Vacuum kilns really are the peak technique for wood drying. They keep the colour as fresh as possible, due to a lack of oxygen, and they dry the wood both rapidly and evenly, creating a minimum of stress and warping. But! For anyone looking to actually build a vacuum kiln, DIY style, there really is hardly any good info out there on the interwebs.
So, writing a “quick” article to help anyone wanting to make a small vacuum wood kiln. My own use case was that I was picking up local timber from tree surgeons, and needed a way to dry it fast and to as high a quality as possible. I was interested both in the lack of internal stresses or cracking in the wood, and that rapid drying in an environment without oxygen could do a better job of keeping the natural colours of the wood. And then, apparently because I hate myself, I decided to try and innovate and use radiative heating.
More commonly – if anything about drying wood with a vacuum kiln can be called “common” – people use hot plates to heat the wood. But this seemed like a less than ideal solution to me, as the heating plated only have points of contact with the wood. On the flip side, plates have a maximum and consistent temperature, transfer will be good at the points of contact, and small convective heat transfer will bridge small gaps, so there are benefits to this technique. If you want to know more about the best way to use hot plates instead then check out this book - Vacuum Kiln Drying for Woodworkers: How to Build and Use a Vacuum Kiln for Drying Wood (amazon affiliate link), which is useful for providing basic info on best practices when building such a chamber. Also the author is a great guy, I’ve had some email contact with him about various finer technical points.
Other techniques would be to heat the outside of the chamber, and let convection take the heat to the wood, or use microwaves, or to cyclically blow hot air/steam through the chamber, between vacuum cycles. Of these, pure convection seemed slow and deeply sub-optimal for maintaining constant temperature throughout each piece of timber. Microwaves, lol, no. Apart from the need for shielding, there is a real danger of hotspots forming, and who wants to play with microwave generation inside of a hot, steamy chamber? Doesnt that require high voltage? Hard no. Cyclical heating from an external chamber, better, but why bother when you can directly heat the air inside the chamber. Enter nichrome wires....
So, I’m just going to give a bunch of brief tips on what I have discovered, apologies in advance for the block of text, but just putting out what I have right now. I'll make it pretty if anyone actually, you know, ever reads this. Hey leave a comment if you do! As of time of writing this, I’m testing my mk3, after mk2 dried a piece of cherry and then failed. Do not ask about mk1. We do not speak of mk1. But in other words, I'm still learning, and trying to add to the body of knowledge on the subject online - almost non-existent as it is.
OK – basic principle then is to string long nichrome wires and apply power through them, so they get hot and heat the wood. Heating is both direct radiative, and convective as the elements heat the air/moisture inside the chamber. How much of the latter, depends on the level of vacuum you are able to achieve, and whether you are running a cyclic schedule, i.e. separate cycles for heating the wood, and turning on the pump to suck out moisture.
So, design tips:
- Maintaining an equal temperature throughout the chamber. Unless you are not running cyclic and are able to quickly hit really high vacuum, there is going to be some gas in there, and that hot air is going to rise. In my mk2 chamber, I could clearly feel that the top of the chamber was way warmer than the bottom. My designs are significantly taller than they are wide, which will emphasise this issue. Also note that the gas is going to be mostly if not entirely vapour, which has higher thermal conductivity.
- Fan! Sounds obvious, but what fan? Will a standard axial fan work under vacuum? I figured it would be less effective, and went with a radial / centrifugal design. But next, can you just buy, say, a 5015 fan and throw it in there? My initial guess was, probably not, with the electrical circuitry stuck in hot vapour, and condensate forming on surfaces etc. But as it turns out, if the fan is just running on 12V you can pretty much run it underwater. After a huge amount of time wasted on a magnetic coupling system (below), I tried just adding two cheap 5015. One of them has since died, but it didnt sound happy to begin with. The other runs, but seems a bit confused - I suspect the speed control was not specced to be running at 5% atmospheric pressure! I have since bought two Sunon maglev fans with tach outputs, so I'll be able to monitor for fan failure. Not yet installed.
- Magnetic coupling system..... I would not advise this, as 5015 fans seem to work, but I'm going to leave this info here anyway in case anyone wants to try. Basically, means you have a couple of magnets rotating outside the chamber, each one set at a slight angle so they drag a pair of magnets on the inside, which are themselves set into a fan impeller. Complications include….
