7 Day Office Thermostat

This is for someone who works from home usually four days a week. Every now and then the days get shifted around, so it needs to be easy to say “I’d normally be here today and away tomorrow, but I’m leaving now and I won’t be here tomorrow”.

Also, this unit is for a small outbuilding that can get very cold overnight, so it has to two temperature settings. One for “I’m going to be here today so start warming up to XX degrees an hour before my usual start” and one for “Even if I’m not here, turn the heater on if it gets below YY degrees.

Design decisions:

• Get the basic functionality working with an Arduino based approach, then later add an ESP32 based WiFi capability.
• Separate out all the 240V circuitry from the Arduino and human interface bits, with two separate boxes.
• Use 3D printing to keep parts aligned and vaguely tidy.

Power switching box:

My first version of this was a decidedly “chunky” unit built of CNC cut MDF and acrylic. The circuitry worked but it lacked aesthetic appeal.

There’s nothing much inside this box, just small power supply and a 20A SSR (Solid State Relay).

I decided to redo the unit, and this time I drew up a detailed Fusion 360 CAD, even to including some of the wires.

With this CAD, I could print out small plastic parts to help align things on a lasercut mdf base. The laser also marked placement outlines to help assemble it. Printed PLA parts glue really well to MDF with wood glue. When I revised the base I just levered them off (taking some mdf with them), then soaked the mdf off to reuse them.

I found a cheap extension cord and chopped it in half to supply the power in and power out cords. A 3m section of CAT5 ethernet cable connects the two boxes, with GX16-6 connectors (I didn’t want to use normal ethernet connectors in case someone plugged one end into a computer).

The controller box:

I decided to use one of these cheap “LED&KEY” boards that comes with 8 x 7-segment displays, 8 LEDs, and 8 push button switches. All they require is +5V, GND, and 3 pins on the Arduino. (I saw these used on a Clough42 video and immediately bought a bunch).

The remainder of the parts were an Arduino Nano (clone), DHT11 temperature/humidity sensor, a DS3231 real time clock with battery, and a couple of TTP223 based capacitive touch sensors.

Note: I later tested the Omron SSR and realised it only drew 11mA, so got rid of the transistor buffer.

This time I drew up a 3D printed case, which incorporated mountings for the various parts, and had holes for the display LEDs to poke out.

I also 3D printed some little “pushers” to poke through the holes and press on the push buttons.

This all looked very good.

One problem is that the thin (2mm) PLA allows the other LEDs (on the Nano, RTC, and LED&KEY board) to show through. It’s not really much of a problem and I could always put some tape over them.

I needed to add some printed labels to the buttons and I decided to go with a new technique. I exported the layout as a DXF from Fusion 360 into Inkscape and drew up a pattern, including the lines to cut out. I passed this to my vinyl cutter (“Silver Bullet” brand) and used its SureCutsALot software to print the pattern with registration marks, then align the cutter to those registration marks and precisely cut out the holes. This worked extremely well and is a technique I’ll certainly use again.

On the final version, I printed/cut the cover in 210gsm card, then put it on some baking paper and sprayed heavily with clear gloss polyurethane. This made it look and feel much more like plastic.

The final result looks quite nice.

Except when you look at wiring on the back.

Here I’ve used the reverse of my usual technique and glued small MDF islands down to the PLA. These are 6mm thick with holes tapped for M3 screws which hold small terminal strips for soldering. It works, and makes the soldering easier, but it’s not pretty.

The software:

The program running on here is pretty simple (about 700 lines) with by far the bulk of it to do with using the buttons to edit the settings. The actual code to measure the temperature and control the heater is only a few lines.

  // CONTROL //
  unsigned long newControlTime = millis();
  if (newControlTime - oldControlTime > controlInterval) {
    oldControlTime = newControlTime;
    if (isValidTime && isValidTemp) {
      if (HereButtonPressed || (!beforeStart && !afterFinish && (settings[setTHISWEEK + DayOfWeekOffset].value == 1))){
        //HERE mode
        float target = settings[setTEMPHERE].value;
        heaterOn = (lastTemperature < target);
      } else {
        //AWAY mode
        float target = settings[setTEMPAWAY].value;
        heaterOn = (lastTemperature < target);
      }
    }else {
      heaterOn = false;
      Serial.println("Control: not valid.");
    }
    //write to digital output for SSR
    digitalWrite(SSRDATAPIN, heaterOn);
  }

Conclusion:

It works!

