Software Engineer Grows Weed pt. 1

[ttystikk]

Also, as far as software changes for those interested:

I've moved away from node.js and went to the JVM. I was using scala as the language anyways (which can compile to JS or JVM code) and in the end node.js just wasnt giving me the timing I needed. The move to java went pretty well, ended up using the pi4j library which has pretty much done the trick for most of my sensor needs.

You get your monitoring and control system dialed in, I'm going to want you to build me one.

 

[Yesdog]

@ttystikk roger that!

 

[blackforest]

Your setup looks really nice. Love the monitoring equipment! My best advice would be for you to start over with non infected plants. I would get rid of everything, sterilize the crap out of everything and start with fresh non-infected clones, or from seed. You are going to be disappointed with a mite infested harvest. Not to mention, right now they the are small in numbers and you can easily get rid of them all. If you let them proliferate, you are going to have a problem 1000x worse. I only grow from seed now because of a similar issue a long time ago. If you are going to go this far with technology, equipment, and expense, you should go all the way with top quality genetics and a clean lab type environment imo. It's a hard pill to swallow sometimes, to start over, but I wish someone would have given me this advice when it happened to me.

 

[ttystikk]

FFS, they're spidermites, not the invasion from Planet Doom.

Azamax is a foliar pesticide. Pouring it on the roots may work, but not very effectively. But by all means, spray the plants with it.

Insecticidal soap, applied twice a week, done.

Just these should be entirely sufficient.

Mighty Wash has pyrethrins, not listed on the label. Keep using it.

Rotate amongst all 3 for best results, but spray every 3 or 4 days.

Add some foliar kelp and get twice the return on your efforts.

You're gonna need to deal with them sooner or later, may as well learn now.

 

[goofy81]

What's the black plotter program? Is it your own? Got a
I'm testing arduino with megunolink . But those graphs just look a lot cooler!

 

[SPLFreak808]

hehe, yeah that UI is 'grafana', the company I work for is one of their sponsors. It's a really solid graphing system, we use it everywhere at our corporate offices, got TVs hanging around everywhere with various teams system stats etc.

So ultimately I have grafana set up to take data from graphite (more like carbon-relay, but graphite stack basically). I have a statsd server set up that takes simple UDP based stats, and that bundles em up and sends em to graphite.

statsd makes logging new data really easy, takes UDP datagrams in a really simple ascii format:
<metric name>:<value>|g

So for the raspberry PI, I'm running node.js and thats interfacing with

* 3 "1-wire" digital therms (DS18B20), you can get 3 for like $10 and they're waterproof.
* MH-Z16 CO2 sensor with UART -> I2C breakout board. This is pretty cheap but kinda janky. The breakout board I got with it lets me reboot the meter so it wont auto-calibrate and stuff
* SI7005 - Really janky little i2c humidistat (get the si7020 over the '05 any day). Works, but it pisses off other i2c devices.
* IoT enclosed relay (these things are fucking sweet). Basically a power strip you can control with any sort of digital circuit. Used to turn CO2 on/off.

So node.js basically just babies the sensors, does some smoothing on the data, and just sends it off to statsd for the metrics, plus uses the Co2 data to turn the relay on/off. Initially I wanted to do controlled venting and stuff to control humidity, but honestly im doing everything i can currently to keep any humidity in there at all.

Here's some of the hardware:

CO2 sensor 'assembly', had to move a few jumpers, which was just a cruel joke in the end. Board works, but its mostly melted.

Jumper i was supposed to 'move' lol cruel joke at best.

Finished soldering on the 'hat'. I started off with a breadboard, but every time i bumped the enclosure something stopped working =\

I am realllllly bad at soldering

That surface mount chip was supposed to drive the therms, but I ended up using GPIO and the linux 'one wire over gpio' driver, which works ok.

I'm not gonna lie, that's some terrible soldering lol, but ive been doing it for a very long time and it does take some practice and a steady hand, in fact the only thing that comes to mind when soldering is trimming bud, the stress feels very simular, burning eyes sore back ect..
Don't worry though, im a generator engineer thats been hanging around the arm vlsa/ic guys trying to gain some skills in your department and i suck 10x more then your solder skills lmao. So you can write firmwares and insert kernel modules? Pretty cool, all i can do is get stuck in boot-loops lol


Shit luck with the mites, clones will fuck you hard if you dont clean them up before taking em home

 

[Yesdog]

@blackforest yea =\ trust me I definitely feel the pending doom. But, a sterile environment for me is almost impossible.... I'm almost more worried about getting rid of the infestation and then just not thinking about it until a month later and they crop back up. I will probably always (at least at this property) always do a preventative spray. My grow tent is smi-outdoors (think outdoor accessible storage area). The real trick seems to be figuring out how to do it cheaply... but azamax for veg, mighty wash for flower. I will probably always use it. This grow is also definitely more of an experiment, even if the mites are under control for this batch, I still may just early flower and take my own cuttings for monster-cropping.

