Yet Another Remote Winter Stove – Part 2 — Roger Caffin

[Roger Caffin]

Introduction

My design goals for the V3 stove were outlined in the previous installment or Part 1. Here we get into the technical details.

Key Design Problems

There are always design problems – but solving them is so much fun. Overall, the plan was to make a slightly smaller (and lighter) stove than the V2 with better pot supports, while keeping the existing hose and canister connector and the Vortex Burner principle. There were a number of important details like how long the heat exchanger section really needed to be and how to connect the hose so the gas flow went against needle taper, but there were two other areas of greater practical significance. The first concerned the air flow into the burner chamber: while it works fine on the V2, was it optimal or could it be improved? The second was how to design the wire legs / pot supports so the stove was easy to assemble but stable.

A preview of the stove

At this point things get very technical, but I have included lots of pictures to explain. This section is by way of a map of the evolution. Or maybe a blog of all my mistakes …

Incidentally, I do have some of the V3 stoves available for sale. I find going into actual ‘production mode’ is necessary to make a good one for myself. It also means I have spare parts. Feel free to ask.

Gas Flow Direction

Gas flow through control valve
I wanted the gas flow to be ‘against’ the needle, to help with cleaning. This one was easy to solve: I would have the jet on one side of the needle valve and the hose connection on the other side, where ‘valve’ means the actual constriction. A real cross-section view will be seen further on: this view is just diagrammatic. The Hose connection is to the left and the needle valve handle is to the right. Obviously, the ‘working’ part of the needle valves is mid-picture. I could even use the quite successful Hose Connector from the V2 stove for this. I would point out that this gas-flow arrangement (flowing against the needle according to the blue arrows) is what is used on almost every upright canister stove, although in many cases that will be for manufacturing convenience. The details will come later.

Air Flow into Burner Chamber
Caution: geek territory. This occupied a lot of my time in the later stages. I wanted to get the air flow and fuel/air mixture really ‘right’, which was difficult in the confined space available at the bottom of the burner chamber.

Fuel/air mixtures

In this context, ‘right’ means somewhere near the bottom of the curve shown here. Yes, it is possible to have ‘too much fuel’ in a mix – although that might better described as ‘not enough oxygen’. How you judge this in practice is hard to describe but is all about the shape and behaviour of the flames. It is not that hard to distinguish the flames from a proper vortex flow from a ‘wrong’ flow once you have seen some of each. You do get some funny flames when the jet is half-blocked and the fuel stream goes off to one side, and an example is shown under ‘Jet Size’ further on. So we launch into a myriad of technical possibilities.

Rim feed
On the V2 stove the air flow comes up through holes in the Stove Base Plate. I had discarded the idea of having holes in the wall of the burner chamber as they did not work very well. But I had also raised the bottom rim of the burner chamber above the Stove Base Plate by the thickness of a washer, to reduce the otherwise excessive heat flow from the titanium wall into the aluminium stove base plate. This massively reduced the contact area between wall and base plate. An intriguing idea popped up: could I increase the size of the gap under the burner chamber and dispense completely with the air holes in the base plate?

Washers between Stove Base Plate and Burner Chamber creating an air inlet.

I tested this idea by blocking the existing air holes in the base plate and varying the number of washers between the chamber and the base plate. Judgement was by looking at the flame shape. Too many washers (>4) and there was clearly too much air coming in; only one washer and there was definitely not enough air. One should note that with a thin single-washer gap (<0.5 mm), boundary effects at the surface of the metals effectively reduce the air flow to much smaller than expected. OK, perhaps we could fine tune the gap to get the right amount of air?

However, while the idea worked fine at high power with 3 washers and also worked OK at very low power with 3 washers, it was not so good at the commonly-used low-to-medium ‘cooking’ power. The flames went ‘soft’ and had yellow tips. That meant elevated CO production and dirty pots. Obviously, the vortex flow and air suction were not happening properly. This was not good. Eventually I decided that the problem was that at medium power, the slower gas jet was not sucking in enough air from under the chamber rim, given a gap suitable for high power. Ultimately, this meant that the aerodynamic coupling between the jet in the middle and the rim air inlet was too weak. I put this down to the distance from rim to jet being a bit too far.

