Hi Scott,

That is an awesome project. I use a lot of Ubiquity gear, and they have a non standard 24V PoE. Would the circuitry still work with 24V instead of 48V?

Bas

@Maximilian-Gerhardt Awesome! Thank you so much for putting this together!

Hello,

I'm rewriting some code that reads an I2C current sensor. In my rewrite I run into a segfault on an i2c write. The code in question is:

def pecmac():
    """
    Support for the Control Everything current sensor. Code has been tested with the 70A/3 channel model
    :return:
    """
    logging.debug('...... i2c create')
    i2c = onionI2C.OnionI2C(0x01)

    #set the verbosity
    logging.debug('...... i2c set verbosity')
    i2c.setVerbosity(0x01)

    command2 = [
        0x6A,
        0x02,
        0x00,
        0x00,
        0x00,
        0x00,
        0xFE,
    ]

    logging.debug('...... i2c command2')
    i2c.writeBytes(0x2A, 0x92, command2)

The first time I run this code, all is well. But on the second run (after a 60s sleep inside the loop) I end up with a segfault.

The output I get (with the actual data collection output as well is:

20180203 21:31:41 Detected our hostname as [OMEGA-F6BB]
20180203 21:31:41 Creating database connection
20180203 21:31:41 .. Done
20180203 21:31:41 Reading sensor data
20180203 21:31:41 .. Reading data from sensor PECMAC
20180203 21:31:41 ...... i2c create
20180203 21:31:41 ...... i2c command2
20180203 21:31:42 ...... i2c readbytes
20180203 21:31:42 Type of Sensor: 4 | Maximum Current : 70A | No. of Channels: 3
20180203 21:31:42 Channel no: 1 | Current Value: 0.376A
20180203 21:31:42 Channel no: 2 | Current Value: 0.000A
20180203 21:31:42 Channel no: 3 | Current Value: 0.000A
[{'Timestamp': 1517722302698L, 'Sensor': 'pecmac', 'Value': 0.376, 'Measurement': 'ch1_current'}, {'Timestamp': 1517722302698L, 'Sensor': 'pecmac', 'Value': 0.0, 'Measurement': 'ch2_current'}, {'Timestamp': 1517722302698L, 'Sensor': 'pecmac', 'Value': 0.0, 'Measurement': 'ch3_current'}]
20180203 21:31:42 Processing the sensor data
20180203 21:31:42 .. Sending data to UBIDOTS
20180203 21:31:42 .... Checking if Ubidots is Online
20180203 21:31:43 .. Submitting data to Ubidots
20180203 21:31:43 {"ch1_current": {"timestamp": 1517722302698, "value": 0.376}}
20180203 21:31:44 {"ch2_current": {"timestamp": 1517722302698, "value": 0.0}}
20180203 21:31:44 {"ch3_current": {"timestamp": 1517722302698, "value": 0.0}}
20180203 21:32:44 .. Reading data from sensor PECMAC
20180203 21:32:44 ...... i2c create
20180203 21:32:44 ...... i2c command2
Segmentation fault

Logread shows:

Sat Feb  3 21:32:44 2018 kern.info kernel: [ 1251.955047] do_page_fault(): sending SIGSEGV to python for invalid write access to 00000000
Sat Feb  3 21:32:44 2018 kern.info kernel: [ 1251.963673] epc = 77ce6864 in libc.so[77cbe000+92000]
Sat Feb  3 21:32:44 2018 kern.info kernel: [ 1251.968855] ra  = 779eac69 in libonioni2c.so[779ea000+11000]

Please note this is with some extra kernel modules to deal with I2C issues as per https://community.onion.io/topic/2312/i2c-detect-slave-method

tx
Bas

Hi,

An update, received the Onion shields and they worked without issues. Working with ControlEverything on their unit

Tx for the advice all

Bas

@ccs-hello said in Ethernet weirdness (no link with several switch):

You really should contact ControlEverything to ask them to perform a thorough test
(compatibility with many Ethernet switches) on that module they sold.
BTW, ask them to test using an Omega2. Omega1 supplies 2.5V on the transmitter side C.T., while
Omega2 does not (N.C. on that 2.5Vout pin.) They also use different Ethernet PHY internally.

Hi,

I had asked them to do that, and they replied they did test. But asked them to doublecheck it is indeed an Omega2

TX!
Bas

And GRRRRRRRRRRRRRR, same issue with the Trendnet.

@luz

Hi Luz,

My dayjob is in IT (with infra structure mgmt as my background) so deal with network issues fairly regularly. This one is curious since I tried:

• Multiple (cheap ish) switches
• Multi Omega's 2+'s
• Multiple ethernet shield (but same make/model)
• Multiple cables

There is 100% correlation to the switches used, any other variable can be changed without changing the result. The same switches have no issues establishing a link with a wide variety of other equipment.

Failed speed/duplex negotiations, poor speed as result of poor cable, etc etc, all fairly common. Don't think I have ever seen something like this.

Amazon just delivered me a cheap trendnet 5V power switch that people have used in IOT projects. I really hope that works

Might end up ordering one of the Onion shields to test

TX!
Bas

P.s, that must be an interesting deployment you have!

Hi,

Unless I missed something, the last Roadmap was up to July 2016. Is there any updates on what is in the pipeline. New modules in the works, software/firmware updates planned, etc

Tx
Bas

@ccs-hello Hi, the fact that it works with one switch without issues suggests (I think) that the hardware side of things is OK? I suspect there is something in the auto negotiation (Speed/Duplex) that fails

Bas

Hi,

Running into some rather curious issues. I have an ethernet expansion dock from ControlEverything (I actually have two, so tested with other one as well).

I found that the Omega is very picky in what it will built a link with:

• Dlink DGS-1008G, all good. Link is established after 10 secs or so and is stable
• Ubiquity Edgerouter Lite. Link comes up and goes down after random intervals (in the second range)
• TP-link TL-SF1005D. Link never comes up

The TP Link is the one I hope to use. Very low power and using 5V so great for combining with the 5V for the omega.

Firmware fully up to date.

Google has not turned up anything useful. All suggestions welcome!

Bas