klog

I have solar panels on my house

And an electric car

If I have 2kW power coming in from the panels, but am using utensils that total 3kW, then the house needs to pull in 1kW from the grid to make up the difference. That costs money, making it a bad time to charge the car.

Alternatively, if I have 3kW coming from the panels but am only using 2kW in the house, then it’s not currently costing me anything. I actually get paid for the excess 1kW which gets sent out to the grid. But it’s still a bad time to charge the car, because it uses up to 2.5kW, so I’d be pulling in 1.5kW from the grid and paying for that.

But if I’m pulling in 4kW and only using 1.5kW in the house, then plugging the car in will be free.

So, sometimes it’s free to run the car charger, and sometimes it would cost money.

Doing these calculations is an annoyance. Especially as the total input changes throughout the day as the clouds come and go, and we turn various things on and off.

I don’t want to recalculate and turn the charger on/off every few minutes. That’s what computers are for!

The commercial solution for this is that you buy something designed for that purpose, such as the MyEnergi Zappi. And I really do recommend you do that, if you have the money and prefer something that just works. It basically looks at what the house is doing, and decides for itself whether the car should be charged, based on calculations and other decisions.

But, if you are like me and like to do things cheaply and hackily, then let’s build a cheap device that does (almost) the same thing, but at a fraction of the cost.

Materials needed

You will need a travel charger, also known as a granny cable. I use the Zencar charger which I managed to buy for €150. Any travel charger will do as long as it can be plugged into a socket in your house.

You will also need a wifi-enabled plug adaptor. I got a BG Home adaptor for about €20. Make sure that it is IFTTT-compatible. You’ll need that.

You will also need a computer that’s permanently on. I have a Linux machine I use as a file server. I used that for this purpose. You could also use a cloud server. A Linux server from Digital Ocean can be as cheap as $4 per month

And then you need a way to measure what power is coming into the house through the panels and the grid. I use a MyEnergi Eddi power diverter which was installed when the solar panels were installed. If you’re using something different, then adjust the code as you need to.

Setup

First thing you need to do is plug your travel charger in through the wifi-enabled plug adaptor. Follow its instructions to make sure it’s working as normal. After that’s set up, you should be able to enable/disable the adaptor (and therefore the charger that’s plugged into it) using an app on your mobile device.

Next, create and log into an IFTTT account.

Once thats done, you can start creating a webhook for turning the charger on/off using a URL.

Click the CREATE button on the top right of the IFTTT page. You will be shown a page with an “If This … Then That” message.

Click the “If This”. Search for and select “Webhooks”, then select “Receive a Web Request”. Give it the event the name “evcharger_on” then click Create Trigger

Click the “Then That”. Search for and select “BG Home” (or whatever is your wifi adaptor’s product). You may need to do some set up for that before it lets you go further. When done, choose “Turn Device On/Off”, and fill in the small form.

You should now see something like this:

We’re almost there. Click to Continue, and do whatever is needed to complete the applet and return to your main account page

To build the URL, you need to know your user API. It’s not obvious how to get that, but as of this minute, you should find it here.

The bit I’ve scrubbed out in red is your API key.

Now, construct a URL like this:

https://maker.ifttt.com/trigger/evcharger_on/with/key/{{KEY}}

Obviously, replace {{KEY}} with your own.

That URL should turn your charger on. Check it in a browser!

Now that that’s done, repeat the above but this time, name the trigger “evcharger_off” and have it turn off the adaptor.

Now that you can now turn your charger on/off with a browser, let’s write a program to do it!

Programming

First thing we do is set up a config file. Save this as config.php

<?php
define('MYENERGI_KEY', '{{MYENERGI_KEY}}');
define('MYENERGI_SERIAL', '{{MYENERGI_SERIAL}}');
define('IFTTT_ON', 'https://maker.ifttt.com/trigger/evcharger_on/with/key/{{KEY}}');
define('IFTTT_OFF', 'https://maker.ifttt.com/trigger/evcharger_off/with/key/{{KEY}}');
define('MYENERGI_URL', 'https://s18.myenergi.net/cgi-jstatus-E');
define('CHARGER_W', 2500);
define('W_MAX', 100);
define('CHEAP_TIME_START', 2);
define('CHEAP_TIME_END', 5);

You should adjust the first four lines above, using your own keys. You can get your own MyEnergi details by following this article.

