site  contact  subhomenews

Designing distiller basin-type proto3

December 27, 2019 — BarryK

This is intended to be the "final" design, for which I will create DIY plans. I have ideas about a collapsible design, however, think it better to stay with the latest basin-type that is working very well, with some tweaks.

The simple basin-type, prototype #2, was evaluated here: achieved efficiency 87% relative to the F-Cubed C1000, and 47% absolute efficiency.

Note: F-Cubed claim 55% for the C1000, but 65% for the C3000. The latter is the same width, but 3 metres long, and apparently the longer air rotation path improves the efficiency.

I am proposing that the DIY plans be basically the same as proto-2, except for:

  1. Thicker insulation in base, 60mm instead of 35mm
  2. Thicker side walls, 19mm instead of 12mm
  3. 3mm glass instead of 4mm
  4. 20 degrees glass angle, instead of 10 degrees

I did two tests of prototype #2, the second with the distiller sitting on extra insulating plywood and a towel, and got higher water output, 2.06 litres up from 1.86 litres.

The thinner glass is expected to improve efficiency. The higher angle of the glass will decrease efficiency, however, the distiller will work better in winter.

I will use a 572x672mm sheet of glass, 3mm thick, one of the pieces that was used in the sloping wick-type prototype #4.

I constructed a side-view in SolveSpace:

Note: SolveSpace hangs on my laptop. This is a known problem with intel video. In my case, the fix was to exit X to the commandline, and run "xorgwizard", and change from "sna" acceleration to "uxa" acceleration.

This is the side view, outside dimensions:


...the yellow-filled area is insulation, mostly 60mm thick, but tapering down to 18mm at the back (to keep the width to 235mm).

I managed to constrain the thickness so that a 235x19mm, 1.2m, piece of pine, will do the job for the sides. Although the glass short-side is 572mm, another 10mm is added on each end for the pine sides. It is proposed to use the router to indent the glass into the wood frame.

The right side of the above snapshot is messy, so here it is expanded:


...only 30mm between floor of basin and top-outside of the glass, but that is enough. The water depth is only going to be about 5-10mm.

The design looks reasonable, so a trip to Bunnings is coming soon... 

Tags: nomad

Further analysis of basin-type distiller test

December 16, 2019 — BarryK

I posted a couple of days ago, a showdown between three solar water distillers:

That was a very hot day, the ambient temperature peaking at 41 degrees C (106 Fahrenheit), with a slight breeze (11-23km/h, as reported by the Bureau of Meteorology for my suburb), and the distillers gave their best. My simple basin-type was measured to be 87% as efficient as the F-Cubed c1000 distiller.

So what is the absolute efficiency of my basin-type? Well, taking the claim by F-Cubed that the C1000 is 55%, that works out that the basin-type is 0.87*55, which is 47.8%. But, can I calculate it myself, from basic principles?

Interesting question. I looked at some research papers, and the formulae given are complicated, or confusing (for me anyway). However, I did find some web pages with basic explanations...

From basic principles, for a day test, we want the energy of the sun coming in, the water output, and a conversion constant which is the energy required to convert the water to distilled water. That constant is called "the latent heat of vaporization".

This is not my field. I was a solar energy researcher for awhile, in 1979-80, but that was in photovoltaics. Besides, I was an electronic engineer. Anyway, from online reading, this seems to be the appropriate formula:

Efficiency (%)

Water-out (kg) x latent-heat-of-vaporization (kJ/kg)

Energy-input (kJ)

x 100

Latent-heat-of-vaporization is a constant, 2260.

Distiller water output is shown as kilograms, but 1 kilogram equals 1 litre. My basin-type gave 2.06kg.

That leaves energy input, in kilojoules. This is the awkward one to calculate. The sun was pouring energy into the distiller, and I took hourly measurements of the instantaneous power input, in watts per metre-squared.

1 joule is 1 watt times 1 second. Joule is an energy figure, so if I break the day into 1 hour chunks, I can just multiply the watts hitting the distiller times 3600 seconds (1 hour), to calculate the energy.

The problem though, is that the sun is hitting the distiller at an angle, quite an acute angle early in the morning and late afternoon. I need to calculate the distiller surface area that is perpendicular to the sun, at each hour of the day. To help with that, there is a graph here:

This is mid-summer here in Perth, Australia, and the glass angle is 10 degrees, so at midday the sun is directly perpendicular to the glass -- so that one is easy, the area is just the area of the glass -- well, less the thickness of the wood frame it is sitting on.