I can think of lots of improvements, but it works and I learned a lot of new techniques.

Alas, the recipient really wants WiFi connectivity so she can change the settings (particularly whether she is here/away today) without going out (in the cold) to the shed. So the next part of this project will be hooking up an ESP32 which will send and receive information from the Arduino (via serial), and present the settings as a web server on the home network.

A Simple Stepper-Based Turntable

Last week I happened across a youtube video in which someone had built a self-aiming airsoft turret. The interesting thing was that he’d mounted the whole unit using just the bearings of the stepper motor. This intrigued me enough that I decided to spend a day (hah!) and knock together a small stepper motor based turntable using cheap and salvaged parts.

I found a suitable stepper and got it working with a general purpose stepper driver circuit I’d put together some time ago. I decided on a 150mm disk and a 150x150x50 lasercut box. I managed the tricky task of reverse engineering the pinion on the stepper to print a matching socket, then set about gathering the parts to make it (Arduino nano and a cheap L298 motor driver board).

The problems of fitting in:

Alas, when I started considering how to fit the parts into the box I’d designed, I realised it would be much harder than I expected. The stepper was a large one, and the L298 board is too large to mount vertically on the side (44mm square pcb will only just fit into 50mm high – 3mm_MDF_top – 2mm_MDF_base, but that leaves no room for connections).

I had three choices (a) go to a bigger box, (b) swap out the stepper for a smaller one or replace the L298 with an A4988 controller, or (c) use Fusion 360 to engineer the exact placement of parts so it could all fit together. I went with option (c) and I’ve spent much of the last five days refining my design.

While this might sound like a lot of work for little gain, I’ve learned a huge amount about working with much more complex assemblies in Fusion, developed a bunch of tricks to improve my projects, and also created my own cad models of devices such as the Arduino Nano, SPDT toggle switches, and pots.

The Fusion 360 model:

It doesn’t look too bad laid out like this, but some of the clearances are tight!

The first prototype:

I got to the stage of laser cutting the box and mounting the first few components before I realised how tight it would be. Here you can see the box (open at the bottom) with the stepper motor, the base (with L298 driver, arduino nano, and a strip of solder terminals), and the disk for the top with the first iteration of printed joiner (the blue six legged thing) to push onto the pinion gear of the stepper motor. On the left is a general purpose stepper driver which I use for testing salvaged steppers.

To make it more challenging, I’d decided that the base piece would sit a bit above the bottom of the box, so that the mounting screws and nuts wouldn’t stick out the bottom of the box, requiring rubber feet. That stole another 6mm of vertical height and made things even tighter.

Assembly begins:

In order to see where I could run wires, I needed to mark the outline of the stepper onto the base. If I was going to do that, I might as well have an “Engrave” sketch with part placement and extra information. Not really needed here but good practice.

The little solder tag section at the bottom is a laser cut piece of 6mm MDF, with a row of holes tapped for M3. The two at the end fasten it down, the other four have a M3x5 screw holding a solderable tag. I find this works well for me, as MDF stands up well to the heat of soldering.

I made up a 6-way DuPont connector cable (my first!) to link the Arduino to the L298 board. To keep the wires from floating around, I printed a small set of troughs to press them into and glued it down.

connecting the Arduino to the L298

It fits:

box (upside down) and base which fits inside it

It’s still a tight fit. I had to cut a hole in the base to let the boss at the bottom of the stepper protrude and tidying the wires away while fitting it together was a bit messy. But it does fit.

I probably could have fitted it together without drawing up a CAD model, but it wouldn’t have been easy.

Connecting to the stepper motor:

This was a salvaged motor and it had a pinion gear attached. I could have removed it, but then I’d just have had to make a solid attachment to the shaft, which would probably require adding a flat, etc.

Instead I decided to leave the pinion attached and print a matching socket. This required me to work out the dimensions of the gear with the help of calipers and Fusion 360’s “spur gear” generator. It turned out to be a 15 tooth, 0.6mm module, 25° pressure angle gear.