I did test this a little further by having some smaller air holes in the base plate working with the gap under the chamber rim. That worked – a bit like the dual jets (Hi and Lo speed) on a small two-stroke engine. But if I am going to have air holes in the base plate anyhow, then I might as well forget about the added complications of holes in the rim. So I did.

Centre feed

At right a much larger hole around jet to act as an air inlet

The next idea I had was to replace the several air inlets which are around the Base Plate with one bigger central inlet: a large hole around the jet. This is shown on the right: the original base plate (and the V2 design) saw the jet boss completely block the centre hole (left). This would bring the air inlet very close to the jet for (hopefully) a better venturi effect. I could vary the diameter of the hole to fine-tune the fuel-air ratio. Well, this did work as far as the fuel/air balance went, but it had an unexpected consequence: a rather poor flame behaviour at most power levels.
Air flow from different inlets

Eventually I understood the problem. The way the jet and air went straight up the middle to hit the splash plate at the top (left hand drawing) meant that it was hard for the flame vortex to form. The flames were not swirling around inside the burner chamber as they should for a Vortex burner. It may also have meant that the mixing of fuel and air was not as good as I wanted, and that could lead to problems with CO emissions. On the other hand, when the air inlets were further out (right hand drawing) the air flow was across the bottom of the chamber and then up the middle. In practice this more curly flow made the formation of the vortex almost guaranteed.

A total change in overall design might well solve this problem, but I wanted to use the 38 mm Ti tubing I had. It had cost a lot, after all, and it looks sort of neat.

Overlapping Air Inlets

Engineering problems with the air inlets

I found some problems fitting the air inlet holes into the space between the central stove body block and the burner chamber rim. (Where the rim should be is shown by a black circle in the next photo below.) The problem was one of mechanical engineering: the web of aluminium between a screw hole or an air inlet and the burner chamber keyhole was too thin for safety (in my opinion) in at least one place, no matter how the holes were laid out. This is illustrated by the faintly pink area and arrow just SE of the SE screw head in this photo. The metal was even narrower in some cases.

Since the Stove Base Plate is on standoffs, above the central block (see black square in the photo below), perhaps the air inlets could creep into the area covered by the central block? In this case the air flow inside the burner chamber would be angled inwards towards the jet a bit more. This was done to the stove base plate shown above, and is illustrated by the yellow area and arrow N of the brass jet.

It was a nice idea, but unfortunately the vortex flame showed a similar problem to that due to the big centre hole. The air inlets were blowing too close to the jet, and the air flow across the bottom of the chamber was just not quite good enough to get adequate mixing. A proper vortex was not being created inside the burner chamber. It seemed that the balance between too close to the middle and too far out was a little more sensitive than I had expected. At this scale you cannot predict such things.

Optimal Solution for Air Inlets

Eventually I really LOOKED at what my engineering drawings were trying to show me and realised where the problem actually was. First of all, the inside edge of the burner chamber rim on the units being used for testing had not been machined as the drawing required. There was a ragged inside edge due to the way the rim had been formed in the hydraulic press on this particular unit. This ragged rim was bigger than it should be and was covering too much of the inside area of the stove base plate. In other words, the real edge was intruding way beyond the black circle. That was easily cleaned up on the lathe, and it gave me an extra 1.5 – 2.0 mm space all around. To repeat: the correct rim ID is shown by the black circle. (Well, actually, the black circle is just inside the proper rim: it was drawn with a fine pen.) An extra 2 mm may not sound like a lot, but in that small space it was.

The effect of 3 holes vs 4 holes

Perhaps more importantly, I could see a fundamental conflict between the square nature of the central block of the stove body with its 4 screws and the triangular or 3-point connection used on the Ti burner wall. The triangular burner connection was fighting with the square layout of the screws holding the stove body to the stove base plate. This meant that there were some serious weak points between some holes in the plate. These are illustrated by the pink areas and pink arrows on the plate to the left. These are worse than in the previous photo: things get tweaked during development.

Inspiration struck. I changed the burner chamber connection from 3 screws to 4 screws and put them near the corners of the square, as shown above on the right. Instantly I had enough space for the air holes clear of the burner chamber rim (black circle) and better mechanical strength as well. And the air flow worked fine after some tuning of the total air hole cross-section as well. Problem solved, so I decided to move on to the next problem.

