A white board excercise on designing a Raspberry Pi temperaturing measuring device

Question

Looking for input on the design of a temperature measuring device using a Raspberry Pi.

I would like to use this forum to do a white board design exercise on a project I am developing. One of the interactions I miss after retiring from a 37 years career in technology, is the ability to sit with a group of techies and socialize an idea to get their perspectives. I would like to try using the Element14 community for such an exercise.

In a closed group, I would know the backgrounds of the individuals participating and from experience be able to evaluate their have opinions. Not all opinions have value. Using this forum doesn’t afford many of the face to face controls. I would like to work around that. I suspect there will be more overhead on my part to evaluate the input. I look forward to interacting with those who wish to participate.

Please provide your input around designing a temperature recording device using a raspberry Pi. The term input is defined as suggestions, recommendations, best practices, directions or questions. The pi_temp u

Looking for input on the design of a temperature measuring device using a Raspberry Pi.

I would like to use this blog to do a white board design exercise on a project I am thinking of starting. One of the interactions I miss after retiring from a 37 years career in technology, is the ability to sit with a group of tech guys to socialize an idea and get perspectives you may not have thought about. I would like to try using the Element14 blog for such an exercise.

In a closed group, I would know the backgrounds of the individuals participating and from experience be able to evaluate their have opinions. Using this forum doesn’t afford those controls. I can accept that. I suspect there will be little more overhead on my part to evaluate the input. I look forward to inacting with those who wish to participate.

Please provide input around designing a temperature recording device using a raspberry Pi. The term input, can be defined as suggestions, recommendations, best practices, directions or questions. The pi_temp unit will be used to measure and record water and air temperatures in and around an Atlantic Salmon fish ladder. The ability to operate without AC power would be an advantage but not a requirement. The unit will be isolated and have no access to a communication network. Data collected should be stored locally and retrieved manually on, no predetermined schedule. The unit may be deployed in rugged conditions exposed to the elements. People deploying and operating the unit, as well as collecting the data may have little technical skills.

Context:

I made an offer to build a device using a Pi, for a non-profit organization that works for the preservation of Atlantic salmon, to record water and air temperature and store the data for collection. This is a volunteer project with no funding provided. All cost are out of pocket. There are no plans to make this into a commercial product. I will write for publication in a technical resource when the project is complete. I saw a manual process and thought, hey I can fix that.

The project provides only the personal gain of applying years of technology knowledge to assist in solving a real world problem. My retirement now affords me time to take advantage of some of the offerings in the Element14 community.

Before starting the final design, I will make a post on this forum containing a list of the requirements I have gathered and hope to achieve in the project. This is mean’t to be a white board dialog exercise (experiment) to solicit input from tech folks. I welcome your participation.

genebren · Answer

Hi Sean, I would start out by pointing out that a Raspberry Pi is an awfully big hammer for such a simple task as capturing and storing a couple of temperatures. This is related to the cost of a RPI and the energy to run one. There are several other available devices that might be better suited to the job, based on cost and potential battery life. Also, it was not clear to me if this was a device that would be left in place (for continuous monitoring) or moved from site to site to collect measurements in an assisted sampling process? Otherwise, this sounds like an interesting project that could benefit those working to assist in the preservation of Atlantic salmon (or potentially other species). I worked with a company that developed survey instrumentation for people involved in monitoring the health of fish populations (length, weight, species, tag info, etc.). There seems to be a few other companies/organizations that are also looking into these same sort of things, so looking first to see what is already out there would be a good idea. Good luck, Gene

mcb1 · Answer

Sean My approach would be different to that suggested so far. I'd use an Arduino based controller with SD card, realtime clock, and a couple of waterproof DS18B20 devices. This will easily log the data for many days, and consumes much less power. When your man attends, he simply presses a button (to halt any recording) removes the SD card and replaces it with another. The button press times out and the relevant information gets written to the new card. At that point you could turn on an LED indicator for xx to show it is working. The options for powering are solar, but you might be able to utilise the small wind powered option I eluded to here. (near the bottom) BT_Sentry : Bluetooth Options Other nice options could be using eINK displays to show status, or other information. Self-adjusting clock with e-display There are other LoRA options to feed data out, or transfer it, but keeping it simple is the better solution IMO. Cheers Mark

