The Allen Hydrautomat, a Working Paint Tin Model.




About: general bloke type of tinkering

A silent system of raising water patented in 1926 by Thomas Gaskell Allen, was alleged to have an efficiency of 80% of the pressure water used, a more efficient device than the hydraulic ram pump at the cost of greater parts and complexity.

Sadly, this wonderful device seems to have faded into obscurity over the years and so I have attempted to recreate a working model with the original principles involved.

Step 1: A Brief Working

Referencing the patent diagram, the system used the pressure of the inrush of water into a closed chamber called the operating chamber A, to pump water in the closed chambers C and Cx up to the next level.

Then the outflow of water in chamber A creates a partial vacuum which is used to draw up water from the open chambers G and D into the now empty C and Cx.

A is connected to the closed chambers C and Cx via an air duct as well as to open chamber G via the water inlet to A.

Prior to the working of the pump, closed chambers C and Cx should be manually filled with water as a requirement for initial operation.

The original device used different lengths of syphon tubing to prevent water from being sucked down from an upper level, but the method of operation escapes me and so I have improvised and installed a check valve on the tops of the inlet and outlet tubes as my primary method of avoiding reverse flow.

The 2nd diagram is my view of the sequence of events, bear in mind that the water source will keep the inlet tin on level 2 constantly full.

  1. The initial stage, new device all tins/ containers are empty, all open tins to be filled is the starting requirement.
  2. The first pumping cycle, default inlet valve position is open with foot valve closed, operating chamber begins filling pressurizing the empty closed tins which should result in bubbling in the open tin on level 4.
  3. The foot valve is triggered with the linked inlet valve closing, creating a vacuum in the operating chamber and the first suction cycle draws up water from the open tins into the closed tins.
  4. The second pumping cycle occurs once the trigger mechanism empties and resets, closing the foot valve and opening the linked inlet valve, causing the operating chamber to fill, once again pressurizing the closed tins. Now the first water output should be seen from the upper pumping stage.
  5. The foot valve opens starting the suction cycle once again.

Step 2: The Frame

The frame was made out of 17mm bamboo uprights and 4mm bamboo skewers for the platforms. My hypothetical design head of water is 400mm and the operating chamber is placed midway between the upper and lower levels.

  • The 1st level is the operating chamber at 200mm.
  • The 2nd level is 200mm above 1, being for the 1st open chamber which feeds the operating chamber and the 1st closed pump chamber C.
  • The 3rd level is 160mm above 2 and holds closed pump chamber 1.
  • The 4th level is at another 160mm and holds an intermediate open chamber.
  • The 5th level is at another 160mm and holds closed pump chamber 2.

The number of pumping levels is unlimited but must follow the 2 design rules, namely that the operating chamber volume cannot be less than the combined closed chamber volumes and that each pumping stage height cannot exceed the working head height, which in my version is 200mm.

I decided to design my model conservatively with a pumping stage height of 160mm which is also 80% of the working head height.

The bamboo skewers were tacked together with hot melt glue and then all joints were lashed with thread soaked in CA glue.

Each level platform was then glued into the main uprights, again using hot melt glue.

Step 3: The Check Valves

Generally a build would have started with the available valve hardware and then designed the containers around them. I went about it differently out of necessity, I had the containers and matching tubing so it was a matter of designing check valves to suit.

A single flare was formed on the 3/8 " copper tube which acts as the valve seat.

  1. The first version check valves were made from a rubber screwdriver bit holder with a piece of copper wire bent to function as the guide. The copper wire retaining method made it easy to bend to one side so as to slide the pipe into the wire guide for soldering, but the rubber proved too hard to seal properly.
  2. For the 2nd version I added a soft silicone overlay in the form of a small section of a nipple from a baby’s bottle, but the silicone or latex had a tendency to catch on the seat rim and didnt always center properly.

  3. The final version was a sliding fit brass tube with a small flare to retain an O ring with a section of a brass earth pin from an electrical plug soldered into the top of the brass tube.

Step 4: The Operating Chamber / Release Valve

At the bottom on level 1 is the closed operating chamber, this is where the vacuum and pressure cycles are developed.

The outlet needs to be braced to provide a rigid seat for the swinging valve and also against the pressure of ejecting water. I have used 15mm dia pipe due to the greater force of the ejecting water as compared to that of the 3/8" tubing.

More on the trigger mechanism in the following steps.

Step 5: The Allen Trigger Mechanism

The trigger valve at the bottom of the operating container is the heart of the system, opening once the water inrush has reached the required height and at the same time closing the linked inlet valve in the open chamber above.