CHARGER_W is the maximum wattage the charger will output. In my case, I’ve set 2500W, which is slightly above 240v * 10A.

W_MAX is how many watts you’re okay with paying for, if there’s not quite enough power to charge the car completely for free.

CHEAP_TIME_START/CHEAP_TIME_END are the hours when your local cheapest tariff begins/ends (that you’re okay with paying). In my case, I’m with Bord Gáis, who have an EV Tariff which runs between 2am and 5am. They’re currently advertising an EV rate of 10.40 per kWh vs a standard 40.82 during normal hours

Once you have that sorted, create a file in the same directory that contains the actual program we’re building. I called mine update-status.php because it’s updating the status of the charger. yeah. not good at names.

<?php
require_once 'config.php';

function getData() {
	// { get current info
	$ch=curl_init();
	curl_setopt_array($ch, [
		CURLOPT_URL=>MYENERGI_URL,
		CURLOPT_FOLLOWLOCATION=>1,
		CURLOPT_RETURNTRANSFER=>1,
		CURLOPT_SSL_VERIFYPEER=>false,
		CURLOPT_HTTPHEADER=>[
			'Content-Type: application/json',
			'Accept: application/json'
		],
		CURLOPT_HTTPAUTH=>CURLAUTH_DIGEST,
		CURLOPT_USERPWD=>MYENERGI_SERIAL.':'.MYENERGI_KEY,
	]);
	$res=curl_exec($ch);
	$obj=json_decode($res, true);
	// }
	return $obj;
}

$prev=intval(@file_get_contents('/tmp/ev-charger'));
$on=$prev;
echo "start\n";
$obj=getData();
$msg='failed to retrieve data from MyEnergi Eddi diverter';
if (isset($obj['eddi'][0])) {
	$eddi=$obj['eddi'][0];
	$grid=intval($eddi['grd']);
	$boiler=intval($eddi['div']);
	$on=0;
	$msg='charging car would use too much grid|grid='.$grid.'|charger_w='.CHARGER_W.'|prev='.$prev.'|w_max='.W_MAX;
	if ($grid+CHARGER_W-($prev*CHARGER_W)<=W_MAX) { // if charging would not cause us to use the grid
		$on=1;
		$msg='can charge without using much grid';
	}
	else if ($grid-$boiler+CHARGER_W-($prev*CHARGER_W)<=W_MAX) { // the boiler is on. turning the charger on will automatically turn the boiler off
		$on=1;
		$msg='disabling boiler to charge car';
	}
	else { // check the time to see if we're in a low-tariff time
		date_default_timezone_set('Europe/Dublin');
		if (CHEAP_TIME_START<CHEAP_TIME_END) {
			if (intval(date('G'))>=CHEAP_TIME_START && intval(date('G'))<CHEAP_TIME_END) {
				$on=1;
				$msg='ev tariff';
			}
		}
		else {
			if (intval(date('G'))>=CHEAP_TIME_START || intval(date('G'))<CHEAP_TIME_END) {
				$on=1;
				$msg='ev tariff';
			}
		}
	}
	if ($prev!=$on) {
		file_put_contents('/tmp/ev-charger', $on);
		if ($on) {
			echo file_get_contents(IFTTT_ON)."\n";
		}
		else {
			echo file_get_contents(IFTTT_OFF)."\n";
		}
	}
}
echo $prev.'|'.$on.'|'.$msg."\n";

echo "finish\n";

That’s it!
When that’s run, it will check the MyEnergi cloud service to see what’s going on with the house at the moment, and will turn on the charger if it’s not already on.

The last think to do is to just set up a cron job to run that once per minute. I’ll leave that to you.

Example

Here’s today’s grid usage so far. the orange on the top is power coming into the house from the grid, and the yellow is excess solar-generated power being sent out to the grid.

the solid block on the left is between 2am and 5am, when power is cheapest (10c per kW), so the program turns the charger on.

next we have a growing yellow curve that increases until it hits about -2.5kW at 9am. then the car starts charging and the excess drops .

about 10am, I unplug the car as we need to go do something in town. you can see the excess spiking downwards as there’s hardly anything in the house running. there are clouds, so it’s unsteady.

at 11am we get back to the house, and it’s time for pizza! so we turn the oven on, which spikes the power to a pull of 3kW from the grid for a few minutes. My wife has also turned on a washing machine …let’s not comment about her timing…

from that point onwards, the car is basically charging. Turning on/off as cloud-cover fluctuates and the draw from the grid goes above zero.