Needle Valves

The goal was for very long needle valve shaft on the V2 to be cut right down in length, and also the tip was to be redesigned to make the control range much larger. Reducing the length of the shaft was simple, but redesigning the tip took a bit more work. First I had to understand the problem: just what was really happening down inside the stove body?

Real needle, virtual seat

In theory (whose, I am not sure), the tip on a needle valve has a distinct shape like this: a fast taper down to a parallel or almost parallel tip. The seat into which this goes should match this profile (more or less): the red lines show an extremely crude representation of what the seat should look like. When the inside corner on the needle sockets into the outside corner in the seat, the valve seals off. Getting this seal right can however be tricky. If you don’t, the stove will not shut off. I had experienced this problem a few times during my stove development saga.

I can machine the tip on the needle valve to the profile shown: that is not that hard with a very fine tip on the lathe tool with a CNC. But boring out the valve seat down at the bottom of a potentially deep hole is another matter. The main requirement is that the small hole into which the tip goes must be absolutely concentric with the surroundings – otherwise the edge of the hole will be ’tilted’ and the needle will not seal against it. A secondary requirement is that the tip of the needle and the hole into which it goes must be ‘small’. In practice this is best done with a specially shaped drill bit which cuts both surfaces at once: that will guarantee concentricity.

Finding a suitable ‘specially shaped drill bit’ is the hard part: they are not specifically made for this task. I imagine some other stove manufacturers might have them custom made to their specifications. However, a lathe centre drill bit comes very close to this profile, and I could easily imagine a lot of stove manufacturers going down this route: its cheap and easy. However, such a centre drill bit has to be rather long to get way down inside the stove body. For the previous V2 stove I was using an ‘extra long’ centre drill which was over 100 mm long and hard to obtain. For this stove it could be a bit shorter, but still longer than a stock unit.

An added complication was that I wanted a 1.0 mm hole for the needle tip to go into. This is rather small: most centre drill bit are bigger (1.6 mm and larger), and long centre drills for this tiny size are very hard to find. In the end, to get what I wanted, I had to have some carbide centre drills made for me in China (CarbideChiu on eBay) to my dimensions. CarbideChiu has done several custom carbide jobs for me so far and they have all been good and reasonably priced.

I explained carefully the profile I wanted, but I did not expect to get it exactly. The specified profile (a controlled small radius at the corner) would require an almost exactly sharp-angled corner on the grinding wheel, and only brand new grinding wheels have this. Chinese production grinders – well, unlikely. But I thought I could manage even if the corner was a little bit rounded.

Cross section of actual needle valve seat

In the end, the carbide drill bits had a somewhat bigger radius of curvature at the corner than I had anticipated. The result of one drilling operation is shown here, with the red asterisks showing the corner. (This is a machined cross-section and it has not been cleaned up.) However, to my surprise, I found that the seriously rounded valve seat corner gave a very nice smooth control and range of control of the flame. From shut off to high power takes at least 1/2 a turn, and the valve seals with light finger pressure. All planned to be this way of course … :)

Then I found out out that the performance of the valve seat is not just a matter of angles and dimensions. Early test units for the stove body were machined from some cast aluminium I had ‘in the corner’. It was a bit of a broken Hercules propeller (friend of a friend at an air force base). On the other hand, production units were to be machined from flat bar of 6060 T5 alloy, when it arrived. The centre drill used at first was a Dormer HSS unit with a sharp corner, not at all rounded as above, but it was a bit short. The needle itself was turned up from stock 2011 machining alloy: nice to machine but not super-hard.

Damage to needle at 1st corner

But things did not work as planned. It turned out that the cast aluminium alloy I was initially using was as hard as steel, and this mattered. (Well, it had been a propeller on a military transport plane after all.) Screwing the ‘soft’ needle into the ‘hard’ sharp-edged seat had deformed the tip as can be seen in the photo above, pointed to by the arrow. I soon noticed this as the valve characteristics changed after just a few operations. This was an ‘oops’ moment for a little while, until I understood the difference in hardness of the two alloys. However, once the flat bar stock arrived I found that the more rounded corner shown previously and the more closely matched alloys used for needle and body managed to bypass this problem, so all was well. The hardness of the cast aluminium alloy was a bit of a surprise though. Never mind: it will be useful somewhere else …

Jet Size

I went into jet sizes in the V2 articles. Common jet sizes on commercial stoves are around 0.28 – 0.30 mm. The bigger the jet the more gas you can get through it for any given pressure, and hence the more power, but you run into problems with air flow and mixing when you try to go too big. I decided to use 0.30 mm as that seemed to give the right amount of air flow inside the 38 mm Burner Chamber, but it is not super-critical.