There needs to be a start-time and end-time for the analysis. I took 8.00am as the start-time, as the distiller was mostly in shade earlier than that, and hadn't even started to warm up at 8.00am. I set the end-time as 5.00pm, as virtually no water was coming out after that.

With the help of the above link, I constructed this table:

Time (am,pm)
Irradiance (W/m2)
Perpendicular area (m2)
Power into distiller (W)
Energy into distiller (kJ)

...the energy is the power times seconds, divided by 1000, giving kilojoules.

The total energy input is the sum of the entries in the bottom row, which is 9867.6 kilojoules. Putting it into the formula:


2.06 x 2260


x 100


The basin-type distiller has an absolute efficiency of 47%

This agrees closely with the earlier estimate of 47.8% ...I kid you not, I did not fiddle with those figures. There is some approximation involved calculating the energy input, but seems to be pretty close to the actual energy input.

This is very pleasing. Various researchers have given different estimates of the maximum efficiency obtainable from a simple basin type, from 30% to 45% (and higher for a "non simple" design, such as ribbed-floor in basin, or a stepped-basin, up to around 60%), so I am at the top for the simple basin-type.

There were a couple papers that I read, where I could not agree with the logic of the efficiency calculations. For example, one paper claimed a seasonal efficiency variation from 14.8% to 31.1%, but it did not consider the angle of the sun, that is, the actual energy hitting the distiller. I thought that the meaning that they had given to the word "efficiency" was wrong, but I won't post links to those papers, as, as I have stated, this is not my field.

Anyway, this is my analysis, let me know if something is wrong! 

Further thoughts on tilted wick-type versus simple basin

I have already indicated that I am now favouring the simple basin-type, due to simple construction and easy-of-use.

There is another point to consider: the survival of pathogens, even the possibility of them breeding inside the distiller.

The simple basin-type exposes almost every part of the interior to UV radiation -- the exception would be just behind the distilled-water runoff at the front of the distiller -- however, I think that the white-painted walls would be reflecting UV into every shaded area. On the other hand, the tilted wick-type has a lot of places where pathogens can hide from UV light. In particular the F-Cubed design, where air rotates around the underside, but even a simpler conventional design would allow pathogens to hide in the wicking material.

Then there is internal temperature, the higher the better. Again, the basin-type wins. An internal temperature of 80 degC might not immediately kill some pathogens, but they are bound to be killed eventually. 

My recent tests highlighted the need to have sufficient water flow through the tilted wick-type, to avoid dry patches. This can result in a lot of waste water. With the C1000, There was about 15 litres in the container, and I had to refill it early in the afternoon. So, mid-summer, rough estimate is to have at least 25 litres in the container, and also mounted high enough, about 1.4 metres above the ground.

With the simple basin-type, there is no worry about water delivery rate. You just put in enough at start of day -- which could be governed by an overflow outlet inside the distiller. In my test, I put in 5 litres, and got 2 litres of distilled water. The left-over water then needs to be flushed out at end of day, to avoid build-up of sediments. Well, probably more importantly, left-over water is required so that there are no dry-spots on the basin floor, which would cause sediments to become baked-on to the floor.

Perhaps running some extra water through the distiller at end of day would help with flushing. The design could quite easily cater for this.

So perhaps comparing the two types of distiller from the water-maintenance side of things is a 50-50 situation. 

Well, no, I think that the basin-type wins, as in a campsite situation I don't really want to have to provide something 1.4 metres high on which to place the water container. The alternative is a small water pump, but that is getting too complicated. 

Tags: nomad

Solar distiller showdown take-2

December 13, 2019 — BarryK

Yesterday I tested three solar water distillers side-by-side:

There was a hiccup with the F-Cubed panel, the cloth becoming partly dry, so have repeated the test today. Friday, 13 December, 2019, in Perth, Australia.

Two changes from yesterday. This time I lowered the water container for the F-Cubed panel, down one rung on the ladder -- so the water container went from 1430mm above ground, down to 1120mm. Secondly, put a bit more insulation under the basin-type distiller -- previously had one 12mm MDF, this time added a 16mm plywood sheet and a towel:


The test proceeded same as yesterday, here are the results:

Top, degC
C1000, wick, basin

Official: 34degC, 23km/h
Official: 36degC, 23km/h

Dry patches in C1000
Official: 39degC, 13km/h
Official: 40degC, 11km/h
Official: 41degC, 14km/h
Official: 39degC, 26km/h

Stopped at 5.30pm.