Conclusion

The thing to pull out of this is that by automating the charging of the car, we essentially charge it for free, letting us concentrate on other things and spend less time worrying about the right time to do things. Just get into the habit of plugging in when we get home, and the car just stays charged.

Obviously there are still cases where this is not enough – if you’ve done a very long drive one day then need to do another very long drive the next day, then there is obviously not enough time in between for the car to recharge for free/cheap. In those cases, just go to your local rapid-charger and top it up.

I designed and built a computerised telescope. Total cost of the materials: €38.40 (including shipping!)

Bill of materials. All the files.

I’ve been working on photogrammetry recently and needed to get some photos of things that are hard to see from ground level, so bought a really cheap zoom lens for my camera.

Turns out it wasn’t great for photogrammetry, but it got some really good photos of the moon!

But, that scope is designed to attach to a phone so it is horizontal, not vertical as would be needed if looking up at the sky.

It occured to me that if I design an eyepiece mount for the scope to attach a small computer such as the ESP32-Cam, then I could point it anywhere I want without risking the camera falling off.

So I did:

I then decided I’d build a full-on mount for the whole thing.

I needed to find the center of balance of the scope and build a strap there that could be used to hold the telescope when turning it up or down:

Notice in the above picture that I have bolts coming out of the sides and a rectangular block there. The bolts are so I can connect the scope to a scaffold, and the rectangular block lets me slot a geared arm on so when the arm is moved, the scope is moved as well. This controls the “altitude” of the telescope:

The next step was to build a very large geared base that the telescope would rotate on. Notice that I have gears on the inside and a groove going all around the top of the circle:

I then created a platform that would sit on that base, and “rollers” to act as wheels (with 10mm pieces of 4mm steel for axles). and attached stepper motors in place to control the azimuth (rotation of the base) and altitude (rotation of the swing assembly):

The final physical work was to create mounts for the electronics needed to control power, then install those electronics and cable it up. I got tired of designing and ended up gluing a few last things in place. My design, my rules. Cat is optional:

A few important things to note about the design are that
1. it’s battery-powered, so there’s no awkward power issues
2. the ESP32-Cam has Wifi on board, so no awkward cabling needed
3. I added a battery charger to the setup so there’s no need to replace the battery (the 18650 battery holders don’t like to let go of the batteries…)
4. the ESP32-Cam has 10 GPIO ports. 2 of them don’t work as far as I can tell, and 1 of the others is linked to the onboard camera flash (very bright!). I need 8 for the project. So, the camera flash now lights up every time the telescope rotates. annoying but I didn’t want to do any circuit surgery.

There were a number of ways I could approach programming. I chose to use a simple idea:
1. the ESP32-Cam uploads a photo to a website every 10 seconds, and downloads and executes move instructions. simple. this way we offload all programming to the website itself.
2. the website displays the most recently uploaded photo to the user, and shows the current azimuth and altitude of the telescope.
3. the user can request the telescope to move to a new azimuth/altitude.

All code and design files were uploaded to GitHub.

Ireland being Ireland, I then needed to wait for the clouds to go away so I could get a photo of the moon.

update (2021-01-03): The first proper photo was taken last night with the camera in SVGA mode. I had to adjust the settings to reduce the exposure quite a lot:

Not terrible, but the camera also has a UXGA mode. Here’s a clipped version:

I’m not sure why that turned out green, but then the camera is probably not designed for looking at the sky at night with low exposure settings.

It would be possible to improve the telescope without replacing too much by changing the OV2640 camera for an OV5640 instead. that’s a 5MP camera (vs the 2MP of the OV2640).

Overall, I’m happy with the project. The end result can probably be cleaned up a bit, but I’ll leave that for another project when I’ll probably rebuild the whole thing and add some new features such as tracking, GPS and compass modules, and automated star-finding. And a better camera.