- I 3d printed an impeller, first in ASA, but that warped and was soon a very sad looking impeller. Fortunately PLA, when annealed, is actually pretty good at temperature. Annealing can create warping however, so to reduce this I printed it up, added the magnets and bearings (annealing causes shrinkage), taped off the bearings, and then set it into a bowl filled with sand. Add a flat plate on top of the sand with some weights on it, to compress and minimise warping, and then baked it at 75C for perhaps 12h. Note with annealing, and I tested this, longer at lower temperature has the same effect as quicker at higher temperature, such as 100C for an hour. However, when doing the process in sand, you need to increase the temps only very slowly, to ensure the same temperature throughout. Low and long seemed safest to me.
- Neodymium magnets are generally not great with high temperatures, so look specifically for high temp rated. I used two D20x10mm “N38SH” grade neodymium magnets for the impeller.
- My first design had a 6mm aluminium plate as the top wall of the chamber, with a singe axle going through, the drive magnets one side and the fan impeller underneath. I was really scratching my head for a while wondering why the fan was turning so damn slowly, and the motor seemed to be working so hard. And then, finally, something triggered in the back of my head, perhaps from school physics classes, something about induced braking? And so I discovered eddy current braking – great. Holding a magnet up to a 6mm aluminium sheet and moving it fast, I could easily feel the drag. I tried replacing with 6mm polycarbonate, but it flexed quite a lot under vacuum - it was at this point I gave up on magnetic coupling and installed two 5015 fans isntead.
- Note that high strength rotating magnets can interfere with thermocouples....
- Another option, at least to help the air mix, is to separate the heating elements into two circuits, one circuit for each side of the chamber. That way you can intermittently heat only on one side, and give yourself a little bit of a circular convective flow. Switch between sides, with a minute or two for each, to keep it even
- Moisture is going to condense on surfaces, and at the bottom of the chamber, unless perhaps you can pull enough vacuum that water boils at room temperature. So for instance, in my mk2 I used thin aluminium sheet epoxied to MDF as the lid, I could release the pressure, quickly lift the lid, and see it covered in water beads. Water also pooled at the bottom of the chamber, so a water drain at the bottom is a must. Best way is to simply pull vacuum to the pump from here, then any water will be transferred to the condensation trap.
- Condensation traps can easily be made from an old bottle of fizzy wine, I like prosecco! I then took a spare wine vacuum stopper (look it up), drilled a few holes and epoxied some tubing into them, with the inlet tube going half way down into the bottle. Easy, done. Normal wine bottles would probably be fine, but fizzy bottles are reinforced for pressure.
- Avoid metals that can corrode, especially if there is a chance of the metal sitting in pools of water. My mk2 design had wires running up from the bottom of the chamber, then current returning via a different wire to the base. So the full 55v of power, applied between wires roughly 4cm (under 2”) apart, with small copper wires at the base connecting to the nichrome elements. With water pooling around them. Kinda similar to electroplating setups! Little surprise when all the copper connecting wires of a single polarity broke and failed after a day….. my mk3 design uses 1mm stainless steel wires for connectors, power applied from the top of the chamber, with all of the wires totally separated from base of the chamber.
- If you want to use MDF then this "can" work.... MDF it is easy to mill and thermally stable, but bear in mind it is VERY porous. Seal with epoxy once, and then at least second, ideally a third time until there is a visible film over all the MDF. Ensure any drill holes for wires are properly sealed. Dont try to speed things up and just do the internal surfaces - vacuum is sneaky, and will find ways in, and internal surfaces are much easier to blister outwards than pulling air in through external surfaces. Again, make sure every surface is fully sealed, there is no such thing as overkill here. Preferably use the same epoxy each time for maximum adhesion - I used thin infusion epoxy for the job, and double glove with talc in between because that stuff is EVIL (I'm sensitised to it, just brief contact causes initial itching, then swelling, then tiny blisters form, finally after a day or two the skin dyes and peels off - not fun). However, in retrospect high temperature epoxy may have been a better idea - check the glass transition temperature (Tg) of potential epoxies before starting out (there is a wide array of different epoxy resins, they are not all the same!).
- Polyester resin might work, if you are careful, but the glass transition temperatures tend to be low, and it really doesnt work well as an adhesive.
- Nichrome heating elements are great but thermal expansion due to heating is significant, so it is essential to build in ways to keep the wires under tension. Easiest is to connect one side using very lightweight tension springs, I am currently using 0.15N/mm. You do not need significant force, and softer means you can more easily build in greater stretch without using long spring sections. A rough estimate is 1% expansion for every 700C rise in temp (about 1260F). My chamber has wire runs around 45cm, so that is an expansion 4.5mm for 700C, or of 0.64mm per 100C.