From the sole photo of the device certain educated guesses can be made, however the method of linking the trigger reservoir to the valve flap remains vague, keep in mind that the default position of the bottom valve is closed and the top inlet valve is open. It then follows that the trigger reservoir will descend as it fills with water from the trigger feed tube, and empty itself at the bottom of its stroke with the valve opening in the process.

The trigger tube can be seen to have an increase in diameter on the downward side, so logic would dictate that an open tube of a larger diameter than the trigger feed tube, rides freely up and down the trigger tube and releases its water load at the bottom of its travel via a hinged mechanism at its base.

An assumption is that the odd U shaped outlet before the valve flap was required to obtain clearance for the trigger mechanism.

The trigger water inlet tube peaks just under the water level of the open chamber above which will allow the water level in the operating chamber to actually rise up the air conducting pipe, thus ensuring greater possible vacuum and subsequent pressure cycles.

Step 6: My Plan a Trigger Mechanism

Plan A, although nice and shiny looking, didn't work, the reason being that the trigger reservoir was too heavy even while empty and the measly 12 grams of water inflow didnt weigh enough to tilt the valve open.

So its back to the drawing board for Plan B, a lighter PVC trigger pipe 20cm long that weighs under 40 grams when empty and carries a water load of 75grams.

Step 7: My Plan B Trigger Mechanism

Plan B has custom made end caps much lighter than commercial 25mm pvc stop ends, and it will be offset from the center line due to it being 20cm long.

First the fittings were soldered and then the copper pivot tube was wired to the end cap.

In use the valve didn't seal as well as it could have, due to the ferrule's hard rubber compound, so I added a softer overlay in the form of a finger cot.

I also made a plastic deflector shield in the form of a section of yoghurt tub, this triggers the opening mechanism much sooner as the pics indicate, producing a faster system vacuum rise.

Next I moved the trigger tube water entry to the back so as to be over the pivot point thereby reducing the swinging resistance of the previous method.

Finally the distance from the counter weight to the pivot was increased to 190mm over the previous version's 80mm, this coupled with the curved link to the rubber stopper creates a much greater seal force, in excess of 250grams.

Step 8: The Open Tin / Inlet Valve

This tin on the 2nd level feeds both the operating chamber below it as well as the 1st closed tin immediately above. It contains the inlet valve to regulate the operating chamber and a method of constant water supply.

  • The first version check valve rubber was made from one of the screwdriver bit holders, a length of copper wire forms both the means of lifting and a way to guide it onto its seat. The top of the rubber was plugged with silicone sealant.
  • Final version was a sliding fit brass tube with O ring seal.
  • The pivot was made from 2 aluminium strips riveted together and clipped onto the tin rim.
  • The 5mm dowel section was drilled for the paper clip hooks and axle.

Step 9: 1st Pumping Stage

This is the simple tin on level 3, the 1st closed tin which has the same body as the tin on the 5th level, just the lids differ.

Step 10: 2nd Open Tin

This was a tricky build because the lid for the below closed tin needs to be soldered to this input tube.

The inter tin copper tubes were offset from the center for obvious reasons( frame structure) which allowed them to be staggered on alternating levels to prevent inlet tubes from obstructing outlet tube valves.

The sequence of pics shows how the offset tube allowed tins to be rotated into place.

Naturally one could cut the copper tube and join with gas or other reinforced tubing once in place, at the risk of more leakage possibilities.

Step 11: 2nd Pumping and Final Stage

Details of the 2nd pumping stage on level 5, the air conduit and pump outlet soldered into the lid, and a water inlet from the open tin below, soldered into the can body.

The air conducting tube connects the closed tins of levels 3 and 5 to the operating chamber and were divided into 3 pieces for easier installation. An irrigation T piece was used with 8mm gas tubing to connect the 3 copper sections.

The plastic container on top is for diverting the output to a measuring jug when testing or a needy area of the lawn.

Step 12: Some Build Points

I found it easier to build from the top down after first making the check valves, mostly because I hadn’t worked out the operating chamber valve design at that stage, but it is easier to assemble the tower from the bottom up.

I studied the photo of the original device and copied the proportions of tin to pipe as closely as possible. I estimated that the tin diameter was approximately 11 to 12 times larger than the tubing diameter, hence my choice of 3/8" tubing throughout, baring the trigger feed tube, smaller was more economical and just meant the vacuum cycle would take longer to complete. Likewise the bottom outlet tube was increased to a slightly larger size for improved performance and my testing indicated 15mm would be more than adequate with 22mm too large.

Another problem, other than the tube kinking, showed up on the trigger tube flexi link to the PVC pipe, an air bleed hole was needed to allow the PVC tube to empty itself when it was in the down position. This leaks water as the PVC pipe starts filling which slows down the tube filling process. On the other hand it gives an indication of when the pipe is about to drop.