My first version just pressed on to the gear (I printed it with a 0.2mm clearance gap). This worked but was hard to install, worked loose, and rattled.

I replaced this with a more complex joiner which worked very well.

And it even works:

You can’t tell, but the battery is spinning smoothly and quietly. The device is crude, but usable for some purposes.

The software:

This was a very simple bit of code.

There are three inputs; ON/OFF (pull-up by internal resistor, short to ground to activate), direction CW/CCW (ditto), and speed (analog input from wiper of 10K pot connected to +5v and GND). To make the wiring easier, I ran all three into analog input pins. In the process, I committed the noob error of forgetting that A6 and A7 on a nano don’t have internal pull-up resistors.

There are 6 digital outputs; the four stepper control signals, and two PWM outputs to the L298 enable lines. This let me reduce the power of the motor when turning at low speeds, making it smoother.

The hardest bit of the code was that I specifically wanted a soft start/stop sequence. If you turn the ON/OFF switch or adjust the speed control, the program will ramp the speed up/down slowly. This makes it much less likely that what’s sitting on the turntable will be thrown off. If you change the CW/CCW switch while the unit is running, it will ramp down to zero, then ramp back up in the opposite direction.

I got the device rotating using the standard Stepper library, but switched to the AccelStepper library as it had ‘set speed’ and ‘max acceleration’ functions. Then I wasted quite a while before I realised that the SetSpeed() function ignores the acceleration settings, and had to configure the ramp manually.

#include <AccelStepper.h>

#define SpeedPin A3 //green
#define CwCcwPin A4 //black
#define OnOffPin A5 //yellow

#define Enable1Pin 5 //black
#define Enable2Pin 10 //white

unsigned long oldTimeStatus = 0;  
const long statusInterval = 50;

int motorSpeed = 0;
int maxDelta = 2;

AccelStepper stepper(AccelStepper::HALF4WIRE, 6, 7, 8, 9); 

void setup()
{  
  //Serial.begin(9600);
  Serial.begin(115200);
  delay(100);
  Serial.println("Turntable");

  pinMode(Enable1Pin, OUTPUT);
  pinMode(Enable2Pin, OUTPUT);
  digitalWrite(Enable1Pin, HIGH);
  digitalWrite(Enable2Pin, HIGH);
  pinMode(SpeedPin, INPUT);
  pinMode(CwCcwPin, INPUT_PULLUP);
  pinMode(OnOffPin, INPUT_PULLUP);
  
   stepper.setMaxSpeed(300);
   stepper.setSpeed(motorSpeed);	
   stepper.setAcceleration(0.1); //ignored by .setSpeed
}

void loop()
{  

  unsigned long newTimeStatus = millis();
  if (newTimeStatus - oldTimeStatus > statusInterval) {
    char workStr[30];

    int sensorReading = analogRead(SpeedPin);
    // map it to a range from 0 to 100:
    motorSpeed = map(sensorReading, 0, 1023, 0, 280);

    int currentSpeed = stepper.speed();


    // read enable
    bool isON = (digitalRead(OnOffPin) == LOW);

    // read direction
    bool isCCW = (digitalRead(CwCcwPin) == LOW);

    if (isON) {

      if (isCCW) {
        motorSpeed = -motorSpeed; //map to backwards
      }
    } else {
      motorSpeed = 0;
    }

      

      int delta = motorSpeed - currentSpeed;
      delta = constrain(delta, -maxDelta, maxDelta);
      int safeSpeed = currentSpeed + delta;

      sprintf(workStr, "S %03d R %03d C %03d -> S %03d", sensorReading, motorSpeed, currentSpeed, safeSpeed);
      Serial.println(workStr);

      stepper.setSpeed(safeSpeed);

      if (abs(safeSpeed) < 60) {
        if (safeSpeed == 0) {
          digitalWrite(Enable1Pin, LOW);
          digitalWrite(Enable2Pin, LOW);
        } else {
        analogWrite(Enable1Pin, 80); //PWM = 1/4 power
        analogWrite(Enable2Pin, 80); //PWM = 1/4 power
        }
      } else {
        digitalWrite(Enable1Pin, HIGH); //PWM = full power
        digitalWrite(Enable2Pin, HIGH); //PWM = full power
      }
      
      oldTimeStatus = newTimeStatus;
  }
  
   stepper.runSpeed();
}

Was it worth it?