About midday, dry patches were starting to appear in the cloth of the C1000, at the bottom. I moved the water container back up to the higher rung, and that fixed it, quite quickly. Lesson: better to have more water flowing through than not enough!

Notice how hot the basin-type got! One good thing -- that will kill any pathogens inside the distiller.

Here are the results:

Tally (litres)
Normalized (litres/m2)

Wow, stellar performance! So, the simple basin-type is 87% as efficient as the F-Cubed C1000 distiller. Pretty good. 

I am wondering if my wick-type distiller might have performed better with more water flow through it. Waste water output reduced to just about nothing in the afternoon, but I could not see if the cloth was drying out. Although I very much like the basin-type because of it's simplicity, both in construction and usage, I might do one more test with the wick-type, with the water container higher up ....maybe. 

Tags: nomad

Solar distiller showdown

December 12, 2019 — BarryK

Today, Thursday, December 12, 2019, in Perth, Western Australia, latitude 31 degrees in the Southern Hemisphere, it is mid-summer. The prediction for today is 40 deg C (104 Fahrenheit) and sunny. I decided it is time for a showdown, a "shootout" between the F-Cubed Carocell 1000, my tilted wicking-type prototype #4, and my simple basin-type prototype #2.

There were individual tests done on these:

In those tests, I published comparative efficiency figures, however, they were based on tests taken at different times of the year, different weather conditions, and also some figures published by the F-Cubed company.

The only way to accurately compare them is to test them together. Which I have done today.

I started putting them out in my backyard about 7.00am. A gusty breeze. They are in the shade of my garage, but that is only a temporary situation. This photo shows the shading:


...which does put the F-Cubed panel at a disadvantage, however, none of them were producing any output at this time, and the shade rapidly disappeared. By 8.00am, there was just a bit of shade in the bottom corner of the F-Cubed panel.

One good thing, the placement, and being mid-summer, avoids the afternoon shade from the patio, so I was able to run the test until late afternoon.

There was some haziness in the sky, which persisted for most of the day, though still got fairly high sun intensity readings -- but perhaps, the distilled water output might not be quite as high as if the sky was clear blue.

Here are the tabulated readings. The "Top" temperature measurements are taken with the IR meter, held about 3 inches above the glass, about 3/4 of the way up. Ambient readings were obtained by pointing the IR meter into the patio.

Top, degC
C1000, wick, basin
No water o/p yet
Official: 28degC, 23km/h
Official: 31degC, 21km/h
Official: 34degC, 26km/h
Problem with C1000 panel
C1000 almost recovered
Official: 38degC at 2.00pm

Official: 39degC, 16km/h
Official: 39degC, 18km/h

The test was stopped at 5.45pm.

There was a problem around midday with the C1000 panel. I had reduced the water inlet flow, by turning the tap from the water container to nearly off. This was because I thought too much water was flowing through the still. I probably should have just lowered the water container to reduce the head. Anyway, at 12.30pm discovered that most of the wicking cloth had gone dry.

This is a bad situation, as it can take awhile for the cloth to wet right through again. When dry, water tends to run over the surface, instead of seeping into the cloth. Anyway, it recovered fairly quickly, and the cloth looked Ok by about 1.00pm. It will be expected that this will reduce the distilled water output.

Notice also the sun irradiance (intensity) was only 750 watts per metre-squared. This was due to increased haziness in the sky around midday.

Here are the results, the accumulated distilled water output for the day:

Tally (litres)
Normalized (litres/m2)

There was only 5.32 litres from the C1000 panel, I would have expected about 6 litres. This test will have to be repeated, probably tomorrow.

The table shows the simple-basin-type giving almost as much output as the C1000, and even higher than the previous test of 1.85 litres (see previous test link at top of this page). That previous test was a hotter day, and higher sun intensity -- so how come I got an extra 0.01 litres today?

In the previous test of the basin-type, the distiller was sitting on a small picnic table, same today, but I placed a 12mm thick sheet of MDF on the table, so there was more insulation, which is known to be an important factor for efficiency.

Note, the distillers were not moved all day. They were oriented facing due North.

My tilted wicking-type distiller gave very poor performance. It is now eliminated. Though, will probably include it in tomorrow's showdown, since it is setup. 

Tags: nomad

Radical rethink of the basin-type distiller

December 05, 2019 — BarryK

I posted this morning some reflections on how the simple-basin prototype #2 achieved a high efficiency, with thoughts to enhance even further:

I have now got an idea how the frame can be collapsible, completely flat, for easy transportation. That is one new thought, that will be explained in a later blog post.