@ttystikk and yea... definitely more of a practice run for me. Granted it's kinda expensive, but I feel like I can get my first failure out of the way pretty quick. I'm also ok with the idea of applying foliars every few days. After seeing how many of them are out in the foothills area, I've lost all faith I'll ever completely get rid of em. I dont have a kelp foliar, but I've heard of people using Floralicous+ as a foliar feed (also has kelp). I think the humic acid causes some ph issues, still gotta read on that more (or just get kelp). As far as the azamax non-foliar, it says on the bottle you can add some straight to the rez. Was reading around and I guess some of the... whatever compound (neem extract?) does make it's way to the plant systemically, not as effect as the foliar, but it does leave a longer-lasting lingering antifeedant in the plant itself.

 

[Yesdog]

@goofy81 so the graphing software is actually called grafana, and the data source i use is carbon/graphite. It's a web-service we use extensively where I work. We use it there to track some several thousand servers, with another few X magnitude of datapoints.

Your software looks pretty cool, is that a custom... Java UI? .NET?

So carbon/graphite is the collection/query software (normally used for plain 'graphite' stats panel), grafana is basically an alternative UI for graphite (and its so so so much better). Then I have a ANOTHER standard service, 'statsd'. What this does is just gives you an easier network API for adding stats to graphite. Graphite/carbon has a kind of complex API, but if you put statsd in the middle, your service can just send really simple 1-line ASCII 'stats' to statsd, and it rolls them up and prepares them for graphite/carbon. I wanted to go full web-based just so I can still view the panel from my phone, and anywhere else via the web. It seems like a convoluted software path, but it works great! It's a lot to set up, but its really not that complicated. I keep all these services running on an external service server, just cause I don't need the Pi heating up my tent more.... but in theory you can run all of it on the Pi.

So basically:

env-contorl -> statsd -> carbon/graphite -> grafana

 

[ttystikk]

@blackforest yea =\ trust me I definitely feel the pending doom. But, a sterile environment for me is almost impossible.... I'm almost more worried about getting rid of the infestation and then just not thinking about it until a month later and they crop back up. I will probably always (at least at this property) always do a preventative spray. My grow tent is smi-outdoors (think outdoor accessible storage area). The real trick seems to be figuring out how to do it cheaply... but azamax for veg, mighty wash for flower. I will probably always use it. This grow is also definitely more of an experiment, even if the mites are under control for this batch, I still may just early flower and take my own cuttings for monster-cropping.

@ttystikk and yea... definitely more of a practice run for me. Granted it's kinda expensive, but I feel like I can get my first failure out of the way pretty quick. I'm also ok with the idea of applying foliars every few days. After seeing how many of them are out in the foothills area, I've lost all faith I'll ever completely get rid of em. I dont have a kelp foliar, but I've heard of people using Floralicous+ as a foliar feed (also has kelp). I think the humic acid causes some ph issues, still gotta read on that more (or just get kelp). As far as the azamax non-foliar, it says on the bottle you can add some straight to the rez. Was reading around and I guess some of the... whatever compound (neem extract?) does make it's way to the plant systemically, not as effect as the foliar, but it does leave a longer-lasting lingering antifeedant in the plant itself.

Azamax is an oil, and adding oils to your res will definitely fuck things up.

 

[Yesdog]

@SPLFreak808 I could write my own modules... if I was a masochist lol. This is one of the reasons I wanted to go with a device that had a newer and fully functionality linux kernel: there's actually a ton of modules out there for popular components.