If I wanted a quick and usable device, this wasn’t the way to go about it.

However, I learned a lot.

My notes file for this mini-project has over 50 notes I marked as “Problems” (yes, I made a lot of mistakes), and at least 30 marked as “Possible improvements”, some of which I implemented, some I may use in future projects.

A tumble drier for seeds

This year we had a good crop of Spaghetti Squash, grown from the seeds we saved from one squash last year. Since that had been so successful, we decided to save all the seeds this year. Lots of seeds. This is just some of them.

One of the problems with squash seeds is that they are wet and slippery, and if you leave them to dry in a clump, they all stick together. The previous batch we just spread out by hand on paper towels, which worked fine. With many more seeds this time, I decided to automate the process. Needless to say, the effort involved in building the device was considerably more than doing it by hand would be, but it was a fun exercise and I was able to try out a bunch of techniques.

The first version:

This was mostly thrown together from stuff I had lying around. The main mechanism is a small worm drive motor connected to some M8 threaded rod supported with a 608 (roller skate) bearing. I lasercut some 50mm wheels out of 6mm MDF with lots of little “V” cuts around the edges like a gear, to get some friction. There’s a second set of rollers, freely rotating, on the other side.

I was going to throw an Arduino into the mix when I realised that all I needed was to have the container rotated for about 30 seconds every half hour, something that a simple relay timer board could do easily.

Problems:

There were a number of problems with this version. The most obvious was that there was nothing to stop the seeds from falling out the open container, but also the friction wasn’t enough. It would all turn when unloaded, but not with a load of wet seeds contributing their reluctance to move.

A quick 3d printed cap solved the first problem (the hexagonal pattern is just infill, with zero top and bottom solid layers). I fluked the dimensions on the first try and it fitted tightly and never came off.

The friction was a bigger problem. I wrapped some fabric tape around the plastic container, which helped, but in the end I had to make a really crude fix. I rigged up a ball bearing on a spring that pressed down from the top to ensure rotation. It worked, but was rather ugly.

Drying wasn’t as fast as I’d like either. I pointed a fan into the open end, which helped.

Time for a rebuild:

While the basic mechanism was the same, this time around I made a special effort to try for a tidier setup. This wasn’t just for prettiness sake. I want to get back into some robotics and wiring layout and connection issues can be major issues in that field, so some practice wouldn’t go amiss.

I used a bigger mounting board this time, and included an 80mm PC fan. There’s a small buck converter providing a reduced voltage (6.7V) to keep the fan blowing, but quietly.

All the wires going to screw terminals got crimp ferrules on them which really improved the reliability. I think it also makes it look tidier.

I was very pleased with the little power distribution blocks (visible at the top of the board). These were 3d printed (in different colours, of course) and took a block of 4 terminals off a row of connectors (what’s known as “chocolate block” connectors in many places). Surprisingly, they’re not glued or screwed in place, just a friction fit that takes considerable force to push in and won’t pull out without the use of a pair of pliers. I used the same technique on the voltage reducer board and there’s another connector just for the fan.

Speaking of the fan, the printed shroud was an interesting design task in Fusion 360. My first attempt, using vase mode printing, failed dismally. However this version worked well.

In the view above you can see the corrugated disks which were tapped and then threaded onto the M8 rod. I was short of M8 nuts so they’re actually held in place by laser cut (and manually threaded) MDF nuts. This was very cheap and worked well (except when I didn’t tap them quite vertical), and I’ll use the technique in the future.

I replaced the container as well. I found a chunk of alkathene pipe (83mm diameter) and chopped of a short section. I printed ventilated caps for both ends so the air could flow through. To solve the friction problem, I glued a section of the high-friction rubber matting sold for lining kitchen drawers, etc, around it. Note – I used PVA glue and just laid down a few stripes along the pipe, waited for them to dry, then rotated the pipe and did more stripes. This worked very well and I had no more friction problems.