I’ve been asked to write down a kind of history of events leading to today with the face shields project I’ve been trying to stay on top of.

On March 24th, I called my local hospital (Monaghan Hospital) because I wanted to help out. I was given an email address to send to. I just found the email I sent:

In it, I said I probably had enough filament to offer 25 shields, and had purchased more so might be able to provide 100 total. (haha – I laugh.)

I had been discussing this idea with some people in the “3D Printing Ireland” group in Facebook, where some other people were also contacting their local hospitals to offer help.

We half expected to be ignored. Who are we? Just hobbiests working in sheds and attics. Who are we to try offer our help to the real experts?

My email and details were passed around to a few people before I was finally in contact with Lorraine Brady, who liased with Cavan and Monaghan hospitals for me. We arranged to meet, and I gave her three separate designs (I think) to bring to her infection control team to examine.

There are a number of popular frame designs that people could choose among. Each of them is a simple design where a transparent A4-sized plastic sheet is “hung” onto nubs in the frame which are positioned so that a common paper punch can place the holes correctly.

The first I became aware of is the “Prusa” model, named after the 3D Printer company called Prusa. I wasn’t over-fond of their design because it was very bulky, but it was tested in conjunction with the Czech health authorities, so I offered that as a design.

Secondly, there is a company called 3D Verkstan, a Swedish company specialising in 3D printing and design. They designed a frame that was about 1/3 the height of the Prusa model. I liked this one, as it was compact and relatively quick to print. Again, theirs was designed in conjuction with local health authorities.

The final design I tried out was one that was a bit like the second but more curved. I’m not sure why, but the original designer of this model pulled the design from websites shortly afterwards.

Lorraine came back to me on the 26th of March and said that the infection control team were happy with the Verkstan model. The only change they wanted was that the visor itself should be cut an inch shorter and the corners rounded on it so it doesn’t catch on clothes.

The Verkstan model is very quick to print. With barely any tuning on my printer at all, I could make one every 50 minutes. The Prusa model, though, took almost 3 hours to print one.

I started recording deliveries on the 29th of March, when I gave Lorraine 42 face shields. I would then give her an average of about 25 shields per day for the next few weeks.

In the beginning, I’m told that the nurses were using the shields once and then discarding them, as is normal procedure in hospitals. But these are not normal times. They were asked not to do this, after the infection control team investigated various ways to clean them so they could be re-used.

Because I (and most people) print using PLA, normal autoclaving is out of the question – the temperature causes the plastic to warp. Thankfully, warm water and soap is all that’s needed to destroy a virus, so the hospital was able to start re-using the frames.

This was important, because it allowed me to then start offering face shields to other people.

It occured to me that I could ask the local council for help, and with a record of delivered face shields, that would give me a stronger case, so I contacted a friend, Paul Bond, who I thought would be a good go-between to start from. Looking back through my records, it looks like I contacted Paul on the 26th of March (earlier than I thought). He immediately said to just go ahead with what I was doing. He knows me through other volunteer work (particularly Monaghan Coder Dojo), so I guess he knew I wasn’t joking around.

Monaghan Town Team is a local events-organising group tied to the council. Paul explained that there was money left over from the St Patrick’s day parade which had been cancelled because of the virus, and whatever I did would be covered from that.

On the 30th, I gave 10 face shields to Dungannon Nursing Home. I actually felt a bit guilty about this, because the hospitals were where all the action was, right? Wrong. Turns out the nursing homes were equally in danger, but they were worse off with regard to PPE (personal protective equipment), as the HSE were not supplying nursing homes.

As more requests came in from other groups, Paul took up the task of delivering the shields. I don’t drive, and he had permission to drive around as he was volunteering his services to deliver pharmaceutical supplies to people.

The requests started to grow, and I needed more printers and filament.

I paid for a printer (an Anet A8 Plus) out of the Monaghan Town Team funds along with a few spools of PLA. In normal times, I get through 1kg of PLA in about a month. I was not sure how that would change. It turns out now that I’m getting through about 3kg every single day. 90 times as much. But I was conservative with my orders in the beginning because I was fully aware that I was using other peoples’ money, and I’ve never been comfortable with that.

