Posts Tagged ‘Still building


Uh… Sort of forgot to say what I’m actually building…

It’s been gently brought to my attention that I haven’t really explained what the hell I’m up to. I kind of started writing that last post without considering the fact that not everyone has been rolling the design around in their head for the last year… I’m trying to create a very modular collection of parts that can assemble into a variety of different arrangements, but for now we’ll just look at the two extremes. The full reflux column and the pot still.

The column is going to look a little like this:

The full reflux column

Experienced distillers amongst you may be saying “um, wtf are you doing?”, so let me explain the design.

For a start, if I want to go commercial, I’ll need a good supply of neutral to experiment with, without having to spend a whole weekend to do a run to get it. So, an upgrade from 52mm (say a generous 2.5l/hr @aezotrope) to 103mm diameter column, which should do at least 8l/hr if I can get enough heat input to it. This will mean I spend less time making ethanol and more time crafting flavoured spirits.

A standard VM

Secondly, I want to try and build a new type of head on the still to see if some more control can be eeked out of the vapour management design. Currently, a VM relies on turbulence and ethanol density to get a vapour split between the overhead reflux condenser (which will return reflux back down the column) and the takeoff port, which leads to the product condenser. While this way of doing business has some serious advantages that I won’t go into right now (see this thread I wrote on home distiller to understand more), the big disadvantage to this is it’s very hard to run in low reflux operation – say for flavoured spirits. I want it to work differently.

Instead of  using a passive split, I want to force the vapour to split between two ports that resize in tandem – If I want to set reflux to 75%, the product should then be 50%. I also want to be able to completely shut down reflux and run as a tall columned pot still. The easy way to do this would be to have two valves. However, that would break a golden rule – never build a system that could allow the still to build pressure. It must always, always be open to the atmosphere.

So, I’m going to put a cross pipe through the column at the top with an aperture cut into it. A machined copper plate will slide along inside it to change the vapour split. It will have a teflon braid (safe with hot ethanol) seal around it. This should force a vapour split, and allow me to “dial in” a specific reflux ratio. This isn’t all my idea, I’ve had help from a lot of people, most notably Harry Jackson for the concept of a forced VM, and Airhill on for brainstorming ideas of how to actually accomplish it.

An illustration of my splitting valve concept

Note that these illustrations don’t actually show all the pipes – the plumbing of the cooling and reflux circuits is excluded.

The split will control the flow between two types of shell and tube condenser. The reflux condenser will be a 4″ / 1/2″ shell and tube, heavily inspired by the Jackson Crossflow condenser.

The crossflow condenser as it is traditionally configured. This is a type of shell and tube condenser. Image credits to Samohon on

By mounting this offset instead of overhead, I can form a pool of distillate, which can be drawn off back to the column as reflux, or via a needle valve, as product takeoff via liquid management. This will allow me to bleed the heads components without letting them contaminate my product condenser.

The product condenser will be a 63mm & 13mm  ‘shotgun’ type of shell and tube condenser.

This is a standard configuration for a "shotgun" type shell and tube condenser. Image credits to Samohon on

If you look at the illustration of my design, you’ll see that the shotgun condenser is designed to be removed, and is also way, way overkill for a reflux product condenser. This is so that it can be reused as the condenser for the pot still.

The reason I’m getting away from liebigs and coils for cooling is simple – backpressure. I recirculate all my cooling water via a pump and a radiator, and the pump really struggles with the coils. shell and tube condensers by nature have a very wide water path, so you can get very good flow rates on a shitty pump.

The pot still head is going to be more simple, I’m not even going to bother drawing it up. a 90 bend at the top of the column, then a triclamp, then a 45, reduced into the shotgun condenser. The triclamp will allow me to set any output height, which is a huge quality of life improver when running a still – remember you can’t use a plastic funnel or anything like that to direct output, not ethanol safe. Here’s a mock up:

Just some parts thrown on the table to illustrate the pot still shape.

