using the power supply we sent you (they were only included with M9s) and see if the problem goes away. Otherwise, I would get my hands on a 2.5 or 3 amp 5 volt power supply. This is what we use: http://www.suntechpc.com/store/index.php?id_product=34&controller=product&search_query=%28Lot+of+250%29+5V+2.5A+Micro+USB+AC%2FDC+Wall+Power+Supply+Converter+Adapter+for+Raspberry&results=7

Yours, 

Branden Christensen

Director, OSOP

USA: +1-206-414-1288; Panamá: +507-6747-1427

Get yours now! Visit shop.raspberryshake.org

--
Some useful links:
 
Manual: http://manual.raspberryshake.org/
Shop: https://shop.raspberryshake.org/
Website: http://raspberryshake.org/
Do It YourSelf Page: http://raspberryshake.org/do-it-yourself
---
You received this message because you are subscribed to the Google Groups "RaspberryShake" group.
To unsubscribe from this group and stop receiving emails from it, send an email to raspberryshake+unsubscribe@googlegroups.com.
To post to this group, send email to raspber...@googlegroups.com.
Visit this group at https://groups.google.com/group/raspberryshake.
To view this discussion on the web visit https://groups.google.com/d/msgid/raspberryshake/6ebaa12a-7e05-4fca-8929-6584166ede35%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Mercalli]

Not sure what that is John. My first thought was, the utility company using some sort of carrier wave communication over the power lines, but the frequency seems too low. Today I saw some spikes on my data I have never seen before. It looks like something counting, 1-2-3-4-5. Does not seem like cultural noise would only sent the spikes up. The sun came out today, so maybe some type of building expansion here (?)

[chris...@aol.com]

Hi John, 

    Try adding some ceramic capacitors across the 5 V supply line ? When you are trying to drive digital pulses into a high capacity line, you need a lot of current. 

    Regards, 

    Chris

From: John Stuart <qxst...@gmail.com>
To: RaspberryShake <raspber...@googlegroups.com>
Sent: Sun, Jan 29, 2017 7:22 pm
Subject: [RShake Forum] Data Tx Noise Spikes at 1 Second Intervals

I have identified some +500 count noise spikes on a PoE (Power over Ethernet) powered RShake.

Now that the Pacific Ocean wave microseisms have died down, my nighttime noise levels are well below 100 cts RMS,  I am now seeing periodic noise spikes on my RShake data stream.

When I activate the Swarm "Wave to Monitor" function, the RShake sends data bursts at 1 second intervals, and the 500 count noise spikes appear every 1 second.

Other periodic spikes are also recorded, probably related to the normal data transmissions.

This RShake is powered over about 100 ft of Cat5 cable with a PoE splitter (DC/DC converter) adjacent to the RShake.  It also has its geophone buried 4 ft underground in a 1" PVC pipe.

So the data related noise spikes could be caused by coupling between the Cat5 and geophone cable (Belden shielded twisted pair. shield connected to RPi's USB jack housing.), or

the spikes may be related to a momentary drop in 5VDC power voltage into the RPi when data transmission starts, or

caused by something else??  (I have never used WiFi on this RPi / RShake, and it does have the Ver.2 auto update.)

Any ideas what may be causing these spikes?  I will continue to investigate, , , 

John Stuart

Lafayette, CA,

 

--
Some useful links:
 
Manual: http://manual.raspberryshake.org/
Shop: https://shop.raspberryshake.org/
Website: http://raspberryshake.org/
Do It YourSelf Page: http://raspberryshake.org/do-it-yourself
---
You received this message because you are subscribed to the Google Groups "RaspberryShake" group.

To unsubscribe from this group and stop receiving emails from it, send an email to raspberryshak...@googlegroups.com.

[John Stuart]

Upon further investigation, , , ,

The PoE splitter was somewhat responsible for creating the spikes, and the extended sensor cable was a willing accomplice (helper).