After much observing of the pvc tube and flexi link, I decided to adopt a slider tube system to minimise resistance. The slider tube yields a better air bleed system and isnt prone to tearing like a hole in the silicone tubing is.

Some figuring with moments about a beam equations yielded a theoretical counter weight of 45g which worked first time, yay for engineering science. That was after plenty calulating volumes of a cylinder and then actually measuring the PVC pipe filled with water. Turns out the theory was pretty close, 2g off, which is neither here nor there.

My main tool in addition to a hacksaw and pliers was the brake pipe flaring kit which also included a pipe cutter. It was during the project that I also bought a small tubing bender to make the work a bit neater.

Lastly, I found it easier to manage the bottom tin by soldering on the outlet tube to the down pipe elbow last of all.

Some Design Considerations.

  • The first rule of design is that the operating chamber volume cannot be less than the combined volume of the closed chambers.

  • The second rule of design is that each pumping stage height cannot exceed the initial working head height.

  • Although not exactly a design rule, but equally important, use rigid containers and piping where system vacuum and pressure are encountered. Plastic tends to flex too much, reducing the amount of working vacuum and pressure available.

Step 13: End Testing

Other than sealing problems in the first test, the 2nd test went fairly well, due to much fine tuning of the operating chamber and the linked supply valve over the kitchen sink. Initially there was a gurgling on the pressure stroke in the closed tins, meaning there was more pressure than water, which could also indicate a leaky valve on the inlet tubes to the closed tins.

After many mods on the trigger tube to extend the down time to increase system vacuum, it occurred to me that the pressure from the open tin would always be consistent as long as the water level was maintained near the top, but the vacuum in the vacuum cycle would diminish as the water in the operating chamber dropped down to a final equilibrium level.

In use 1.2 litres is flushed out the bottom valve in order to pump 225ml out of the top stage.

First video is a general top down view to capture all the goings on and the second one is a closer look at the operating chamber valve.



    • Sensors Contest

      Sensors Contest
    • Backyard Contest

      Backyard Contest
    • Colors of the Rainbow Contest

      Colors of the Rainbow Contest

    59 Discussions


    Reply 3 years ago

    Nope not in common use, I dont have a use for it either, I just built it as a "bump".(a bump as in keeping a forum thread alive, for an interesting device in the real world.)

    Hydraulic ram pumps seem to be the V2 upgrade for using water to pump water.


    5 years ago

    Applause to reviving "forgotten" patents, I will be building one, graciously taking notes from your redesign. I did notice a few key points in the patent that may add to the efficiency and "cost" debate. :) The inverted "u" discharge would create more siphon than a horizontal. And I think the first open container level would be at the waterline of the body of water your lifting from (adds to the pressure entering the inlet; the body will try to stay level), in his case a moving stream (inlet upstream = pressure, discharge down = suction) which adds more to total power. As to "cost", if I could fill a cistern by harnessing ambient energy, I wouldn't care how much of the source I cycled through the pump. Thank you again for posting.


    5 years ago on Introduction

    You mentioned the kinking of the brass tube being a problem, if you pack the tube full of sand (or even possibly rice or some other granular material) and then bend it I think it will retain a circular cross section. I haven't done it myself but have heard it works.

    So this pump is powered by atmospheric and water pressure differentials? What kind of volume would you say it can pump in an hour? Given a large enough reservoir at the top can it run indefinitely as a closed cycle? Or does it need a stream or flowing water to work? Lots of questions, sorry :)

    5 replies

    Reply 5 years ago on Introduction

    Two methods for bending tubes include pouring hot tar and then chilling the tube. After the bend is made the tar is heated to flow out of the tube. Ice with dish soap added is another method if you have some dry ice or a freezer large enough to freeze the mix in the tube. Also molten lead can be allowed to harden before the bend is made. I don't like the use of lead around water bearing pipes. Lead is frequently used to get the lovely bends on trumpets and other brass musical instruments.


    Reply 5 years ago on Introduction

    I didnt use any brass tube in the build, the copper tube on top bend was filled with sand blast glass beads (very fine) and plugged both ends. I suspected that the glass beads might be the problem and changed to sand. While the ripples were a lot less, there were still quite a few. This, in my opinion is because of the cheap bender I was using, its able to take both 3/8" and 10mm which means theres too much clearance on the 3/8" tube and hence it ripples.

    Approx volume would be 18L in an hour with 96L being wasted.

    I tried the closed circuit thing while testing in the kitchen sink, it loses 1.5L and the pumping only adds 200ml, so it definitely needs a constant supply.

