Cumulus, USB, and 1-wire

[Charlie]

Well I did it.

After replacing my Fine Offset outdoor unit twice, the anemometer once, and having to modify my wind vane to get decent readings, I made the decision to build my own station. For me the criteria was simple – it had to be reliable and had work with Cumulus. But it also had to work within a reasonable budget… this is a hobby, after all.

About the only way to work with Cumulus easily was to use the “easyweather.dat” method, and I played with it for a few weeks, but it really wasn’t very satisfactory for all the reasons Steve regularly mentions, and did not allow for some of the newer functions like solar and UV measurements.

So after some discussion, prodding, and help from 41South, I bit the bullet, and set out to design and build a station with a USB interface that can talk directly to Cumulus.

USB is way harder than it looks. So a solid 3 months later, and a prototype station has been running happily on a breadboard for a couple weeks, with Cumulus collecting, displaying, and publishing data, and I’m starting to lay out a circuit board to eliminate the chicken wire and bubble gum currently holding things together.

The outdoor sensors are 1-wire based, mostly because 1-wire transports information nicely, and Colin had a handful of these devices. It does not have an RF link to the outdoor devices, although that could be added someday. I wanted to focus on functions and reliability first.

Since there is still time to influence the final functionality before I pay for boards, I thought this might be a good time to poll the folks in the forum for ideas. Do you have a favourite sensor I should support? Is there a function (perhaps software filtering or averaging) that should be included? How about power? Battery backup? Thoughts and opinions?

I confess I’d like to sell a few copies of this thing once it’s complete. I think there’s a big hole between the Fine Offset and the Davis devices in terms of quality / value, (certainly price!) and for those that like to tinker with hardware, there are not a lot of options out there, at least not options that fit easily with Cumulus. This could be the first commercial product designed specifically to work exclusively with Cumulus! (commercial being defined as more than one unit in use and a few dollars changing hands)

What do you think?

[mcrossley]

Charlie, I'd be interested to see what you come up with here. For me I would just want the basics (wind, temp, humidity, rain) plus, depending on cost Solar Rad & UV. Throw in cloud detection (day and night, though night would be of greatest interest) and I might be interested.

How are you doing the Cumulus interface, emulating another station?

[Charlie]

Hi Mark,

I'm emulating a Fine Offset 3080 ("Fine Offset with UV and Solar"). I chose that one to emulate simply because it seemed to not be missing any common sensors, it was relatively straight forward, and I had a real (semi-functional) "Fine Offset" so I could trace USB transactions to compare to help debug.

I have the same sized data logger capability, and it even works with Steve's tool to configure the logger interval. I toyed with making the memory larger - it was doubled for awhile - but decided there was little value in making it bigger, and indeed some risk of future incompatibility if Steve ever changes how he talks to F.O. devices, of if somebody decides to use some other software. It has capability to erase the logger if it ever gets corrupted somehow, as well as change the logger interval, but in normal use it just drops the oldest reading to make room for the newest once it gets full.

Regards,
Charlie

[mcrossley]

So are you using the Hobby Boards sensors?

[Charlie]

So are you using the Hobby Boards sensors?

As usual, the answer is "it depends".

Indoor humidity uses an HIH3610 directly connected to the microcontroller.
Indoor temperature, and system barometric pressure comes from a BMP085 sensor, which is extremely precise, and is also directly interfaced to the processor.

The outdoor temperature, humidity, solar, and UVI use Hobby Boards modules. Rain uses a standard counter – the Hobby Boards dual counter will work, in conjunction with the old F.O. rain gauge or one of several commercial devices that use the same counter chip. The anemometer and wind vane can be the original Dallas version, the new Hobby Boards ADS version, recycled F.O. devices, or roll your own. I've tried very hard to make it as flexible as possible, without creating the need for an end user configuration interface, so it detects what you are connecting (within constraints) and uses it.

[fadiaz]

Hi Charlie,

This sounds like an interesting project... please keep us advised on how it develops...

[Charlie]

Thanks, Fadiaz!

I have one last sensor to code - the rest have been working happily for some time. My PCB layout is about 75% complete as well, and really only being tweaked to fit the housing I want to put it in. I expect to have production quality stuff in about 3 weeks.

I am looking for advice and opinions, though.

Are there functions I should be including beyond those found in a 3080?
What sort of sensors do 1-wire enthusiasts typically use? Are they built or purchased?
How important is a wireless outdoor unit? (wireless is a huge source of erroneous data)
What has been people's experience with "wired" weather stations? Good/bad beyond the obvious install issues?

[nitrx]

I don't understand why you're doing this, is it cheaper ? The sensors well I mean the windvane, rainunit and anemometer must also be availble for input ?

[Charlie]

Nitrx,

There are several suppliers of 1-wire based sensors including anemometers, wind vanes, and rain gauges. I'd probably make my own if it was not for the up front costs of molds and other start up fees. The advantages (I am told) of the commercial 1-wire instruments is reliability, ease of repair if they do break, and in a lot of cases improved acuracy. Most people with low cost stations (F.O.) complain of these issues, as well as data spikes, noise, lost contact, battery life, USB hangs, and so on.

My personal F.O. experience included most of these issues, but for me, the cost of replacing failed bits fairly regularly made me want something more reliable, with better long term value for my cash investment. My budget, however, does not run to a Davis with all the bells and whistles - this is after all just a hobby, not critical to my health or business.