DAB · Answer

Hi Sean, I think you will find a good group here at element14 to discuss technical issues with. After I retired about 15 years ago I found a great group here and have enjoyed many technical discussions. The Rpi will work with most of the digital sensors, but it lacks any A/D capability for analog sensors. If you need to do remote sensors I would look at the more energy efficient processors and dump the data to SD cards. DAB

colporteur · Answer

Not sure if posting to my own comment is acceptable or if editing the original post is preferred. Good and bad to both. I have started the build (bill) of materials. BoM -Pi 3 -3A power supply -16G SD card -weatherproof enclosure -DS3231 Real time clock Module -2X DS18B20 Temperature sensor Design criteria under consideration Provide air and water temperature Management of data collection schedule Management of data storage How to communicate with a Pi to extract data Enter contest for PiJuice battery unit Sean DS3231 Clock Module DS18B20 Temperature sensor

Problemchild · Answer

Hi Sean I recently was doing a test on the RPI on a Bitscope single an quad carrier and I needed a 3-4 foot square Solar panel to even start to drive a Pi. It wasn't a permanent set up and had no battery but it shows that you need quite a lick of power to run these devices even whilst starting up as well as a requirement to charge a battery for dark operation. Agreed a zero or a PI with the power management configured well would be much better but it shows how you must size your panel quite generously for reliable operation !

dougw · Answer

If you need to minimize power for a PI system, you could rig a timer on your power supply so that it only supplies power to the PI briefly every hour to record temperatures. The PI could actually turn the power supply off when it is finished collecting its readings each hour. That way only the timer is drawing power 99.9% of the time.

colporteur · Answer

Thanks Gene your point is valid. I have limited my device choice to the Pi because I am just playing in what I like. I also consider an Ardunio as an option but would like to explore Pi first. Currently there is one site (fish ladder) that is visited twice a day, to record the water temperature and release any fish contained in the counting gates. A rather long extension cord provides power to a camera used by a TV station at the site. Not sure if that will be available going forward. MY battery thoughts were, what if the unit was deployed at a site that had no power. I recall reading about a weather project using a Pi that was solar powered. I would consider a power support option in design. I'm thinking Pi zero is the best choice because of lower power requirements. Really appreciate your input.

shabaz · Answer

Hi Sean, The Pi may be better suited to aggregate the information from multiple devices, and process it, and upload it somewhere. The Pi would excel at that. I'm not sure of the entire requirements, but from the ones listed, a microcontroller approach would be better because of the points already mentioned by others. The Pi has been used in unusual locations (e.g. ballooning, etc) but it could be better to design or use something already rugged, lower power and ready for running at (say) cold temperatures. But I know nothing about the Atlantic, or Atlantic salmon apart from eating it : ( so I'm just assuming things like temperature could be an issue.

rew · Answer

I don't optimize led-on-time anymore. I use (green) leds that are "uncomfortably bright" when on 130 microamps, so I run them around 50microamps.....

Problemchild · Answer

Look in to setting up a LoRa based network or just point to point at low bit rates it's capable of several KM paticularly if you can get it to be line of sight with each other!

shabaz · Answer

Hi Sean, Bluetooth can be used to transmit data, however this implies that the laptop also needs some software written to fetch individual records or specified content. It could be far easier for your particular scenario to use the Pi as a web server accessible by WiFi, and let the laptop user browse to it using IE, Chrome, etc. That way no additional software is needed to be written to access/download the data or view graphs, etc., and any laptop can then be used, or mobile device, without having to write an app. Google search reveals the Pi can be configured in AP mode, I've not tried this but it's a reputable site: https://learn.sparkfun.com/tutorials/setting-up-a-raspberry-pi-3-as-an-access-point