Flames coming out of burner: partially blocked jet at left, clean jet at right

I am bringing this up here to illustrate a point. These two photos show the stove going at full power (valve fairly open). First look at the flames on the right, and see how they are coming up from the burner. Compare that with the flames at the left: they are not coming up very far. What’s more, it took longer than expected for the stove on the left to boil the kettle. What is going on here?

The difference between the two cases is simple. The jet on the left is partly blocked. Less fuel is coming out of the smaller hole, but because the needle valve is more open (to get more power), the stream of gas coming out of the jet is going faster, and that is sucking in more air. More air: shorter flame, but it is still giving less power. This is not theory: it is hard experimental test data. The lesson here is that if the flame is tilted to one side or is looking very short, the jet may be dirty, and it should be cleaned. With a small lens and good lighting it was possible to see the lump of dirt on the inside of the jet. I detached it with a proper tool (a stove ‘pricker’). After that the stove went much better. This problem can happen with any inverted canister stove as the source is dirt and wax in the canister.

Leg Design

It’s a bit hard to explain how the leg design evolved. First, I got the idea that a bent leg or wire should come down from the pot, which the pot support is holding up in the air. Well, obviously. Then that wire should bend back up towards the stove in the middle, to hold it (the stove) up in the air. The bottom round corner would act as the stove leg. Apart from anything else, this approach gave me totally independent control over the separation between the base of the pot and the rim of the burner. I could hold the pot far enough up that the flame would not be quenched and giving off CO; how far up was an independent variable. I liked this idea.

A separate issue with pot supports is that some of them a too close to the burner in commercial designs, leading to two problems. The first problem is one which has been mentioned here at BPL a few times: when the Ti pot supports get red hot and the pot is heavy, you can find the pot supports bending under the load. Definitely an oops issue.

The second problem is that large pot supports located too close to the burner can quench the flame for a while, leading to CO emissions. If you are cooking inside your tent in a howling storm, that might not be a good idea (although you always get a bit of air movement even inside a tent during a storm).

Clearly then, a good design would have the pot supports up a bit and away from the centre, so they don’t get too hot and don’t quench the flames. In addition, vertical wire legs located outside the flames are going to be fairly strong.

New Ti wire stove legs / pot supports, in holes in stove body with retaining clip

I wanted simpler legs which could make the stove ‘free-standing’. The design shown is fairly obvious. But how to attach the legs to the stove body? Well, I did not want the legs to wobble all over the place, and it seemed that poking the end of the wire leg into a drilled vertical hole would provide lots of constraint. Of course, the leg would still be free to rotate in that vertical hole, but it (the leg) would not fall over sideways. Six degrees of freedom had been reduced to just one or two. But if the legs rotated too much, the stove could still fall over.

A bend in the leg right at the mouth of the hole meant that the wire would come out sideways, and could be pinned between two solid ‘things’. That would stop the leg from rotating. Wire pegs inserted into the stove body were possible but they would likely be too close to the edge of the stove body for solidity. A couple of bumps and the holes could start to breach the wall. That’s poor design.

On the other hand, if the ‘pegs’ were actually part of the stove body, there would be no question of the their strength. All I had to do was to start with a solid lump in the right place and machine a slot through it for the 2.4 mm Ti wire I wanted to use. What width slot? I did look at making it 2.4 mm wide to really constrain the leg, but I could see a few problems in the field with a scrap of dirt jamming a leg in a hole. It’s never smart to create problems. Perhaps 2.5 mm wide? As I had 2.5 mm wide cutters for my CNC and could not at the time find 2.4 mm cutters, this seemed just fine. (For the technically inclined: you can buy 2.4 mm PCB cutters, but they won’t handle aluminium. I tried. The cutters chipped.)

