WeatherDuino Forum

I'm working on assembling my WeatherDuino system, and I see that both the RX and TX boards call for the use of a 7809 linear voltage regulator. These are simple, easy-to-use devices, but they aren't very efficient--the voltage drop is simply dissipated as heat. At 12 volts input, this means that 25% of the input power is being wasted. If the actual voltage is higher (12 volts nominal is usually 13.8-14.1 volts), 35% or more of the input power is lost.

For the RX board, this isn't really a problem; it's designed to be plugged in, and the power use isn't that great anyway. For the TX board, though, it could be. It's a wireless transmitter, and I'd expect most installations will be using some sort of battery to power it. The wasted power from the linear regulator means the battery/solar panel/charging system/etc. will need to be about half again as large as would otherwise be required.

An alternative would be to use a DC/DC buck converter. The simplest way to implement this would be to use a module like the CUI V7809-500 (http://www.digikey.com/product-detail/en...ND/1828604), which is a drop-in replacement for the 7809, perhaps with some adjustments to the capacitor values. This would give ~95% efficiency across the input voltage range, and wouldn't require a heat sink. The obvious drawback is cost. A 7809 is dirt cheap (I bought 5 of them on eBay for $0.99--if they arrive, that is), while one of these modules runs around $7. I'm sure there are other suitable modules out there, but I have no idea if there are less-expensive options. I'm sure nothing comes close to $0.20 each, though.

Another option would be to design a discrete buck converter on the TX board. That would likely be less expensive, but it would complicate the design considerably. There might not be room on the PCB to include this circuitry, and/or it may require surface-mount components, making it harder to assemble.

I haven't seen much indication of the power draw of the TX board, so I'm really not sure if this is a significant issue at all--thought it might be worth considering though.

Hi,

The buck converter won't be 100% efficient, so I doubt if the power saving when dropping to 9 volts would be worthwhile. It might be more significant if dropping directly down to the 5 volts required by the basic ATMEL microcontroller.

(03-08-2015, 15:38)danb35 Wrote: [ -> ]I'd expect most installations will be using some sort of battery to power it.

Using an off-the-shelf Arduino board does have the disadvantage that the power drain is much larger than for a "custom" transmitter/controller as used in most commercial stations. For example the power consumed will be literally 1000s of times higher than for a "Fine Offset" manufactured transmitter (which uses two AA cells). So I think most "Duinos" will use either a "line" (mains) supply, or quite a substantial Solar Panel and (rechargeable) battery combination.

Of course the two "prime movers" (Werk_ and _Bob) are located where they probably have more solar power than they know what to do with. But for those of us in higher latitudes, solar powering can be quite problematic. There is a thread on this forum where somebody was attempting to design a "reliable" solar supply, and (IMHO) the figures were quite frightening, but perhaps over-pessimistic.

For "Solar" installations there are numerous opportunities for adopting more sophisticated switching and control functions to improve efficiency, but the design is far from trivial. Some of those that have tried, untimately just suggest to increase the panel and battery size by 30% and use "simple" electronics!

Cheers, Alan.

No, the buck converter won't be 100% efficient, but the datasheet for the one I posted calls for 95% efficiency for an input voltage of 12-14vdc. That's a pretty big win compared to 65-70%.

What I have in mind to power my station is a 10W panel/charge controller, and a 12V/7AH sealed lead-acid battery. My gut feeling from looking at the device is that this should be overkill to a large degree, but I don't recall seeing anyone post actual power consumption, and I'm not seeing smaller panels and charge controllers a lot cheaper. I already have a few batteries around--rather past their prime at this point, but I think they should still work.

I designed an Arduino-powered device a few years ago (specifically, a Teensy 2.0 (http://www.pjrc.com/teensy/)), and it only drew around 100 mA while operating. Of course, all it was doing was receiving serial data, storing it to an SD card, and printing data to an LCD, so I wouldn't be at all surprised if the TX board draws more than that, but I wouldn't think it should be much more, at least when the fan isn't running.

Hi,

It really depends where you're located.

Here in the UK we can get a really "grey" week or more in mid winter where little if any battery charge would be obtained. A 7 Ahr battery can only suppoprt a 100 mA continuous drain for less than three days. Personally, I would not consider a "10 Watt" panel, nor a 7 Ahr battery, to be "compatible" with a 1 watt (100 mA, 10v) continuous load. At the other extreme, in a more "favourable" location and time, the battery could become fully charged in a day, with many hundreds of percent "excess" power.

But (in that other thread) I believe that Werk expected the drain to be significantly less than 100 mA; I can't comment myself as I haven't made any practical measurements.

Cheers, Alan.

ADDENDUM: The thread appears to be here

(03-08-2015, 15:38)danb35 Wrote: [ -> ]An alternative would be to use a DC/DC buck converter. The simplest way to implement this would be to use a module like the CUI V7809-500 (http://www.digikey.com/product-detail/en...ND/1828604), which is a drop-in replacement for the 7809, perhaps with some adjustments to the capacitor values. ...