I asked a few friends to lend me their printers, and they did! On the 28th, I was given two Anet A8s and an Anet A8 Plus (printer models). Unfortunately, they were each broken or not fully assembled, so work was required. Over the next two days, I got two of them up and running, but one of the Anet A8s was broken beyond repair (the motherboard was burned out).

Una-Minh Kavanagh of WeAreIrish.com wrote an article about what we (the 3D printing community) were working on. Various other news papers also wrote articles. Northern Sound, Anglo Celt. I remember speaking to someone from AP, and one of the guys who loaned me his printer “volunteered” me to speak to Joe Duffy (radio broadcaster). I’m told that was well-received.

The local Monaghan Town Runners group offered a large donation through Paul (a member of the group). I did not ask for the money, but gave a list of materials that I needed for the project – four new printers and a load of PLA – and they went ahead and ordered it!

A friend had asked me why I didn’t start a fund-raising campaign. I’m not comfortable touching other peoples’ money, so I much prefer to spend my own money, and let other people donate material instead (like how I handled the Town Runners offer).

But, it also occurred to me that there were a lot of frustrated people out there that wanted to help but didn’t know how, so creating a fund-rasing campaign so they could donate and I could use the money raised to buy materials and tools for the project actually made sense.

I created a GoFundMe campaign on the 12th of April, asking for €500, as I thought that would be all that I need to cover whatever requests come up.

As of today, the figure raised on that is €4580 (of which €4418 is available after fees), and I have spent about €4000 of that, on about 100kg of filament, thousands of binder covers, one more printer (still waiting on that…), various replacement parts for broken bits on the other printers, and some tools to help speed up cutting of visors.

The orders were starting to grow, so I needed to find ways to produce the shields faster and with less effort. Here’s an example of about 6 days of orders (Apr 8 to Apr 13):

The Verkstan model takes about 50 minutes to print on a standard printer with the default 0.4mm nozzle. This meant that I could produce at most 16 frames per day, and needed to be next to the printer at all times so I could reload every hour.

Luckily my day-job is in computers, so it wasn’t a big deal to work from the shed and just take a few minutes every hour to reload the printers.

But those 8 hours of sleep bothered me – there must be a way to “stack” the frames such that I could get some sleep and wake up to a load of frames ready to separate.

“Stacking” printed objects is not straight-forward. If the space between objects is too small, they will bond together and break if you try to separate them. If the space is too large, then the printer will start printing “spaghetti” and you will just waste filament.

Even if you get the space just right, there are additional complex problems associated with curves, that are not straightforward to solve.

It took me about a week of testing and fine-tuning to solve the problems. In the end, I was able to create a stack of 20 face shields, separated by 0.6mm, with a tiny bit of carefully placed “scaffolding” in between the shields to help keep the shape. Combined with a larger nozzle size (0.6mm), I managed to reduce the print time to about 30 minutes per frame.

The result was not as smooth as when the frames are printed out one at a time, with stray plastic “hairs” making the frames scratchy, but the fact that I could now print out a lot of frames every day made this a price I was willing to pay. And even as I felt guilty for giving out scratchy frames to nurses and carers, I also felt it was justified because I could give out more of them. Also, I wasn’t charging. What are they going to do – demand their money back?

I managed to increase the output from 16 per day per printer with hourly reloads, to more than 40 per day per printer with just 2 reloads. I have 6 printers running right this minute, creating at least 240 frames per day.

On April 24th, I wrote a post explaining the various tricks I used to speed up production. Partly so other people can use my experience, and partly so I can remind myself if this happens again.

The donated printers were all Anet printers, just like my own. Unfortunately, those printers are very cheap but not very good. I like them myself, but that’s because I like to tinker. So when I bought one more printer, I got yet another Anet A8 because it’s what I’m familiar with and comfortable fixing.

Resources were very low a few times over the next few weeks. I had underestimated both how quick I was burning through PLA and how slow the post had become because of the virus. At two points, I had to turn off some printers because I did not have any PLA for them to print with.

I started overcompensating for that, ordering larger amounts. I now have enough filament on stand-by that I can keep printing for about 3 weeks before running out. That’s better than needing it and not having it.

Over time, the orders started getting fewer but larger. I started delivering to some larger hubs such as the Armagh COVID-19 Response Committee, and the Bravo Charlie Tango – Bikers Coming Through groups.