Almost every core component in my system (besides some of the raw sensors themselves) has a kernel module involved with pretty exceptional support. Here's what I got, and the linux support:

1. 1-wire therms - Full kernel support if you have a kernel module for the 1-wire 'master'.
   1. 1-wire GPIO master - easy as it gets! Designate a gpio pin as your 1-wire master, add a pullup, kernel does the rest.
   2. Various master chips - far better solution really, the 1-wire impl is missing some functionality that helps with stability. I used the DS2482 (that tortured chip i tried to solder on, hilariously does work.) I think the linux kernel has support for about 8 different standard 1-wire master chips.
   3. After the 1-wire module and the master modules are loaded, I get this for the therms ( the last t=#### is the temp in milli-C):
      pi@pi1:~/pi4j $ ls -la /sys/bus/w1/devices/
      
      total 0
      
      drwxr-xr-x 2 root root 0 Aug 5 09:23 .
      
      drwxr-xr-x 4 root root 0 Aug 5 09:23 ..
      
      lrwxrwxrwx 1 root root 0 Aug 5 09:23 28-8000002803df -> ../../../devices/w1_bus_master1/28-8000002803df
      
      lrwxrwxrwx 1 root root 0 Aug 5 09:23 28-800000280fcf -> ../../../devices/w1_bus_master1/28-800000280fcf
      
      lrwxrwxrwx 1 root root 0 Aug 5 09:23 28-800000280ff7 -> ../../../devices/w1_bus_master1/28-800000280ff7
      
      lrwxrwxrwx 1 root root 0 Aug 5 09:23 w1_bus_master1 -> ../../../devices/w1_bus_master1
      
      lrwxrwxrwx 1 root root 0 Aug 5 09:23 w1_bus_master2 -> ../../../devices/w1_bus_master2
      
      pi@pi1:~/pi4j $ cat /sys/bus/w1/devices/28-800000280ff7/w1_slave
      
      15 02 ff ff 7f ff ff ff b3 : crc=b3 YES
      
      15 02 ff ff 7f ff ff ff b3 t=33312
2. CO2 sensor - NDIR-Z16 (UART) with a UART -> I2C board (SC16IS750). The SC16ISxxx also has a kernel module that just automagically connects to it over I2C and provides linux with a new serial tty. However... I needed support not offered by the module (GPIO to control calibration), so just opted into the user-space implementation again.
3. Humidity sensor - plain I2C, rather simple implementation in user-space.
4. As far as everything else, I use I2C-dev. The I2C kernel module is required for any other kernel modules using I2C, but it also opens up a user-space interface that you can use to write 'user-space' drivers (this is what most raspberry/arduino samples do). In fact, most devices use a small subset of I2C/smbus that can be implemented using just read/write on the file handle for the i2c bus.

Making your own user space drivers is actually pretty efficient depending on the bus. Implementing 1-wire over GPIO would be almost impossible in user-space because of the timing requirements (tight bit banging and stop conditions, etc). I2C on the other hand seems perfect for it. My humidity sensor takes 35ms to respond sometimes when its warming up.... thats a long time! So at the end of the day my cpu stays at about.... 3% SYS and USR, and thats polling 5 devices 3 times a second plus emitting all sorts of data across my network (and running on full Java .

 

[SoOLED]

automated DWC, DWC is one of those things you kind of have to live with your plant on a daily basis. smell, taste, observe, test. IMHO: DWC, for first time growers never ends well.

I can see this going much better with some drippers, hoops or EB/FLW.

with plants that small you can dip them, get some press and seal, or plastic wrap: to keep the clay in the netpot. turn it upside down and completely dip the plant in a bucket of mite death. that way you know every inch is covered. then spray your tent with water/isopropyl alcohol, don't hit your plants with it. just the tray and walls, wipe down everything with it too.

if you can lower you temps too, mites love summer fun, mites hate cold.

 

[Yesdog]

@SoOLED I actually do have some drippers (really hard to see in the last pic, first post has better pics showing it). Up until recently the plants have actually been a whole foot away from the water. They touched down last week and have been just taking off since (also raised the water level). Wanted to promote 'more' roots, but ended up with really long ones.

Which brings me to the dipping lol. I have.... no idea how to go about dipping the plants with their current root structures. I've got at least 12in of roots coming off of each of these. I imagine its not a great idea to touch the roots, but I don't know how else to shield them while i dunk the girls.

@ttystikk hmm it says on the bottle it can go in the rez, maybe the surfactant or something gives it the solubility? Or its just GH trying to get me to use more of their products =\

 

[Olive Drab Green]

I set out to make an ultra high-density grow in a 2' x 4' space. If things went as planned, I could fill that space with the maximum amount of buds and save several thousand dollars off of my usual quarterly weed expense.

To do this I was going to have to science my way to success.

What I ended up with was an over-engineered spider mite cave, but thats neither here nor there.

Goal was to have a sealed tent, with a cool-tube light, CO2 injection, and no spider mites.