I design things for a living, and when I looked through the F.O. design, I realized a lot of it is pretty good, with a few fundamental flaws that could have been fixed for a small incremental cost during design time, but are now unrepairable. So I decided to do my own. The 1-wire direction was chosen after some prodding and support from another forum member. Worst case, he & I will have pretty nice stations. Best case, several others will too.

[nitrx]

Thanks for you answer Charlie, I know the FO-stations are not the best but for the price I can't expect more, but I'm sure following your interresting project (it wasn't a question of negative purpose but of curiostity)

Ron

[fadiaz]

Hi Charlie,

First of all thanks for your reply and thanks again for your work in this hard project.

I do not know if this kind of input is what you look, but I will try to answering some of your questions to guide you better in this project. BTW, I also like much that you take into consideration not only your needs for this project, but needs of other possible users.

In the budget range of personal weather stations, basically what I personally look is the basic functions/features, like what the FO WH-1080/3080 stations have. Indoor Temp/humidity; Outdoor Temp, Humidity, Wind speed and direction, Pressure, Rain. Solar readings not a requirement but a plus.

Having some kind of display or console to see the readings without a computer is important, in case I loose the PC (this becomes more important during events like hurricanes). But I guess if the price is right, one could manage with just the PC and cumulus reading.

I have mostly used weather stations that are wired to an external sensor, and send wireless the data to the internal sensor/console; and that is the setup I mostly prefer. But again, if price is right, I guess a complete wired station (like yours) will not be a "deal breaker" despite my preference, if all other requirements are to my liking.

The ability to get replacement sensors (easy and cheap) do would be a great plus also. Stations do require occasional maintenance and is sometimes frustrating to buy a 100USD station to then replace a small one component at a price of 50USD plus shipping.

Thinking wild here... maybe having a station capable of both, completely wired and wireless from outdoor to indoor might be a good feature But I guess will also add more on the development and selling costs...

BTW, the reason I like more the set up wired sensors in the outside, but wireless to the inside (like the FO), instead of all wired like your, is a simple logistics one I find like you that wired is less prone to error than wireless, but wireless is easier to get inside the home without drilling holes in the walls or windows... So I find the best middle approach is like what the FO stations do... all outside sensors wired to the External TX and then an internal RX to receive that wireless signal.

I also like stations (like the FO) that have the sensors wired to that External TX Unit, but is not all in just one big piece. That way one can place better the sensors, like raising the wind instruments higher in a pole and/or putting the outside temperature sensor in the shade.

Well hope input like this is what you were looking for. I am not saying build your station like this, I am just letting you know what a customer like me would be normally looking, based on my personal preferences. Hope I understood what you want. I also apologize for any part of what I say you do not understand. Although I do have a very good written and oral english, I do must accept english is not my primary language. And I sometimes may forget a word or two in moments like this; and those words may be crucial for the communications to flow.

If I can help in any other way with more information, or clarify anything I told here, I am at your orders...


Francisco

[Charlie]

Thanks, Francisco (and Ron and Mark too) for your feedback, it's great to have someone to exchange ideas with.

Cost is certainly going to be a challenge. Since I'm building a few units in a local manufacturing facility instead of making millions of units in China, I think there's no way to come in under the F.O. in terms of monetary cost. However not having to climb on my roof when it's full of snow, or spend an hour editing out rain spikes, or redoing cold solder joints, is worth a few extra pieces of silver to me at least.

I'm also hoping to leverage parts people may have bought in the past (1-wire sensors, rain gauges, etc.) to make the overall cost lower. I may offer some competitively priced sensor devices if there’s some interest, but for now most of us hobbyists have a box or two of bits in the workshop gathering dust that we can leverage to keep the cost down.

I wrestled with wireless for a long time, and for now at least, decided against it for several reasons. First, it’s a huge source of data errors. There are certainly ways to make the link more robust, and there are error detection and correction algorithms, but it’s another thing to go wrong. Then there are power issues. Wireless implies batteries, which imply serious design constraints around power consumption. Certainly doable, but months more work. Next there are regulatory issues. Once you start transmitting things, you find that every country has different frequencies and allowable emissions limits. Here in Canada, I can do pretty much anything under 100 mW effective radiated power without license. Not so in a lot of the rest of the world, and applying for a single license would destroy any hope of having something at a reasonable cost.

So if I stay away from wireless, this becomes a standard low voltage electrical product with effectively no certifications required, other than declarations around EMC and safety. Simple and more robust, and supplying the data when you want it.

Do you think lack of wireless is a “show stopper”?

[steve]

Lack of wireless would certainly make me think very hard before buying a weather station. I wouldn't enjoy digging a trench out to the sensors for the cable!

Would you not need a CE mark if you were to sell these in Europe?

[fadiaz]

Hi Charlie,

As I said before, lack of wireless will not be a "deal breaker"... but surely will be a "-1" in the evaluation.


Francisco

[Charlie]

Would you not need a CE mark if you were to sell these in Europe?

The CE mark is a good question. As near as I can tell, there is no requirement on this sort of gear. It's certainly below the definition of low voltage devices, has no mains connection, no RF transmitter, and attaches to a computer somewhat like a mouse or keyboard. I suppose you could make a case for an EMC Class B requirement, basically not further degrading the computer's compliance, but that's about all I could find. Possibly ROHS for materials (lead free soldering).
CE is a self declaration for this sort of gear; basically stating what it meets and does not.