The Verdict:

This was an interesting exercise, and it did the job I needed. Running for about 12 hours it turned a wet sludge of seeds into a convenient batch of dry seeds. Since it only rotated for 30 seconds every half hour, the seeds didn’t get thrashed around.

Now I can dismantle it back to parts until next year.

Storage for Boot (Trunk) of Toyota Caldina

The boot of our family car is usually full of ‘stuff’. Jackets and blankets in case of a breakdown, tow rope, leather gloves, etc. These tend to get shoved to and fro to make room but flop around and spread out. This annoyed me more than usual the other day, when I had to clear everything out to get access to the floor.

I didn’t take a photo of the mess in the boot, but this is the sort of junk that was in there, hurled over into the back seat.

Looking at the boot, there’s an area just behind the back seat where things could be stored vertically if there was a suitable container.

I took some measurements and used Fusion 360 to work out the angles. I could have drawn up a cad diagram of the unit but it was just as easy to lay it out on paper.

This is what I came up with. It’s a simple box with two partitions. One side is angled to match the slope of the back of the back seat. The other is dropped down a bit to make access easier. Cardboard was the material of choice for this. (The rendering above was done after the fact, while experimenting with Fusion 360’s new sheet metal workspace, which can also be used for cardboard boxes).

I found a suitable chunk of cardboard from our huge collection in the shed and we marked out the main shape and cut it out. Folded up and held with clamps, it looked pretty reasonable, and fitted firmly into the space I’d planned.

When I put it down on the floor, it was immediately inspected.

The weak point will probably be the back (front as you look at it) wall. It’s already got two layers of cardboard over much of it, so I filled in the gap, then covered it with another complete layer. Probably overkill, but cardboard is cheap and light. We slathered it with pva glue and weighted the sandwich of layers down with exercise weights and wood.

I flipped it the next morning, and glued the angled face. The partitions were added with more pva and some tricky clamping. I added a rim of clear tape just for looks

This is what it looks like in place for a test fit. I mucked up the measurements slightly and had to cut a notch for the handle which releases the sunshade fitting. A lot easier to change in cardboard than something harder.

And this is what it looks like full of stuff. It absorbs a heap of bits and pieces and still leaves most of the boot clear.

 

Christmas Lights Obelisk/Pyramid

Barbara wanted some pretty Christmas lights. It was a bit late for Christmas (i.e. it was January). So?

We started with some 100×25 Oregon, ripped in down to 50×25. Here I’ve chopped four 1600mm lengths to form the sides of the pyramid/obelisk. We found a nice looking christmas tree on Google Images then used a graphics package to measure the angle – turns out that 80°  from horizontal was the angle we thought looked good.

Some simple trigonometry gave us the size of the base (close enough to 500mm), to get the angle we wanted. Some more 50×25 oregon and various offcuts from the mitre saw gave us a base.

We braced it up for strength, since I planned on placing one or two concrete blocks onto the base to weight it down. (We get strong winds around our place).

Some 7mm holes and M6 bolts attached the ‘legs’ to the base. Forgot to allow for the 25mm offset from the ends of the pieces of wood, so our angle won’t be quite correct.

A very sophisticated (not) mechanism fastened the top together. One zip-tie.

Here’s the result so far, with a human for size comparison.

Chicken wire! Messy stuff to work with, but a compressed air stapler made it easy to pin it down. We only put wire over three sides, leaving the fourth side open to allow placing concrete blocks (for weight) and christmas tree lights (for pretty).

We put it up on an earth bank overnight, to check it wouldn’t blow down.

Barbara fitted the lights through the chicken mesh. Took quite a while. Here we ran a quick test in a darkened room. Looking good!

Come nightfall, it was very pretty.

Laser cut boxes to tidy kitchen bench

A quick project, most of it done the night before a party.

My wife complained about the mess on the kitchen bench. Lots of little bits of “stuff”, of many sizes. I thought about making a partitioned tray, but decided we might do better with an array of small boxes, allowing flexibility to use the space efficiently, and to swap boxes around if one section outgrew its container. Also, as anyone who’s had to get pointy things (e.g. drawing pins) out of a fixed compartment will attest, it’s much easier to lift up a box and empty it out.