    Hard Way;

    Bake the sand before you pack the pipe to remove moisture. Fill and pack the pipe using a tamper a few inches at a time. Solder caps to the ends with one cap drilled with a small hole on the end to relieve and steam pressure that may build should there be any moisture left in the sand. Using a propane or map plumbing torch, I prefer map it's hotter, clamp the pipe in a vise and heat it mostly on the outside of the bend when you see green in the flame it should bend relatively easily. Go slow. You can make very complicated bends this way.

    Or if you prefer...

    Easy Way;

    Go to a building materials store and look for soft copper in the plumbing section. In the tools section most likely right next to where the soft copper tubing/pipe is merchandised, you will find bending springs that go around the pipe to help form the bends without much distortion in the tube.

    If you want line straight pipe when you unroll it. First lay the roll on the ground like a wheel step on the end and unroll the length you need plus some extra. drill a hole through a 2x4 that the tubing will barely fit through and put one end of the tubing in it. with the 2x4 close to the vise clamp the tubing in the vise tightly and pull the 2x4 down the length of the tube. This should go a long way to getting the tube straight.


    Reply 5 years ago on Introduction

    frozen soapy water is what gets used for brass musical instruments like trumpets.
    same should work here.


    Reply 5 years ago on Introduction

    Thanks so much for pointing that out :-)

    Soapy water was left out of a "How its made" episode I saw last year.


    5 years ago on Introduction

    How has this been working?

    Can you tell me, how much water does it pump upwards, in an hour?

    4 replies

    Reply 5 years ago on Introduction

    from checking the video, it looks like 3 pumping cycles in 1m, so 3 X 225ml X 60 = approx 40 L pumped for 216 L flushed.


    Reply 5 years ago on Introduction

    I came to the same conclusion, but would like to know from the OP if it did pump that much water, or maybe even less?

    The pump isn't very efficient, so the question is, what needs to change, in order not to loose the flushed water everytime?


    Reply 5 years ago on Introduction

    I assume by OP you mean the author, thats me, I didnt run it for an hour, it would have been a tedious excercise with the watering can, hence calculating it from video from the tests I ran.

    This style pump is more efficient than the standard hydraulic ram.

    It seems you dont quite understand the working principle of these devices, it functions as a result of the water being flushed for the vacuum cycle, no vacuum cycle = no water available in the higher stages for pumping.


    Reply 5 years ago on Introduction

    With "OP" I meant, the "Original Post", really referring to the instructable itself :)

    I understand the working principle of "these devices", but just think it's quite unproductive. Surely it could be improved upon, somehow? I haven't build one yet so I don't have the knowledge and experience you have gained already but would love to build one when I have some spare cash to buy everything needed, since I don't have anything that's needed for this kind of project.

    From a different point of view, I guess if the flushed water is being flushed back into the water source, no water would be wasted and essentially you'd have a free water pump, but it would be nice to see it being more efficient.


    5 years ago on Introduction

    Google found this explanation with pictures:

    1 reply

    Reply 5 years ago on Introduction

    Yea, I tried wading through it but gave up. He's dissecting the patent drawing and not the actual photo, which means he's going to get thrown off the track as per the usual vague patent application's design, in which they say just enough to get the patent, but not enough for you to actually understand it.

    Google found another hydrautomat view in a 1922 Popular Science mag, which in my opinion was most likely a temp installation sited in the Upper Wandle River for the official demonstration at that time.

    On page 53 it can be seen that the dia of the trigger tube changes in size on the beginning of the straight downward section after the bend, yet nowhere is it discussed in the patent as being a sliding fit reservoir of water to trigger the opening of the bottom valve.
    The most detailed description that I've seen is from The Engineer16th June 1922, a journal of that era, was...
    by means of some such mechanism as that indicated by V, closes the valve W.

    I've seen more meat on a used dog's bone. :)

    You'll notice Arto's very impressive graphical disection barely touches on the bottom valve much less the triggering method, also it can be seen from his inlet valve diagrams that his sole basis of info appears to be the patent drawing.
    If you compare my simple inlet valve in the open chamber with the patent drawing, you can see how the inventor tries to derail one with his drawings.


    5 years ago on Step 13

    This took me back to my Myst days and it made my eyes mist at the cleverness of this Instructable on so many levels. Kudos for resurrecting an extremely clever patent. Double-kudos for building a working model. Given the problem of global warming and diminishing energy supplies, this device is very much needed!!!

    3 replies

    Reply 5 years ago on Introduction

    Exocetid, did you take part in creating Myst? Or was it only a game to you?

    I think the Steampunk style today would make for a wonderful Myst game world.


    Reply 5 years ago on Introduction

    No, I am not one of the Miller brothers, although my secret name sounds like it. My son and I enjoyed the entire Myst series since I have been a Mac user since 1984.

    Steampunk copied Myst ?


    5 years ago on Introduction

    So this works indefinitely without any source of energy? What if you add a hydraulic generator next to it? Would it be possible?