Another rethink is how to achieve a flat black floor for the basin. So far, have been spreading black silicone sealant, two 300gm tubes required for a 0.35 - 0.4 metre-cubed floor.

There are various problems with spreading silicone sealant. One is to find all pin-head holes. Another is to get the surface smooth. Then there is the messiness of the application.

A solution is a glass floor for the basin, with a thin black silicone sheet stuck on it. This would give perfect flatness, that will not warp, and suitable for very high temperatures.

The problem is to find a supplier of a large enough black silicone sheet. The exact dimensions yet to be determined, but somewhere around 700x550mm, 1mm thick. Last night I made some online enquiries to Australian suppliers, today only received one reply.

Reglin Rubber, based in Victoria, sent me a quote for a roll, 1200mm width, 10m long, sheet thickness 1mm. The quote is AU$148 including GST, and AU$40 postage. That would be enough for 28 distillers.

The biggest that I could find on eBay is 500x500mm, ditto Amazon. However, on Aliexpress I found this: 


...1m by 1m, 1mm thick, black silicone, AU$18.76 plus GST, and AU$27.05 for EMS ePacket postage (13 - 20 days delivery to my home, from China). Grand total AU$50.39. Rather a lot of money for a piece of silicone sheet, but want it, so ordered it.

EDIT 2019-12-09:
Yikes! I have received another reply. From Complete Rubber, based in Victoria:

Thinnest Black Silicone we can offer is 1.5mm Thick
1000 x 1200mm piece costs $499.00 + GST + Freight
We can supply 0.5mm, or 1.0mm Thick in RED or Translucent colour if acceptable
0.5mm - 1000 x 1200mm  Piece @ $441.00 + GST + Freight
1.0mm - 1000 x 1200mm Piece @ $385.00 + GST + Freight (Yes cheaper as is more common size purchased in the Market Place)

...a bit of a jump from $18.76 direct from China, and $148 for 10 metres from Reglin Rubber. 

Tags: nomad

Reflections on the basin-type proto-2 test

December 04, 2019 — BarryK

Yesterday was hot, ambient peaked at 41 degrees C (105.8 Fahrenheit), mostly clear blue sky, light breeze, mid-summer here in Perth Australia (Latitude 31 degrees), so the conditions were right to obtain maximum distilled water output that my basin-type prototype #2 is capable of. Here is the report of the test:

The panel produced 1.85 litres (0.49 gallons), with an effective glass area of 0.35 metre-squared, which calculates at about absolute efficiency of 48%, extraordinarily high for a simple basin-type distiller.

This blog post reflects on how that high efficiency was achieved.

There are three main factors:

  1. White painted inside walls
  2. Glass close to water surface
  3. Very low water depth

1. White painted inside walls
I posted about a research paper that obtained a 6.8% absolute efficiency gain by doing this:

In my case, I used white silicone sealant, Prosil 10.

2. Glass close to water surface
There are several research papers that have found the closer the glass to the water surface, the higher the efficiency. To achieve this, many designs have gone for a stepped basin-type still.

In my case, it is a compromise. The glass is about 15mm above the water at the front end, and about 115mm away at the back side. The distance of the glass at the back side is kept low by a very low glass angle of 10 degrees.

The downside of the 10 degrees is that for the latitude of Perth, 31 degrees, the efficiency of the distiller is going to drop right off in winter. The acute angle of the sun to the glass is going to mean most light will be reflected off the surface of the glass, instead of going into the distiller. I was thinking of using an external reflector in the winter.

3. Very low water depth
I only put 5 litres of water into the distiller, giving a water depth of 14mm average. However, my prototype #2 has a very uneven basin floor, due to misadventure experimenting with expanding foam, so the depth was variable, but there was water covering the entire surface.

There is one research paper that experimented taking the water depth down to 5mm and efficiency kept increasing as depth was reduced. This paper tested depths of 5, 10 and 20mm:

...they achieved output of 1141, 758, 305 millilitres in a day, from 5, 10 and 20mm water depth respectively. Quite an extraordinary variation.

So, if I had got the basin floor more flat, and setup the still to be very level, I could have put in less water and achieved higher efficiency.

Raising the efficiency further

Apart from lowering the water depth, is there anything else that can be done to further increase the efficiency of the simple basin-type distiller? Yes, there is...