Initially, I’ll just be using a keg boiler. Longer term I think I’d like a small hot water cylinder. They’re all copper here, and about a 100-150L boiler with pre-installed elements and drains would be very convenient I think. Either that or I’ll look into getting a stainless boiler fabricated for me, or salvaged.

Hope that clears up what I’m up to.



A still is born. Well… Conceived?

Who would trust me with one of these?

I’m finally getting started on my new still.  It’s going to be a modular system that can be reconfigured into various reflux column heights, and the humble pot still. It’s going to be 103mm, 63mm, and 13mm copper pipe, and stainless triclamp ferrules.  No Copper fittings if I can avoid it, although I might end up getting a couple of 4″ endcaps imported from aussie. Why not use copper fittings, I hear you ask? Because in NZ, buying copper fittings is like pulling down your pants and grabbing your ankles. You’re going to get reamed. Even at trade price, 103mm fittings are well over 100 dollars, closer to 200-300 for some of em.

My stash of triclamp unions.

So, this build has to be adaptable, modular, and made entirely from ferrules (I came across a supply of very, very cheap triclamp unions in a fair game of chance) and pipe. How hard can it be right? I’ve never soldered much, especially not with an oxy-acetylene torch, but as an engineer mate of mine said the other day “Yeah bro, people stupider than you use lathes and gas welders every day”.  With that sentiment in mind, I set out on some deep end learning (the best sort of learning).

First challenge – Get the copper pipe to mate nicely and strongly with the stainless triclamp ferrules. The pipe ID (~102mm) is approximately the same as the ferrule ID, but the ferrule has an extra 1-2mm of wall thickness, so the OD is more like 107mm, compared to the pipes 104-105mm. Two clear options – expand the copper over the pipe, or somehow get it fitting on the inside of the ferrule.

Second, related challenge – Make sure that when the ferrules are brazed to the copper, they will be square – this is important, as the even distribution of downwards flowing reflux in a reflux column is a big help for column efficiency, and crooked ferrules lead to leaning columns.

Third challenge – Make bends without buying fittings. This one seems pretty easy. Cut the pipe at precisely 90-desired angle)/2, rotate the two halves 180, then carefully braze them together. Then braze on my triclamp ferrules to each end without completely fucking the joint I’ve already made.

So, Leaving the whole “not paying bullshit prices for fittings” thing for now, I started to sort out that ferrule mating and squaring problem. I actually had a bit of a false start on this one, and built myself a very, very homemade sort of mandrel expander, 4 segments of steel that could sit around a lathe chuck, which you could then wind out to expand the pipe. This did leave challenge three as a complete nightmare though – how could I seat that square for brazing? I didn’t come up with anything, so after wrecking one end of a column section and completely failing, I decided to look at a new method – get the copper inside the ferrule.

I couldn’t think of an easy way to go about contracting the pipe. I calculated the relative coefficients of expansion with heat, and the distance I needed to move them was greater than I could do by freezing the copper and putting the stainless in the oven. The stainless is way, way too hard to think about expanding, even if I could do it even enough to still seal. So, I had to turn to the lathe.

Lathes are pretty cool, especially ones like this with a good autofeed and digital readout.

I ended up turning a wider ID for the ferrule, about 8mm deep into it. I figure that 8mm is enough surface for a good lap joint, plenty of space for the solder to wick down and make it strong. Stainless, especially in grades  like this (316L), is a tricky metal to machine. It’s very hard, kind of stringy, and needs a lot of care when working. Bear in mind when taking anything from the following process that I’m a rank novice at fabricating.