a531016 · Answer

Sean, That is certainly the approach I would (and have) taken in the past. I like to affordability and compatibility of the BS18B20 sensors, and you can buy waterproof versions for the water sensing. You don't need any ADC circuitry, and a one-off "offset" calibration should do the job. I have had up to 4 connected in series to the single gpio on a PI. I have done a couple of things with the readings: save to text file, and SSH to retrieve Store in a mySQL database and serve by web page (Apache and PHP) as table, download file or graph or send as recorded to an API to save remotely (maybe best for a battery option, so the results are saved as you go). I have used Cron to request the recording (15 min intervals), although depending on your required log frequency and time period, a python script may be better. Of course a hybrid of "local cache" and send to server is probably wise - belt and braces! David

colporteur · Answer

Thanks for the feedback on temperature sensor. My plan was to use two of them. I have ordered waterproof versions from an online website. Relatively inexpensive. I have no experience with the devices. What I found online regarding them appears to be pretty straight forward. I have some experience with DHT11 & DHT22 temp sensors. Ideally I would like to use SSH, but there is no network. Now I was thinking of using a serial connection but that requires user intervention to make connection. I was hoping there was some suggestion of bluetooth capabilities. I also thought of making the Pi an access point and seeing if a connection could be established that way. Any suggestions along these lines or others would be appreciated. If I consider battery supply, then reducing it power footprint (i.e. transmitting signals) would be important. Appreciate your insight. Sean

a531016 · Answer

Sean, A little more detail may be helpful if you can provide it? For what period will the device be deployed between human interaction? What logging frequency do you need to achieve? Are there any GSM networks (texting the data could be an option)? From what you have said, and thinking about power conservation, I think appending your logs to a text file as they are recorded would be more efficient than running a database engine. Also having a script that fired when a usb flash storage device is attached is more frugal from a power consumption point of view than anything wireless (bluetooth, wifi or GSM). This would allow a human with no skill to retrieve the data, but you would want some LEDs to indicate the copy and completion status? Does that sound like a viable option? Also I would avoid installing a full OS with a GUI (command line only) and maybe consider an older PI or a PI 0 for the project, as they have lower power consumption, and processing power or RAM shouldn't be an issue for this project (my before mentioned project is running on a RPI B). David

danielw · Answer

I'm leaning toward the approach of mcb1 . I've relegated my PI B to monitoring temps in my house / garage using DS18B20 sensors. If I wasn't wanting to access the data wirelessly, and I didn't have the PI sat in a box, I would have gone with an arduino for the lower power consumption. Over the past year I've had a few lockups that are fixed with a power cycle. Weirdly it hasn't done it during summer (Yet) It's in the garage and it's about 30 - 35C in there most days at the moment so it can't be an overheating issue! I've used AVR / Arduinos before but instead of SD card I used a separate eeprom. Data storage was infrequent and sequential so I wasn't concerned about exceeding the maximum number of writes in the eeprom. It Could run from 12V and NiMh battery. When plugged in the battery charged up. The clock and calender was implemented in software, and ran from battery for a few days between charges using the low power modes of the AVR to increase battery life. Also if I can I fit an LED indicator. We sometimes omit them to save pennies, But it costs when you are trying to support a 'box' over the phone and don't even know if power is getting to it. If there's a processor I try to use a regular heartbeat, 2 or 3S that slows to 6S when on battery. I also use as short an LED on time as possible to minimise power consumption.

shabaz · Answer

As mentioned, a web browser would eliminate complexity. On the one hand you mention the users have limited knowledge, but on the other hand you mention the execution of scripts to pull data, so it's confusing.

Not only that, but as mentioned, pulling data by executing scripts would mean you've now got to code an application on the PC too, and maintain that. Using a web browser means Microsoft/Apple/Google will maintain the browser for you.

shabaz · Answer

Hi Mark, Some of the water-clear red LEDs (typical 5mm ones for example) are nicely visible end-on, at 100uA (and can even cast red light dimly at that level, onto a white surface), but still glow red at lower currents too. For the Smart Doorbell project, I used an orange LED that was surface-mount (but fairly large, so easy to hand-solder or attach wires) which at 2mA, was bright enough to see even in reasonably strong outdoor sunlight (the unusual summer here allowed that to be noticed!) at 2mA. It was this one: VLMO30L1M2-GS08 VLMO30L1M2-GS08