 

We weren’t too sure what would be a good base size for the boxes, so I ran off a few samples of different sizes. We tried a 40mm x 40mm, and 50mm x 50mm, but quickly found that 60mm x 60mm worked well. We settled on 30mm high. Since these were just test boxes, I didn’t bother flattening the top edge, as that was the slowest step. These are all made from (cheap) 3mm mdf.

All of these boxes were designed with the wonderful (free) “Tabbed Box Maker” extension for the (free) Inkscape program. http://www.inkscapeforum.com/viewtopic.php?t=18315

The interface is very simple – just put in your dimensions and instantly get all the tabs worked out. It’s *much* slower doing it by hand, as all the tabs have to be a bit wider than the slots, to account for the laser’s kerf (cutting width).

The laser cuts quickly, but do enough cuts and it adds up. For 5 of the 60×60 boxes and 5 of the 120×60 boxes, the cutting took 17 minutes. I really need to realign the laser as I’m sure I should be getting better speeds. Here are the 50 pieces required.

And here is the remainder after I cut them out. This is a 600×400 piece of mdf, worth about $1. Alert viewers may notice a problem with the shapes. I carefully flattened one edge of each side, so the the top of the box would be flat, not crenelated. However, in a moment of dumb, I flattened one edge of the base (far left) as well. Had to recut those.

Here are a few of the boxes. Default settings in tabbed box maker gave me parts that fitted together easily but tightly, and could be squeezed into firm position. They held together quite well, but I added a bit of PVA wood glue just for certainty.

The result looked very promising. I burned a second set of 5 x 60×60 and 5 x 120×60, and one long 300×60 for scissors and a letter opener. Much tidier.

I wanted to store pens and markers as well so I made a double height box (60x60x60) and some dividers.

When you get the measurements and calculations right, laser cut parts just slide together in a very nice way.

 

Here’s the final (so far) result. It does spread things out more than the original piles, but at least you can find things. When we decide it’s finished growing, I’ll make a laser cut tray that just fits around the whole set.

Straw and cardboard vegetable bed

As a bit of an experiment, we knocked up a couple of beds for vegetables.

---

 

The outer frame is just four length of Oregon 100×25, joined at the corners with screws. I added 8x 45° braces in the corner as well. The result was fairly rigid. I did try nailing them together with my framing nailer but it didn’t work well. Here you can see Barbara watering the ground inside the frame, which is simply plonked on the ground.

We covered the ground with a couple of layers of cardboard from packing boxes.

We marked out a grid of points (about 200mm spacing). Then I got out an electric drill and a 75mm (3″) holesaw and drilled 48 holes through the cardboard. The slowest part was digging the cardboard disks out of the holesaw.

I didn’t have any proper soaker hose, but this “sprinkler” hose was on special so we wound that between the rows. Once the straw goes on, it should still let us water under the mulch.

Then the planting! Here Barbara and Rachel are placing seedlings (cabbage, beans, etc) and putting pea straw around them. I dibbled a bit of dirt out of each hole with a grubber, and Alexandra recorded which seedlings went where. Ignore the messy wood pile in the background.

Final result doesn’t look to bad. This represents about $5 of wood, $2 of screws, a $7 sprinkler hose, and about $20 of seedlings.

If it works, we can easily replicate a bunch of these.

 

 

Seedling propagating bed – two week update

First – there are seedlings growing!

And quite a lot of them. The first seedlings came up after about 7 days, and now there is at least one seedling growing in 215 of the 350 pots.

An automatic cutoff when the bed is flooded.

This is a simple float switch. When the float (a section of pool noodle) is lifted by the water, the wire rotates around a bearing and presses the microswitch which cuts off power to the pump. There’s a “control panel” visible to the left. Pressing the square button powers up a relay which turns on the pump, but it also supplies power to the relay so it keeps running after you take your finger off the button. When float rises, it cuts power to the relay which switches off, and stays off until pressing the button starts the cycle again. There’s also an override switch, a fuse, and an LED to light up.

FISH!