The heart of the beast is the reservoir, which is mostly a DWC system, but with an additional dorsal irrigation hookup that would aid in plant growth via drip until the roots reached the rez.

The brain of the beast is a Raspberry PI that is hooked up to: 3 temp sensors (out-tent, in-tent, and rez), CO2, and humidity (both bottom tent). These log to a "grafana via graphite" metrics logger, plus additional software to control CO2 injection.

Monitoring:

So far i've seen explosive plant growth! But I started with clones that have spider mites. So I'm fucked.

This post is open for discussion! This attempt of the grow is an almost definite failure due to spider mites. Pt 2 coming soon.

About the mites:

You're not even in flower yet, so you aren't fucked. Get some neem oil, a full sized water bottle, organic castile soap, and a teaspoon and mix in 1 tsp neem and 1/2 tsp soap into a full water bottle of water. Don't use this in late flower, though.

 

[disbeverk]

I've had good luck with "Greener Cleaner" for spider mites on babies... it's derived from soybean oil, rather than neem. Neem is tried and true though, give the Azamax a shot first. Remember to spray during lights off. If the plants aren't looking strong and healthy, might want spray 'em again with plain water to rinse the treatment off. Anything you spray to kill the bugs is going to stress the plant somewhat.

I've also rigged together a raspberry pi, a relay board, and some sensors to manage my aero grow. I'm currently working on the interface side of things...

Exciting to see what everyone is building!

 

[Yesdog]

@disbeverk I sense your UI-foo is strong (jealous!). I think just from the posters on this thread, we could have a pretty strong team for some kind of bigger project

ahhh also super jealous of your dynamic config. What kind of relays are you using? I found this cheap fully assembled relays that work awesome. Wish I had a big relay box tho, just afraid of doing AC wiring =\

EDIT: also, what kind of therms are you using? The maxim ones give you milli-C, yours seem to offer only 1 decimal point? Even with the milli-C ones I have, the jitter in the graphs annoyed the piss out of me. I ended up applying a rolling average to the reading, so when it jitters between 2 readings like that, it smooths it out to actually show the middle value (is about a 1s average, so therms are a bit less response than normal, but I think my sensors only have a 300ms response anyways).

 

[disbeverk]

@disbeverk I sense your UI-foo is strong (jealous!). I think just from the posters on this thread, we could have a pretty strong team for some kind of bigger project

ahhh also super jealous of your dynamic config. What kind of relays are you using? I found this cheap fully assembled relays that work awesome. Wish I had a big relay box tho, just afraid of doing AC wiring =\

EDIT: also, what kind of therms are you using? The maxim ones give you milli-C, yours seem to offer only 1 decimal point? Even with the milli-C ones I have, the jitter in the graphs annoyed the piss out of me. I ended up applying a rolling average to the reading, so when it jitters between 2 readings like that, it smooths it out to actually show the middle value (is about a 1s average, so therms are a bit less response than normal, but I think my sensors only have a 300ms response anyways).

I'm not running HID lights through it, but I'm still running the first JBtek 8 channel relay board I first started testing on over a year ago. It's been rock solid for all my fans, pumps, and 17w T5. Not in a hurry to switch my lights off mechanical timers though.

That sensor in the screenshot is a DHT11... I believe it only reads in whole C, but I honestly don't recall where the dec is being broken down... sensor, backend, db, or web app. The sensor is only being pulled every 30 seconds.

 

[ttystikk]

I've had good luck with "Greener Cleaner" for spider mites on babies... it's derived from soybean oil, rather than neem. Neem is tried and true though, give the Azamax a shot first. Remember to spray during lights off. If the plants aren't looking strong and healthy, might want spray 'em again with plain water to rinse the treatment off. Anything you spray to kill the bugs is going to stress the plant somewhat.

I've also rigged together a raspberry pi, a relay board, and some sensors to manage my aero grow. I'm currently working on the interface side of things...

Exciting to see what everyone is building!

There's a wet spot in my pants...

 

[Yesdog]

Update on the plants (still no sign of mites! think the Hot Shots strip is doing the job great):

Sorry for the shit picture, was taken through the viewport at lights-on. So... these 4 clones i started with were all supposed to be the same damn pheno. What is going on with this right plant? Looks totally different =\

 

[Chef420]

Janey Mack! Is that aluminum foil? Get that out of there. Lol
Seriously though, it creates hot spots on the plants. Doesn't diffuse like Mylar type products.

 

[Yesdog]

@Chef420 I swear I haven't heard of that until you and somebody else mentioned it to me in the past hour lol