A ferrule during machining

When I set the flange in the chuck, I clean both surfaces, tighten it slightly, and then get a rubber mallet and gently tap all around to make sure it’s sitting square. I set the lathe very slow, 112 RPM seems to be a nice spot for me. I tried 80 and it’s great too, but 112 gets it done quicker without too much extra heating. The auto feed speed is set to about 0.05 mm/rev (not sure on what the correct notation is for that unit? distance travelled per revolution of the workpiece). I’m only cutting about 0.5-0.7mm per pass, so about 5 passes required to turn out the ferrule to fit the pipe. As for cooling, I just stand there with a bottle of cutting fluid and slosh a little bit onto the tool and the workpiece every now and then, once or twice per pass is working really well for me. You can probably see a bit of the fluid pooled on the ferrule in the above photo, a pink liquid. I’ try to tweak the depth / speed / feed until I get a nice spiral swage coming off in long strings, and I pay attention to the colour of it – a pale straw colour is acceptable, this is the point I’m aiming for in terms of efficiency, although sometimes I’m going slower if I’m not paying attention. If you start getting purple, slow down, more fluid, give it a rest, etc. That should be a drama though, because the cutting fluid will be boiling constantly by that stage.

The results sit tight and snug, though the slight deformations of the pipe from cutting can mean that if you want them really snug you need to give the pipe a slight squeeze in the vice or something. After the first one, which I had to seat with a hydraulic press (overkill), most of mine have been fine to seat by hand after I’ve filed a leading edge onto the copper.

Filing a leading edge onto the copper to help it seat in the machined ferrule.

After filing, they came together nicely, square, and almost perfectly flush on the inside:

Nice and flush

Nice and square.

Also appealing about this method is the fact that copper expands faster than stainless when heated – that means these joints will get tighter, not looser, as they are brazed, or as the still is operated.

So, with the ferrule joints no longer a problem, I needed to get started on making some copper bends out of my pipe. Off the shelf bends are normally die-cast – this is way beyond my resources.

A 90 bend being trimmed down to size in the drop bandsaw.

The fittings I’ve needed so far have been pretty easily created by using a drop bandsaw running nice and slow to cut for a brazed miter joint. I could cut at 45 and end up with a 90 bend like pictured being worked on in the drop bandsaw above, or at 22.5 and end up with a piece like below. I actually made that 90 on top of a long section to be the column on my pot still before realising that I’m wasting a potential 30cm of column height on the reflux still by not putting a union in below it – now I’ve cut it I can take that section and pack it. Modular!

A home made 45 bend with a ferrule seated on one end

I still haven’t decided if I’m going to make 4″ endcaps for my shell and tube condenser (reflux condenser for the column), or get someone to buy them for me in aussie and send them over. It seems that their prices probably make it worth it compared to the time cost of more complicated fitting fabrication than simple bends.

As for the actual brazing, I’m still a novice with the torch, but I think the after some TLC with a file and emery cloth, my joints aren’t going to be too ugly. First of all I lightly sand and flux both surfaces, then press fit, then reapply a bit more flux to get it sitting on that lip of the stainless:

well fluxed and seated ferrule

Then I get out the oxy torch and set the flame pretty neutral, maybe a hair on the carburizing side. I heat only the copper side of the joint, and maybe 5cm up from the joint. Copper transfers heat so well that you’ve got to work like this to avoid buggering the stainless. Then it’s just a matter of getting the workpiece up to a flow temperature for my solder rods (silfos 15, got a kilo or two of it for free), and soldering the joint. I try to keep the torch moving across the work so that it doesn’t overheat a specific area too much.


I’m going to experiment with different ways to “lay” the solder onto the joint, and see if that gets me neater. here is a snap of a joint that I’ve quickly hit with a wire brush. More cleanup to come.

Hot off the press

For now, at least things are finally happening on this build, I’ve been talking about it for almost a year now, it’s good to put solder to copper.

I’ll be doing some more brazing shortly, and then working on my shell and tube condensers. That will probably be my next post. When this is all done, I’ve got a lot of scotch I need to run! I’m filming the whole build, so when I’m done I’ll stitch it up into something entertaining.