In total, I have given away about 7000 face shields so far. I am not sure yet of all the accounts, but the cost per face shield has averaged at between 50c and €1.

I’m still printing at full pelt, and expect to continue for the next week at least. I hope there isn’t a second wave, but if there is, I am prepared to continue.

I fully expect that all material I have paid for (and will continue to buy until the fund run out) will be used. People are going to be asking for shields for weeks after the “peak” has been reached. So far, every single one that I have produced has been donated within a day or two, and I don’t expect that to stop for a few weeks.

Financially, I’ve received help from Monaghan Town Team, Monaghan Town Runners, a GoFundMe campaign, a donation of €500 from Bank Of Ireland (through NewsTalk), and a grant from Monaghan Council.

More photos documenting this project: mine, Bronwyn’s.

I’m sure I’ve missed a lot in this – if there are any questions, please ask!

Over the last few weeks I’ve given away more than 3000 face shields to hospitals, nursing homes, home-care companies, vets, and even a few fast-food companies.

In order to ramp up production, I’ve had to figure out ways of producing more shields, with less effort, while keeping a good standard of production.

The tricks here will help you speed up. The results will not be perfect, but they will be fast and scalable. This article assumes you are already producing face shields and juts want some ideas to speed up. I’m not going to walk through the basics.

Visors

The visor is the simplest piece to produce. Take an A4 clear PVC binder cover, cut an inch off the side, round the edges on that side, and punch holes in the other side in precise places based on what frame you are producing.

Speedup level 1

To cut many sheets to size at the same time, take a prepared visor sheet and place it on the near right corner of a work surface you don’t mind cutting into (if you’re left-handed, reverse these directions).

mark a line along the top edge of the visor in heavy marker. stretch the line out a few inches to the left, and all the way to the right edge of the table.

take a pile of maybe 20 sheets and square them to the table corner. place a metal ruler on top so it is in line with the marked line on the table. clamp the ruler down (you can get reasonable clamps for about €3 in any discount store).

you can now run a craft knife along the ruler edge to cut a few sheets at once. Angle the blade gently in towards the ruler, or the cut may veer away.

To quickly and accurately punch up to 3 sheets at a time: Buy two punches that are the same make and model. get a small wooden plank. screw one of the punches to the plank. carefully place the second punch on the plank so that if you put a sheet in them both and press down on both, you get the correct holes. Screw that punch down as well. Now you don’t need to measure. just push the sheet in, punch punch done.

to cut the curves, just use a scissors right after doing the punching.

Speedup level 2

Here’s where I am at the moment.

Buy a pillar drill and a band saw. They’re about €149 each in Lidl. Well worth it if you’re at the stage where you’re producing so many shields that visor cutting is eating up hours of the day.

Create two “templates” by getting two plywood or MDF sheets and drawing a prepared visor onto them, then very carefully drill 5mm holes into the correct place through both the sheets. Get this right. Then you can cut out the rest of the shape of the visors. This part is not important to get right.

Take a full block of sheets (including its wrapper) and place between the templates, then clamp this in place.

Put the 5mm drill bit in the pillar drill. With the drill off, carefully pull down the drill until it is in place in the template and touching the visor. This is to make sure you’re aligned. Lift up again and tun on the drill. Push the drill bit down through the vinyl in a few passes, each pass going a little further. This gives the drill time to evacuate the plastic from the previous pass before going further.

Then just repeat for the other holes.

Next, on the bandsaw, prepare the “fence” so it is 1 inch from the blade.

You can take the block of sheets from the template now. keeping the wrapper on the sheets means they’re all still held together so you should now be able to push the block through the bandsaw to cut an inch off. To do this, use the pushstick thing to push the block in. Don’t press the block in with your hands as you may influence the blade to veer off. Or you may slip and have to go to hospital, which is not a good idea right now.

Once that’s done, you can round the edges easily using the bandsaw.

Easy! I took my time with this earlier and was able to prepare 100 visors in only 3 minutes 40 seconds.

Frames

This one is tricky. We’re all using different frame types, so use whatever instructions make sense to you. I’m not going to walk through everything button by button (that’s why god invented search engines) but will explain the steps in general.