Prototype of stove base made from plastic

I made up a plastic prototype (HDPE) for the leg system, and it was fine. I made up an aluminium prototype, and the legs fell out of the holes. Oops, but that happens with soft plastics and prototypes: you run into the second degree of freedom. So a retainer plate was needed to keep the legs in the holes, but I wanted to be able to remove the legs for packing. So a retainer plate with slots which could be rotated to lock the legs in place was needed. There was a bit of experimenting with the exact design details, but that worked out OK. Some 0.55 mm Ti sheet was strong enough, although 1 mm aluminium 7000-series alloy might also serve if hard enough.

Titanium retainer for legs

You will see that I show four legs rather than the conventional three which many stoves use. The advantage of using three legs is that the pot is always stable on three points; the disadvantage is that the stability of the pot is actually quite poor. A pot is easily tipped off a 3-point support. Try it and you will see. On the other hand, a 4-point support is a lot more stable, but you have to get the legs right or the pot will wobble. The obvious solution here is to use a proper wire-bending jig to get every leg almost exactly the same. Commercial wire-bending companies do this, and so could I. (Actually, I already had a suitable bending jig: it was just a matter of adjusting it to suit.) The less obvious solution is to note that the pot is heavy and the legs are wire: the legs are slightly compliant. Very small discrepancies (<1 mm) don’t matter too much.

Stove Body

The design of this is of course the core of the whole thing. Some people try to use a 3D CAD modelling system for their design work. I have several 3D CAD systems, but they are all a pain to use when you are in a creative mood. You have to spend so much time fussing over the technical details of how to mate tab A with slot B (correctly, rather than back to front) that you lose track of what you are doing. So most of the design stuff goes on in my head, and that is followed by simple 2D plan-and-elevation sketches (AutoSketch, brilliant) for the details. Exactly how the design is generated inside my head – I have no idea. Anyhow, the stove body is obviously a major part of the exercise, but I will skip over a lot of the fine tuning.

Cross-section through middle of stove body

Technical detail: while the photo suggests that the edges of the holes are a bit rough and wobbly, that is only an artifact of cutting the stove body in half. The bores are smooth, albeit with some rub marks. There is also some messy-looking machining fluid still on the internal bores. I didn’t clean it all up very well, did I?

Clearly I need a ‘central block’ (green square) to hold the jet and the needle valve seat and the four leg holes. I mentioned previously how having a long needle valve had led to some thermal instabilities, so obviously the needle valve needs to be as short as possible to avoid them. Equally obviously, given the direction of fuel flow, the path to the jet has to come out of the needle valve cavity. The way to have all this is to put the hose connector on the opposite side of the central block, in line with the needle valve. So on the right I have the needle valve with O-ring seal and thread; on the the left I have the hose connector. Now to satisfy the detailed constraints.

It was desirable that the hose connection should not be too short, so that I could have a little heat exchanger section inside it, as on V1 and V2 stoves. A (relatively) large bored hole with some aluminium welding wire inside it worked fine on V1, and the same thing works fine here. The aluminium rod heat exchanger fits where the red rectangle is to the left of the valve seat. It is just a bit of 2.4 mm aluminium welding wire, cut to size and with an angled end so it can’t block the flow.

Now for the top and bottom faces of the stove body.

Top and bottom of stove body

This shows the top side of the Stove Body on the left and the underside on the right. The underside shows the leg holes and the anti-pivot stoppers for the legs, and the central boss for the leg retainer. The leg retainer is shown in a previous photo of the legs. The top side shows the screw socket for the jet and four short standoffs for holding the Stove Body to the Stove Base Plate. The leg holes are visible from the top as well.

The hidden constraint here was the commercial availability of aluminium flat bar. Only a limited range of profiles is available in Australia. Fortunately it proved possible to fit all the bits into an available 40 mm x 16 mm profile in 6061 alloy. To be sure, I could have started with a much larger lump and machined it down, but that would add a lot of machining time to each stove body and create a lot of wasted swarf, all to little benefit if it was not needed.

To be continued

The original version of this technical section turned out to be far too long to be posted in one go, so the second half will be posted in the next installment. After that we will have some test results – actual measurements.

 

 

[James Marco]

Thanks, Roger!

[Jerry Adams]

I like the sequences of figuring things out

[Miles Spathelf]

Nice work Roger!

I’d be interested in purchasing a V3… I’m a bit of a stove junkie.

[Gunnar H]

An engineering cliffhanger… Looking forward to next part.

[Roger Caffin]

A Hollywood cliffhanger – where you know the hero will win …… :)
It works.

Cheers

[Author]

Posts