- When selecting what diameter and resistivity of nichrome wire to use, bear in mind that higher resistance wire is very thin and flimsy, and accidents happen. Therefore, having as many wires in series as possible may be preferable to running in parallel, as it will allow for selection of larger and more robust wires. This does of course mean that should any wire fail, the entire run will fail - but I dont consider that to be a bad thing, it will give you a good indication that something needs fixing. Better that than unbalanced heating of the wood.
- My chamber is 160mm external diameter, 60cm tall, and I have at the time of writing 16 nichrome wires, each running around 45cm, generating just 150W of heat. This works, and I can get the wood up to 100C, but is a little underpowered in terms of the time it takes to heat the wood. I'll upgrade to 250W when I have time.
- If in doubt, suggest designing for greater heating power than you expect to need, and use a PWM output on a mosfet or the likes to regulate actual power consumption. I use STP40NF10 mosfets, which are more than adequate power wise and can be controlled directly with the 5v out from my arduino nano (via a 220Ohm resistor and a 10K pulldown). Using the 500Hz PWM pins (10 and 11), switching time is not a problem. Flip side of giving more power than you need, is the potential for safety issues, so obviously a balance needs to be struck.
- Ensure the nichrome wires and springs are easy to access and service. I now have them all connecting to the lid assembly (3d design to follow), so the whole section lifts out. The wood hangs from the lid also, so the entire assembly slots into the chamber and seals.
- Silicone sealant for bathrooms etc makes for great vacuum seals, use it liberally but bear in mind that once silicone has been added to a surface, you wont have much luck with other adhesives. I had my CNC machine carve a thick groove in a sheet of MDF for the main vacuum seal, and after sealing the MDF with epoxy I added enough silicone so it was well raised above the surface. A flat plate pushed against it to create the vacuum seal. At first I used greased aluminium for the flat plate, but it still stuck badly to the silicone, next iteration I added cling film against a flat plate. If the silicone is thick, you don’t need to worry too much about whether the surface is perfectly flat – smooth yes, but not so much flat. The vacuum will pull any surfaces nice and tight against it, although you might need to physical push them surfaces together when you first switch the pump on. But if you do want a flat surface and, like me, are too cheap to pay for a proper engineers surface plate, then granite kitchen saver/chopping boards makes for a great alternative. Use fast setting silicone for the seal, and leave it for a day or two minimum – the air contact area is likely to be small, so you need to give it time for the setting reaction to propagate through the silicone. Don’t rush it, same with any silicone you add to the build. I tried to rush at one point, and soon regretted it!
- Alternatively just buy a thick silicone sheet. Readily available, my next design will use a 6mm sheet. Much easier.
- For the chamber itself, I am using 160mm underground drainage pipe, cheap and readily accessible (from toolstation here in the UK). 6" pipes seem to be easy to source in the US, a little smaller but close enough. Works great for vacuum, no problems, and I’ve taken the wood up to 100C without issue. But then I also have insulation on the inside…
- Add aluminium foil to the inside walls, and if you can add an extra thermal insulation layer between the foil and chamber walls then even better. Neoprene or silicone rubber are good options, remember vacuum conditions make standard insulators like closed cell foam or cork unsuitable. I bought 2mm neoprene, epoxied it in place, and then self adhesive thermal aluminium tape on top. Make sure you rough up the surfaces before gluing, and suggest cutting into strips, and using high temp epoxy or possibly silicone to stick it in place. I had issues with the rubber bubbling up, though not even sure how it happened.
- Make sure you have a fuse on the nichrome wires, in case two of them happen to fail and touch the aluminium…. a 5A fuse works for me.
- Pump – focus on diaphragm types, and look for a design that is happy with vapour condensates. Honestly most diaphragm pumps are probably good, but I have not tested this in any depth so cannot say for sure. Welch are a well known manufacturer in this field and from what I can see, their pumps are great - if you want to pay. My first vacuum pump was a cheap Chinese rotary vane pump that wasn’t designed for continuous use pumping at near atmosphere. And like most rotary vane pumps, it was not designed to pump water vapour either. So it quickly died on my mk1 when I was trying to find leaks. Avoid!!! I can tell you that I am now using a KNF VP 1HV pump, which in the specs states what it is designed for pumping high vapour loads, but you wont be able to find this model, so take a look on ebay etc to see what is available. There are lots of manufacturers and I only have theoretical guesses regarding what vacuum pressure is really required.
- My diaphragm pump isnt happy turning on when the chamber is under vacuum, it needs to be around 750mbar before it can start pumping. One option is to use a solenoid valve to let a little air in before switching on, or at the start of the heating phase. In this setup, relieving much of the vacuum before heating, isnt a bad thing anyway. You dont really want to run a heating cycle when under high vacuum, as moisture is still going to be getting pulled off - meaning heating will first dry the outer layers. Ideal cyclical heating would mean taking the entire piece of timber to a uniform temperature, and then switching on vacuum.
