1-Wire Data Logger Ideas

[skyewright]

Pretty much all that sda said goes for me too.

PS. Forgot to add...

Configurable sampling interval and logging interval by device.
Some devices you might want to sample and log every minute or less.
Some devices you might want to sample every minute and log the average every 5 minutes.

For wind gust, it would be highly desirable for the gust to sampled more frequently than the logging interval and the highest retained through to the point of logging. Logging of that highest once a minute would be fine.

One of the appeals to me of 1-Wire over my existing OS sensors is the much better wind/gust sampling rate (the OS wireless wind gauges (I have WMR928 and WMR200) transmit at ~15 second intervals, and so far as I can tell the gust speed is assessed just once per interval, not maximised over the interval).

Regards, David

 

[dpenezic]

I plane to use NSLU2 from Linksys with some Linux distribution and OWFS. All other thing I will program myself. I understand that isnt very handy device, but for me will do a great job.

 

[steve055]

Hello Eric,
It has been some time since there has been activity on this thread. Not sure if it is still active.

Are you still giving consideration to developing a logger? Have you started developing your logger ?

I want a device that can record various sensor data "continuously" without creating great big carbon footprint.
Your idea of using a PLC is right on the money.

For my part, I am interested in a 1-wire standalone logger that can be periodically connected to a PC for data download.

* An Ethernet connection is probably easiest, though serial or USB is acceptable.
* Enough storage for a weeks worth of data, perhaps even a month (holidays) is another requirement.
* Capable of recording temperature and humidity, counters and voltages, and support for weather stations. Probably 40 sensors in total.
* Ability to run from 12V would be an advantage.

Regards, Steve
in the land of OZ

 

[skyewright]

* An Ethernet connection is probably easiest, though serial or USB is acceptable.
* Enough storage for a weeks worth of data, perhaps even a month (holidays) is another requirement.
* Capable of recording temperature and humidity, counters and voltages, and support for weather stations. Probably 40 sensors in total.
* Ability to run from 12V would be an advantage.

Earlier in the thread I mentioned Meteohub. Since then a "plug-in" interface has been added to Meteohub. All the plug-in needs to do is write appropriately formatted data to stdout.

Using OWFS and a bit of C++ code, I have a test plug-in feeding 1-wire data from a HB temp / hum / solar board and a barometer board to Meteohub. At the moment the code is very basic (just a proof of concept) but it does demonstrate the practicality of the general idea.
The Meteohub and the plug-in are running on an ALIX 1D that draws around 5W, and runs on a 12V supply.

 

[sda]

Data acquisition is a hardware/software issue, but data logging is strictly a software issue. IOW, getting the temperature/voltage/etc from the device requires specific hardware, drivers, and software. Recording that information at specific intervals over time depends only on software.

The data logger should only record values it gets from the sensors. Temp/voltage/on-off, etc. It doesn't need any logic to interpret, display or analyze the data. Those tasks are best done by whatever is the consumer of the data. ("consumer" is another one of those B.S. newschool compsci terms like ETL.)

My personal 1wire data logging runs on a Win2k server and logs to text files. It samples the sensors every 15 seconds and writes the 1 minute average to the data file using a directory structure of \data\2009\03\20\sensor.txt. Simple, fast, reliable. The web display software also runs on the server (IIS). The Win2k server is always on, so this setup is not using any more power than it would otherwise. The only problem with this setup is if I have to do anything to the server that requires a restart, the logging goes down during that time period.

My long term goal (just because) is to build a dedicated pico itx logging machine wth a SD card or even a SSD disk running xp/2k to do the logging. On a regular basis the logs would be shipped to the main server, or the main server can copy the logs. This is easily done using plain old file sharing.

Any standalone logging device is going to have to do four basic things with the data - capture, record, archive, distribute. These are primarily software functions. The advantage I see to doing data logging around Windows or Linux is that enhancements and fixes are easily done with a software update, and adequate replacement hardware is readily available should the logging machine fail. There's also the advantange that the basic components (file system, networking, etc) of the system are already in place and the logging process is just an application.

That's my 2 cents, which is what I had left after I asked Obama to give me change for $1.

 

[ericvic]

Hello Eric,
It has been some time since there has been activity on this thread. Not sure if it is still active.

Are you still giving consideration to developing a logger? Have you started developing your logger ?

I want a device that can record various sensor data "continuously" without creating great big carbon footprint.
Your idea of using a PLC is right on the money.

For my part, I am interested in a 1-wire standalone logger that can be periodically connected to a PC for data download.

* An Ethernet connection is probably easiest, though serial or USB is acceptable.
* Enough storage for a weeks worth of data, perhaps even a month (holidays) is another requirement.
* Capable of recording temperature and humidity, counters and voltages, and support for weather stations. Probably 40 sensors in total.
* Ability to run from 12V would be an advantage.

Regards, Steve
in the land of OZ

Steve,

The project had gotten shelved because of lack of time but I have started working on it again. I am not going to even guess when it might be ready because something else will probably come up to derail it.

I will release more details when I have things solidified a bot more.