As the water flowed across the bed, then sat in the reservoir, it started forming algae. We started considering options to stop the buildup, then hit on the idea of advancing towards our eventual goal of aquaponics by adding a couple of fish to the system. We’ve always intended to have goldfish as part of the larger system we’re planning, but we had to do some furious research to see if it was workable.

The reservoir is renamed “the fish tank”

A visit to a pet shop netted us two quite small fish, some fish-food, a thermometer, and an air pump. I modified an old sieve by cutting holes for outlet pipes and wires, then placed two water pumps (the original pump and a smaller one) inside the sieve. A collection of stones from the beach at Birdlings Flat and a couple of chunks of brick all serve to hold the sieve firmly against the bottom of the tank. I’m fairly confident that the fish won’t be able to get near the pump intakes. The air pump is outside the tank, with a hose leading to the bottom of the tank. We used it before the fish moved in to make sure the water was well oxygenated but haven’t kept it up as (a) it was really too loud and vigorous, and (b) it ran off 240V whereas everything else is running off 12V (solar) power. As an alternative, the second (small) water pump inside the sieve feeds out to a hose which splashes water back into the tank, adding oxygen in the process.

The algae has gone, the fish seem happy, and they have definitely grown.

Anti-cat barrier and air circulation fans.

We strung some loose strands of bright 4mm rope on three sides to dissuade the cats from jumping up onto the seedlings. Seems to have worked so far.

We’re quite worried about “damping off”, the fungal disease(s) that like to destroy young seedlings. The seedling pots are quite moist and, while we haven’t had any problems so far, conditions are quite good for fungus to develop. Many sources recommend fans moving the air about to dry the top layer of soil, prevent stagnant moist air from sitting around, and also toughen up the young seedlings. I didn’t want to use a mains powered fan so I grabbed eight PC fans salvaged from various computers and hooked them up. They’re hooked up in pairs, with each pair joined in series so the fans, designed for 12V, only get half the voltage (6V) and turn over slowly and quietly. It took a surprising amount of time to hook them up but they seem to work very well. The biggest problem, and one that almost caught me out, is that each pair of fans should be of similar power. I was amazed how many different ratings there were in the my fan collection – each of the four pairs in use here is a quite different rating.

More controls

Each step we take seems to require more controls, especially adding the fish. I really need to add some timers to the system, and perhaps a water level detector for the fish tank. Perhaps a moisture detector to tell how damp the soil is. Lots of options. I’ve started working up a system based on an Arduino microcontroller.

A prototype seedling propagating tray

A recent project has been building a “flood and drain” bed for raising seedlings. It’s all rather rough and experimental, but showing promise.

Newspaper pots. Lots and lots (200 here, 350 when complete) of little biodegradable pots. Each group of 50 sits in a “seedling tray” which is has an open grid at the bottom.

The structure is built of 18mm MDF, with two layers of polythene sheet to make it waterproof. The polythene is trapped by a thin strip (18mm x 10mm) of mdf which is screwed down. That applies lots of even clamping pressure to the film so it doesn’t rip. Also visible is the hose that delivers water from the reservoir (currently a 50L “fish bin” just visible below the table).

If the bed fills with water, there will be quite a bit of pressure trying to bend the walls outwards. I was going to put in cross braces, but ended up just using a strong method of fastening the walls. Vertical steel bolts (M6x65) come up through the base and into TAPPED holes in the walls. I recommend this technique for anyone who wants to really lock an mdf construction together. Note: the holes were drilled 5mm, then tapped with an ordinary M6 metal tap in a cordless drill. Very fast and very solid.

This all sits on a chunk of material from an old office desk, which in turn rests on two homemade trestle legs (made from pallet wood). The tray is 2200mm x 600mm, and the walls are 100mm high. There are seven seed trays along the length, and an area at the extreme left which is where the drain is fitted.

Inside the tray we routed some grooves along the base, hoping that the water would press the plastic down into the grooves and help to drain the bed smoothly. It didn’t. A simple work-around was to string some 4mm poly rope underneath the seed beds. This lifts them up enough that there is no water permanently pooled around the base.

The seed trays don’t go all the way to the end. The last 200mm or so is routed down about 4mm deep to act as a drain, and has a “bulkhead fitting” installed to connect to a hosepipe. This was one of the points I was most worried about, as any leak here would get into the mdf and really damage it.