- You can buy insanely cheap diaphragm vacuum pumps from China at about £20 ($30) a pop, they run on 12V and claim 65mBar of vacuum. This is, of course, a total lie - my measurements show them struggling to reach 300mBar. However, the pump does make for a great fist stage. When I added it in front of my of higher quality pump, I was able to rapidly speed up the exacuation of the chamber. Not so noticeable in the first 500mar, but the time to reach 150mbar was halved. Better yet, maximum level of vacuum is also increased. And the pump seems to have no problem with vapour condensate (I suspect this is general for diaphragm pumps) It might be possible to buy a number of these pumps and run them in series to reach decent vacuum levels, but I have not tested this. Suggest not adding to two pumps very close to each other, and perhaps even adding an extra moisture trap between, so even out the vibrations in pressure draw. Finally, this also aids the above problem, since the cheap pump is in line I can turn it on first, and it creates enough pressure / flow on the outlet for my main pump to switch on, when it otherwise would have struggled.
- Cheap second hand diaphragm pumps can get debris on the inside, importantly in the valves, which will decrease performance. My own pump has specs that say it should reach 25mbar, when I bought it the best it could do was 120mbar, but after opening and cleaning just one of the two sides, it is now reaching 47mbar. Opening and cleaning was pretty easy.
- On that topic, I’m just going to present my thoughts on the “desired” level of vacuum in a slightly disorganised manner, so lets go. Consider that vacuum is a decent thermal insulator, although this effect is somewhat reduced as water vapour has a much higher thermal conductivity than air. And while the boiling point of water is obviously heavily affected by pressure, the energy required to make the water transition to gas, the enthalpy of vaporisation, is not heavily affected – in fact, the lower the pressure the more energy is require to transition from water into steam. The difference is not large, but it is there. So, what do we make of this? In essence, the way I read it is that the process has a degree of temperature independence, in that we are bringing the boiling point down, but we don’t need to be too strict about achieving super high vacuum. Consider, what does the thermal equilibrium equation looks like. Roughly the same amount of energy is going to be required to boil off the water, whether the boiling point is 30C or 60C. We are using radiative heat (hopefully), and there will be losses through the walls either via convective heat transfer or the aluminium foil on the inside not being a perfect reflector. These losses will be smaller if we don’t need to reach such high temperatures, but vacuum being a good insulator means it is not as big a deal – even less so if you have foil and insulation on the chamber walls. The temperature of the wood itself will, at least while it is still very wet, be limited to the boiling point, and rate of drying is then a function purely of the amount of heat energy you can transfer to it. Which is, obviously, energy input minus losses. If losses are small, does it matter then the precise temperature? So, perhaps it is not necessary to achieve the highest of high vacuums for this task? Also then bear in mind that wood has more elasticity at higher temperatures, which is a plus, but if your chamber is plastic you’ll want to be careful with the maximum temperature it reaches. However! This is my own guesswork, as there is unfortunately not much info out there. It would be useful to know how much of the heat from the nichrome wires is reaching the wood in a purely radiative manner, compared to being caught by the vapour in the chamber, but this is not something I can reasonably test.
- Running a cyclic kiln schedule seems to be the safe approach. Bring the wood up to a temperature, let it settle and even out, turn off the heating elements and switch on the vacuum pump until the wood cools to a lower threshold. Rinse and repeat, at least at the start of the process when the wood is wetter. Continuous heating may be safer, once the drying process has slowed down a little.
- One flaw of cyclic heating, is that when releasing vacuum, the obvious way is to pull outside air into the chamber. But this is relatively dry air, so when heat is applies, the outside of the wood will start to dry first, until the air is saturated with steam. This may create a degree of unequal drying, and potential issues with case hardening, internal stresses etc. Also, probably overkill but another thing to bear in mind, is that during normal commercial kiln operation, oxidation of the wood at high temperatures, is one of the mechanisms that causes colour changes. Therefore, to counter both these issues, is may be a good idea to vent the outlet from your pump (which will be in large part water vapour) into another chamber, perhaps even put a bowl of water in this chamber and heat it to create more steam in the air. Then when cycling and releasing the vacuum, do so by pulling gas in from this chamber.
- Simple 6mm outer diameter flexible tubing, such as those used in irrigation systems or aquariums, is fine for vacuum tubing - no need to buy anything expensive. Same with cheap tee connectors for joining or splitting. Clear pipes in my experience tend to be weaker, and can compress slightly flat, but dont close. The only pipes I had issues with were larger clear tubing, or thinner wall clear.