I used 4mm thick window glass. Here are two more factors to increase efficiency:

  1. Thinner glass
  2. Low-iron glass

I have a collection of glass panes, that were cut for me by Casey at Perthglass, used in the various prototypes, and I just happened to have that 4mm piece available so used it. However, there is a research paper that has determined the efficiency improves as glass thickness is decreased.

This overview paper references another paper that shows efficiency improvement going from 4mm down to 3mm glass:

...that "other paper" is not free to download. Unfortunately, most academic research papers are seen as a business opportunity by companies such as Elsevier.

It was found that efficiency improved by 16.5% going from 6mm to 3mm thick glass.

I did locate another free-to-download paper that investigated glass thickness, can't locate it now.

Point-5 is low-iron glass. I have used ordinary window glass, that has some iron in it. This reduces the efficiency. However, low-iron glass is expensive. This paper reports that low-iron glass gave an increase of 6% efficiency compared with normal window glass:

I don't think that I will bother with obtaining that extra 6%. Will stay with normal cheap window glass, but certainly will go for 3mm, unless the panel is large and the extra strength of 4mm glass would be desirable. 

EDIT 2019-12-06:
I found some academic papers, free to download, that analyze glass thickness:

4mm, 4mm-two-layers, 4mm-two-layers-air-gap:

Comparing 4, 8, 12mm thickness:

Comparing 4, 5, 6mm thick:

3, 4, 5, 6mm low-iron:

...interesting, that last one determined that 4mm glass gave the highest output, not 3mm. They all obtained significant efficiency improvements going down to thinner glass, except for that last paper.

There is another consideration to improve efficiency: insulation thickness. My prototype has 35mm expanding foam under the basin floor, but tapering to 12mm at the back, and the wood side-walls are thin only 12mm. So, pretty obvious that this could be improved.

So, add this as item number 6:

  1. Thicker insulation

But how thick? This academic paper analyzes various thicknesses and different materials:
pdf d/l: depends on the material, but the ball-park figure they have got is 60mm thick is optimum, beyond which efficiency does not improve much. So, my prototype can be improved with thicker insulation. 

Tags: nomad

Testing basin-type solar distiller prototype-2

December 03, 2019 — BarryK

Today, Tuesday, December 3, 2019, Perth, Western Australia, it is mid-summer and the forecast is 39 degrees C, with scattered clouds. A good day to test the second prototype basin-type solar water distiller.

This distiller is pushing the limit of low angle for the glass, aiming for the narrowest possible design. The glass angle is 10 degrees and the depth of the distiller is 140mm -- well, kind of, it is constructed with 140x12mm pine on the sides.

Some posts on design and construction of the basin-type prototype #2:

So, 10 degree angle, white inside walls, how will it perform? The angle is probably going to be optimum in summer, and dismal in winter. Anyway, I set it up early this morning, around 7.30am. Initially in the shade of my garage. Orientation due North. Here it is, while still in shade:

img1 is sitting on a small camping table, but could have used any old cardboard box.

By 8.30am it was about 45% still shaded and hardly even warm.

8.45am: Sun just over entire panel. Sun intensity: 890 W/m2, ambient temperature 34 degrees C (yes, already hot!). Holding the IR meter about 3 inches above the glass, about 3/4 the way up, temperature is 32 degrees C, back side of the distiller is 32 degrees C. There is a slight breeze.

deg C
deg C
deg C



Official temp., via phone. Wind 24km/h
Only a very tiny amount of water out
Still not much water out
Some wispy clouds


Official temp., via phone, wind 13km/h
Lots of puffy clouds, ait mostly still or slight breeze

Moved the still slightly, facing sun at 1.00pm
Cloud cover increased, puffy clouds. Slight breeze


Official temp. Wind 24km/h
Clouds mostly gone
Blue sky, no clouds
No water coming out

Packed up at 5.45pm. Collected distilled water: 1.85 litres

The temperature readings of the glass are very high. This is obtained by pointing my IR meter at the glass, about 3 inches away, 3/4 of the way up. So it is not the inside temperature that is being measured, only an indication, and I think inside it would be even hotter.

So, my previous thinking that the inside of the distiller would not reach 80 degrees C was wrong, and I need to think carefully about what materials are used, that will retain integrity at that temperature.

The effective surface area of the glass is 0.652*0.542 which is 0.35 metre-squared.