One of the things about polythene is that it’s really hard to glue to. Glue, or silicone caulking, won’t stick to it. I did some experiments with gluing a disk of plastic to the polythene which did work (light sanding, alcohol wipe, blowtorch at a distance to react free radicals with the surface layer) but in the end I decided to just trust the pressure fit of the rubber washer in the bulkhead fitting. So far (though it’s very early days) it’s been fine. If I build another one, I’ll epoxy a few inches of the mdf around the hole so it’s waterproof.

The bulkhead fitting is about 4mm high which prevents draining the last of the water away. Half a dozen pieces of ordinary string (held down by an old 12V battery which proved to work better as a heavy weight than as an electrical device) wicks the last of the water away perfectly. The plastic is dry at the end of the draining period.

DO NOT LOOK AT THE “TEMPORARY” WIRING!

Just in case a leak develops, the whole apparatus is sitting inside a rectangular ring of mdf, again with polythene inside to form a catch basin. The screwed down strips to hold down the plastic on the tray worked well – much better than simply stapling it as we did for the catch basin. Even when stapled through little bits of cardboard, the plastic rips free if you look at unkindly.

For safety, this is a solar powered setup, in this case with a hefty 12V battery intended for a lawn mower or similar ($100 from Super Cheap Auto). Just visible at the back (by the window) is a 1.8W 12V solar charger ($20 from The Warehouse).  I’ve actually got a proper solar panel and controller I can hook up if it needs it.

The tray is a fraction longer than the support it rests on, leaving room for the bottom of the bulkhead fitting in the drain section. This leads into some perfectly ordinary domestic hosepipe. Partway down the hose is a tap fitting. This can restrict the speed with which the bed drains. The more it is closed down, the longer the ‘flood’ stage of watering lasts. Currently it takes a few minutes to fill the bed about 10mm deep with water, and about half an hour for that to completely drain away.

Inside the reservoir is one of these, a 12V, 840L/hour, BRUSHLESS dc motor. The advantage of a brushless motor is that there are very few moving parts. Basically there’s some circuitry inside that turns DC into AC and uses that to spin the rotor. The supplier claims 30,000 hours operation, compared to some brushed dc motors (e.g. bilge pumps) that often fail after about 200 hours. There are lots of discussions about the problems of brushed dc motors on hydroponics websites. These were about $25 off Ebay.

If you wondered about the little paper pots, these are made from newspaper strips with the gadget shown here. It’s just a slightly tapered former with a hollow at the bottom and a handle, plus a matching base. I turned these (on a metal lathe) from some branches cut from our eucalyptus trees. There’s various instructions for making them on the net, or you can buy one (e.g. from here). The plan is to plant the seedling complete with the paper pot which should save a lot of fuss.

Hopefully I can put up an update in a few months to say whether all this worked. I have high hopes, but time will tell.

Worried about hornets

We’ve been reading alarming things about the Asian Giant Hornet. They are huge! 50mm long (2 inches), with a 75mm wingspan (3 inches), and a stinger 6mm long (1/4 inch).

Nasty creatures. And the reason we worry is that we plan to get some of these:

Nice, friendly, dumb, but easy meat to giant killer hornets.

The answer is pretty obvious. A 6mm stinger may go through a sheep’s wool, but it won’t go through the protective scales of one of these:

A Pangolin – scaly but oh so cute.

A bit of glue and some wool, and we will have the Q-ship of the sheep world:

(Actually, that’s a wool-coated armadillo not a pangolin, but you know what we mean).

Any giant hornet that stings that beast will regret it – instant sting breakage will ensue.

BUT

Do the stings on giant hornets grow back? We really don’t know and it makes a big difference. If they don’t grow back then we’ll be better off hiring a pangolin or two from Pangolins-R-Us for part of the season. But if the stingers do grow back then we should clearly buy our pangolins outright.

Does anyone have any advice?

---

Disclaimer: New Zealand doesn’t actually have giant hornets. Or pangolins. It does have millions of sheep, and we do intend to get a few. However, this was just one of those strange conversations one sometimes has, in the car, on the way home.