Eric

 

[collectors]

I have another thread up at the moment. But a way to log multiple mains fuse board circuits for analyzing power consumption on each individual circuit would be nice with a 50 way circuit monitor for my own use.
But i am sure smaller units would be good as well.

 

[steve055]

I have another thread up at the moment. But a way to log multiple mains fuse board circuits for analyzing power consumption on each individual circuit would be nice with a 50 way circuit monitor for my own use.
But i am sure smaller units would be good as well.

Hi "Collectors"
I want to monitor power as well.

VERIS INDUSTRIES produce a Branch Circuit Monitor (H663)(www.veris.com). These devices can be configured to monitor 1, 12, 24, up to 42 branch circuits, and apparently VERIS can build them using 1-Wire comms, and they will also supply split-core CTs .......if only the bank would give me another mortgage. B)

 

[djbowen]

Eric,

My view would be to keep it simple. As others have mentioned do data logging and ignore the presentation/manipulation bit - this can be done with the PC.

I like your idea of very low power - something that could run detached occasionally would be neat. And outdoors would be better.

As for what to sense - a one wire collection role (aka act as a hub with delay) and direct sensing of voltage/current for external projects. External connectivity - ethernet or usb/serial. A 1-wire download routee might be cool.

I do similar stuff with NSLU2's but it is not very neat or weather tight - plus although it is low power it is not low enough for alternate power sources.

Happy to discuss further.

Dan

 

[geoff]

Different people will have different needs. I built my own logger, for what it's worth the thought process went like this ...
Requirements were ...

• 24x7 operation
• support data collection and also "outbound" control of ds2408 etc
• LAN attached (wired or wireless)
• tolerant of power loss
• customisable (I like to tinker ...)
• silent
• cheap (of course)

I had to compromise on most items. I bought one of these ... http://www.norhtec.com/products/mcjr/index.html plus a DS9490R plus a couple of Compact Flash cards. That plus my 1-Wire network draws about 6W of power, and I also added a baby UPS. From then on it's all software.

You need to borrow a screen and keyboard to get started, but as soon as the basic system is installed you can discard those.

I installed a minimal Debian system on the box plus OWFS. The owserver part of OWFS is an easy (?) interface to an attached 1-Wire device (at least it supports anything that I have on my network). owhttpd also gives you a web interface to view the status of all the devices. So far everything is free, open source software, but the next step needs a bit of effort. I wrote some PERL scripts which basically:

• scan the 1-Wire bus to see what's out there (that was trickier than I expected because it has to cope with a hub).
• use a config file to define what I want done with each device (ie what files to read, how often, and what to do with the data)
• loop (forever) scanning the various devices for data at various time intervals
• log the data to a database
• generate a "health check" html page which raises an alarm if a sensor is missing, or whatever

This logger script runs continuously. I have a separate and similar script to do "outbound" stuff, basically swicthing relays etc.

It works pretty well, there are lots of issues to overcome along the way. For example Compact Flash doesn't tolerate too many writes so I actually work entirely in memory, only copying out to the CF "disk" once a day (or whenever there is an alarm like the UPS is nearly out of battery). The data can be stored in any form you like, or just buffered and streamed to some other system for processing. I'm a backup fanatic, so there needs to be scripts etc to backup the data.

My setup is too personalised to be useful to anyone else, but with a bit of work it would be possible to come up with a product that was something like this.

If interested, a subset of the data (weather stuff) is up on the web at daveysbayweather.com There's a link somewhere on that page with some more description. Notice that my solar sensor has (temorarily?) died ...

 

[RadPa]

What I'm looking into a data-logger is in the configuration options (i.e. what to log).
- it should have the possibility to grab time-stamped sensor values / adc values / alarms statuses / switch statuses at defined (configurable) time intervals.
- it should have an option to record data when pulled from the devices above.
- it should have an option to record data only when the pulled values change.
- it should have an option to record "heart-beat" data (i.e. the values are not changing but need to make sure the sensor is still up there).
- it should have the option to save the results in 1-wire memory chips for a limited time (i.e. memory full).

paul.

 

[tparvais]

Hello Eric

Any news on this project ?

Thank you

 

[ericvic]

No news at this point. It is still on the road map but others things have had to take priority at this time. Hopefully will have something in the not too distant future.

Eric

 

[skyewright]

No news at this point. It is still on the road map but others things have had to take priority at this time. Hopefully will have something in the not too distant future.

In the meantime another possibility for folks to consider is something like a "Bifferboard". There are people using OWFS on that, e.g.
http://sites.google.com/site/bifferboard/H...lackware/1-wire.
The Bifferboard runs Slackware Linux and gets by on a couple of watts!

It says 1W on http://bifferos.bizhat.com/ but I think that may be in an ideal case? I use one configured as a Meteoplug and in practice I find power usage varies, peaking at just over 2W during ethernet activity and with a USB device connected, measured by having the "wall wart" PSU plugged into a power meter. Some of that extra is probably innefficiency in the PSU, and I can't claim that the power meter is super accurate either?

Just a thought...