When I removed the extended sensor cable, the spikes were no longer visible, but there was still a 1 Hz FFT spectrum peak.

When I removed the PoE splitter, and powered the RShake with a 2.4 Amp USB batter pack, the spikes were gone!

Questions:

1. Are the spikes caused by a momentary voltage dip in the 5 VDC power bus causing a dip in the ADC converter reference or input signal conditioning? 
2. Where on the RPi or RShake boards can I attach a capacitor to the DC power bus (per Chris's recommendation)?  What size?
3. Where is the best place to ground my extended sensor cable's shield drain wire?

Below are some photos with valuable lessons for anyone doing a PoE installation:

Initial installation with coiled Cat5/power cable and grey sensor cable.  The RShake was recording 800 to 1,000 counts spikes when data was transmitted (see RFD2F on 2017/01/28).

PoE Splitters: Note bottom splitter has a section of shared data & power cable.  Top splitter has separate data vs.5 VDC power cable.

Data spikes disappeared when using this 2.4Amp USB battery pack. Conclusion = momentary current surge was the root cause of data spikes.

As-left configuration: Cat5, 5VDC power, and Ethernet connection cables are separated from the grey extended sensor cable. The sensor/geophone is 4 ft. underground, down the 1" PVC pipe visible at right..

The data spikes have been greatly reduced, but not eliminated.  

Well, that was fun.  Hope I didn't miss any earthquakes. ;)

John Stuart

Lafayette, CA

[Branden Christensen]

Hi John:

Buenas tardes.

I owe you an answer to some questions. I had to talk with our electronics engineer first to confirm some things. Here ya go:

Questions:

1. Are the spikes caused by a momentary voltage dip in the 5 VDC power bus causing a dip in the ADC converter reference or input signal conditioning? 

The power input to the Raspberry Shake was made as robust as possible.  However, power spikes and dips can and do still affect the Raspberry Shake.

 

Also, running any long cables to/from the 'shake will cause issues.  The Raspberry Pi (RPi) itself does not have sufficient filtering on its power bus to get rid of the noise coming in over long cables.  This is then passed to the Raspberry Shake.  The Raspberry Shake filters much of it out, but not all.

A long cable to a geophone is an issue unto itself. Wires must be twisted pair.  

1. Where on the RPi or RShake boards can I attach a capacitor to the DC power bus (per Chris's recommendation)?  What size?

See response to next question. 

 

1. Where is the best place to ground my extended sensor cable's shield drain wire?

Shields are best connected at the geophone end but this is problematic in this case.  Thus there is a point on the board where the attachment of a shield was foreseen.

Attached is a picture of the top of the Raspberry Shake-v4.  Marked are the spots where an input filter cap can be attached and where a geophone shield can be attached.

The shield must not make any electrical contact to anything grounded, nor to either signal wire.

The input power capacitor placement is directly on the +5V from the RPi.  We are not sure how much capacitance that the RPi will allow on this line.  However, we expect that up to 50uF can be tolerated. If we had the schematics of the RPi 3 Model B, then we could look into this, but I don't think that the RPi organization have released full schematics.

Branden

[Mercalli]

Thank you John and Branden
There is a lot of good "trailblazing" information here that some of us will use to great advantage on our installations.
Some of this exceeds the original goal of a network digital seismometer you can place on the floor under your desk,
but the instrument performance also exceeds what was expected, (at least as I and many others see it).

[jeffersonite]

Hi John,

I've seen these in short period (L-22) stations. The sensor is acting like an antenna and receiving impulses whenever the modem or some other source of radio is sending data.

You can try to increase the modem and sensor separation, or create a Faraday cage around the sensor (and possibly its cable). The former is an easy fix. Let me know if try the latter.

Jefferson

[kw123]

"" It also has its geophone buried 4 ft underground in a 1" PVC pipe. ""

how did you do the mechanics of 

1. drilling a deep hole

2. attach the mic to what at the enc

Karl