Another alternative is using a 1€ DC/DC converter like this:

[Image: m6K-zF7pP4CF4DmAKCFp5Ig.jpg]

Wiring it to replace the 7809 is very easy.

(03-08-2015, 18:54)danb35 Wrote: [ -> ]I designed an Arduino-powered device a few years ago (specifically, a Teensy 2.0 (http://www.pjrc.com/teensy/)), and it only drew around 100 mA while operating. Of course, all it was doing was receiving serial data, storing it to an SD card, and printing data to an LCD, so I wouldn't be at all surprised if the TX board draws more than that, but I wouldn't think it should be much more, at least when the fan isn't running.

The total average consumption of the TX v2.50 board is around 60 to 70 mA, it obviously increase if you use a fan for the radiation shield

(03-08-2015, 18:54)danb35 Wrote: [ -> ]What I have in mind to power my station is a 10W panel/charge controller, and a 12V/7AH sealed lead-acid battery.

Depending on your latitude it may be more than enough, or not
My system is running for 3 years with a 10W solar panel and a 14Ah sealed lead-acid battery, and some times it is powering two TX units at the same time. Never was any problem, even in the winter.

(03-08-2015, 18:05)AllyCat Wrote: [ -> ]Of course the two "prime movers" (Werk_ and _Bob) are located where they probably have more solar power than they know what to do with. But for those of us in higher latitudes, solar powering can be quite problematic. There is a thread on this forum where somebody was attempting to design a "reliable" solar supply, and (IMHO) the figures were quite frightening, but perhaps over-pessimistic.

Yes, here in Portugal, we have the luck of having lots of sunshine hours, even in the winter, and we really know what to do with it: By luck, we also have magnific beaches, nice girls and bars

I think that most AU regions have plenty of sun too.

I'm aware of the "study" from hvalentim, which have concluded the need of a 80W solar panel for a specific location in the UK. Quite exaggerated, I'm sure.

(03-08-2015, 22:30)Werk_AG Wrote: [ -> ][ By luck, we also have magnific beaches, nice girls and bars

I think that most AU regions have plenty of sun too.

We have nice girls and bars here too

My latitude is 31 deg N, and I get a fair amount of sun. I guess I'll have to see what the winter's like.

(04-08-2015, 02:01)danb35 Wrote: [ -> ]My latitude is 31 deg N, and I get a fair amount of sun. I guess I'll have to see what the winter's like.

Your location does seem to be sunny. Using hourly averaged data over 10 years from the nearby Savannah Airport and assuming an unobstructed horizon, for a 24/24h 1W draw, PvSyst recommends a 9Ah battery coupled with a 7W solar panel (targeting 4 days of autonomy with a 5% chance of loss of load).

But don’t take my word for it, grab the tools below and do your own reasoning, trying different scenarios.

PvSyst
http://www.pvsyst.com/en/software/download
Fully function evaluation mode, limited to 1 month.
Closest location included in default weather data is Atlanta. You may wish to grab Meteonorm to get data for a location closer to you.

Meteonorm
http://meteonorm.com/en/downloads
Works in a demo mode for an unlimited period. Saving is restricted to five sites.

Instructions on how to import data from Meteonorm to PvSyst:
http://files.pvsyst.com/help/meteo_import_meteonorm.htm

Also, bear in mind it may be wise to plan ahead for future stuff you may one day add. For instance, the values above do not consider having a fan running (easily another 100mA).

The less you discharge the battery the more it will last. Longevity wise, you ought to design things as to ideally not taking more than 25% out of it and never more than 50%.

As a matter of contrast, for Glasgow, Scotland, under similar circumstances (1W 24/24h draw, clear horizon and 9Ah battery), the simulation recommends a 37W solar panel.

Concerning the DC/DC converter I am using a TracoPower TSR 1-2490 myself, and thus far found no problem. It fits nicely, protruding just a couple of mm on the external side of the RX board. Of course, in the end, it does have the exact same utility of a LM7809, except it costs more, it has a potential risk of generating radio noise (cannot comment on that since I do not have a WD unit nor do I use the retransmit feature of the base RX unit) while being a bit more efficient and taking in an higher voltage if needed (up to 36VDC against the 35VDC supported by the Fairchild LM7809ACT).

Edit: Corrected the maximum input voltage of the LM7809ACT.

If you are using the 78xx series regulators for your application which you mentioned here, then ofcourse you are wrong. They are not such reliable for high current applications due to their heat problem.
If you are looking for the the any buck or boost converter. Then go for them. They have better efficiency and reliabilty.

pcb cost

Looking more carefully at the specifications sheet of the LM7809ACT the maximum INPUT voltage actually seems to be 35V for models outputting 5 fo 18V.

https://www.fairchildsemi.com/datasheets/LM/LM7805.pdf

I do agree they would heat a lot a more with such values, which should not be the case with a decent 12V solar regulator.