How does this compare with the other distillers? I have not yet tested the F-Cubed C1000 distiller in mid-summer, only in winter, when I got 3.1 litres. So I will use the claimed mid-summer output, claimed by F-Cubed, which is 6 litres. That is a 1 metre-square panel.

Here is an earlier post discussing the claimed performance of the C1000 panel:

I am no longer going to compare by litres/hour, as the different designs work differently. The basin type is slow to get going in the morning, as it has to warm up, and internal stored heat may cause it to produce water a bit longer in the afternoon. This gives a different water output profile then the tilted wick-type, so it is simpler just to compare the total water produced in a day.

Another factor: with the sun oriented differently in the sky in mid-summer, the shade from my patio that came across the distiller late afternoon in the winter, can be avoided in mid-summer. meaning that today I got the sun on the distiller for more hours in the afternoon (albeit at an extreme angle).

Another note: I did move the distiller slightly, so as to avoid the afternoon shade from the patio, and also changed the orientation slightly, facing the sun at 1pm whereas before it was facing 12 noon. I probably should not have changed the orientation, even though only slight.

Normalizing the water output of my distiller, so as to compare with claimed output of the C1000:

1.85*(1/0.35) => 5.28 litres

That means my prototype #2 is 88% as efficient as the C1000 distiller!

I tested basin-type prototype #1 here:

Which produced 1.525 litres, though I did not leave it out quite so long in the afternoon, as it was not moved to avoid the afternoon shade. Prototype #1 has a larger piece of glass, 0.493 m2. Normalizing to compare with the F-Cubed:

1.525*(1/0.493) => 3.09 litres

Much less efficient. I wrote it down somewhere -- I think the glass angle was 35 degrees, rather high. Lots of side walls to loose heat. Interesting, the glass didn't get so hot.

Where to next? Enough prototypes! Basin-type prototype #2 has excellent efficiency, enough for me. The absolute efficiency of the C1000 is claimed to be 55%, so that means my prototype #2 is 0.88*55 => 48.4% -- that is very high for a basin-type.

I think that the "final design" needs a bigger glass angle, to perform better in winter, and I am thinking of 20 degrees. Also, the proportions of the glass can be changed, wider, less depth from front-to-back.

Tags: nomad

Using SolveSpace to design distiller

November 28, 2019 — BarryK

I wanted to explore the dimensions of a distiller constructed with different timber cross-sections. For the current basin-type prototype, I used 140x12mm dressed pine, and the distiller has a glass angle of 10 degrees.

For the next prototype, and what I intend to be the final design, I am thinking that a bigger glass angle would be better. Bunnings sell pine in profiles of 140x19, 184x19, 235x19 and 285x19mm. I would like to go for 19mm thick for the improved insulation.

Figuring out the dimensions and angles of the distiller is non-trivial, and there are many constraints. Fortunately, we have the superb SolveSpace application, that I wrote about recently:

I am not interested in the 3D aspect, just want a side view, to explore what angles can be obtained with different timber profiles. I have created a file in SolveSpace v2.3, in Easy Buster, and uploaded it:

Note that SolveSpace will be in the next release of Easy Buster, and ROX-Filer will recognise the mime-type, so users will be able to click on .slvs files to open them in SolveSpace.

Here is a snapshot of the file, exported to PNG and edited in mtPaint:


This is exploring using 235mm pine, which is the distance between A and B. CD is the bottom of the distiller, that will sit on a table. AE is the glass. Here are the constraints:

Lines AB and AE have to be at 90 degrees
Lines AE and ED have to be 90 degrees
The insulation thickness (yellow-filled area) is 35mm, except at the back where it tapers to 8mm
The length of ED is 63mm
The angle between AE and CD is 19.33 degrees
Lines AE and BC are parallel
Line CD is horizontal, as is top of insulation

With all of those constraints, the distance between A and B is 235mm. The beauty of this app is that any measurement can be changed and the result immediately seen. The glass measurement of 572mm for example, or the glass angle. 

A little detail about this design. With a glass angle of 19.33 degrees, and the angle of AB-AE being 90 degrees, it will mean that midday in midsummer in Perth, Australia, the AB back side of the distiller will be in shade, causing up to 40mm shade on the water. For most of the year, no problem. The reason for having the sides of the distiller at 90 degrees to the glass, is that is simplifies construction. Anyway, the shading can be avoided by putting some tapered insulation on the inside back wall.

EDIT 2019-11-29:
Playing around a bit more:


....for 235mm timber and vertical back side, can get the glass angle up to 23 degrees.  

Tags: nomad