I’m using the “Europe” version of the Verkstan model, as that’s what I started with, and my “clients” are happy with them.

There are two versions of each Verkstan model – a “solid front” and the original “plate” version. We’re going to use the plate version.

Using the solid front makes sense if you’re using a dual-extruder, but most people are not.

First, create a “cookie cutter” outline of the frame. To do this, I open the STL in FreeCad and export as PDF. This creates a top-down image of the frame. I then open that in Gimp, delete the background and the grayed-out “plate” part of the front so I’m left with the frame outline.

Select the outline and “shrink” the outline by 1px and save that.

Now load that outline in Inkscape and convert it to an SVG.

Then convert the SVG into an STL using this tool. I think the size I used was about 180mm, and 0.6mm height (twice the height of your nozzle’s line-height). tick “reverse winding-order”, untick “base plate”, and then re-render before downloading the STL.

That STL has a lot of nodes in it, so I then use Meshlab to reduce it in size. I think I reduced mine down to about 2000 nodes from …who knows.

Next, open FreeCad and import the Verkstan model.

I turn the model over so the plate is at the top and there is a space under it. This way the printed plate will always have a nice finish on the “bottom”, and there won’t be a problem of the second frame in a stack’s plate drooping down and sticking to the first frame.

now import the cookie cutter outline and manouvre it into place on top of the frame. It will actually be better (for reasons you’ll find out) to just manouver the frame into place under the outline.

In my case, this just involves moving 60mm in Y.

Using the Parts menu, convert both STLs to solid. you can then delete the imported STLs.

Now go into the Draft menu, and create separate matrices from the frame and the outline (so you have 2 matrices).

Change X and Y multiple to 1 in both. Change Z multiples to 5. If you want, you can use a Spreadsheet object to make this dynamic. The Z distance should be 5.6mm (5mm for the height of the frame, 0.6mm for the height of the outline)

Okay – select both the frame and outline matrices, and export as STL. You can now print that.

After your first print you’ll notice some really bad “stringing” issues. I don’t have any left so can’t photograph them, but they look a little like this:

The reason this happens is that when the printer is laying down lines for the outer wall of the second frame, it’s pretty much drawing on empty space. So the line pulls away from the print and follows the path of the nozzle until the nozzle touches against the frame enough for the printed line to stick.

So now we get to the fine-tuning.

Print out a stack of 2 frames. Take note of where the stringing takes place.

Create a new FreeCad file and import the cookie cutter outline. Convert that to solid.

Now, wherever you noticed stringing take place, put a little line right across the outline from the inside to the outside, using a cube block that’s 0.6mm high, whatever length is required, and 0.2mm in width. I use 0.2mm because it’s a little less than half the width of my nozzle. we want this to be an easy-to-remove line, so don’t make it heavy.

Now you can save that all as an STL and import into the main model and repeat the creation of the Outline matrix (delete the old one).

Load up in Cura or whatever. If you look at the print path at those new lines, you’ll see that the printer skips from inner to outer, etc, which forces the line to reconnect with the main model, making it stronger.

The next print of the 2-stack should be better than the first.

Repeat until happy.

Then create a 20-stack and print.

Oh – almost forgot. There’s an art to separating the stack as well. Place the stack down on a cutting surface on its side so there are 20 end hooks touching the surface. Press a blade (I use a butchers knife) down between each end hook one at a time. Turn the stack over so the other 20 end hooks are touching the surface. Repeat. Now place the stack on its “back”, hooks in the air. You should be able to place the blade between the end hooks of two frames at a time, and just push down to separate them.

I was thinking recently that I don’t really do as much as other people. I’m always admiring what other people get up to, and wish I could be as productive as them.

But then I realised that the reason I feel like I don’t get much done is that I’m not comparing myself to one person. I’m comparing myself to many.

So, I’m going to try keep a semi-daily record of things that I actually get done (not counting my day job, of course!)

Starting with – I’m currently finishing off some pigeon-hole shelves for Bronwyn to use for her knitting. They’re way too large, but they’re also my first attempt. Maybe I’ll remake them in a year or so.

I’m also strongly considering building a new clavichord from plywood or OSB. The last one, I got to the stage that it could make a sound, but I didn’t complete it. This time, I have the tools and the work-space, so I should be able to get a bit further.