Compressor

This page last updated: 16 May 2021.

Reference

For reference, the sizes of pipe threads used below are as follows:

BSPP means BSP parallel (P for parallel), which is the default for BSP (AKA G).  BSPT (AKA R) means that the male version of the thread has an outside diameter which is tapered (T for taper), increasing from the size given above slowly along the length of the thread; this means that a seal is formed purely by the mating of the threads themselves (with some PTFE tape), no shoulder on the end of the pipe with a gasket/washer etc. is required to form an air-tight joint.  The female version of this thread is just BSPP (AKA G) but can be denoted Rp, presumably to indicate that there is no reliable shoulder to butt against; the mating of the threads must make the seal.

Some Numbers

1. be as silent as the Jun-Air 6-25 machine (i.e. 45 dBA), since it is run in a loft workshop,
2. be able to deliver more air than my existing Jun-Air 6-25 machine and for roughly two minutes; the motor on the 8 bar, 25 litre, Jun-Air 6-25 machine would normally kick-in within 30 seconds in my application and is not meant to run more than 50% of the time (with a maximum running time of 15 minutes),
3. be of a vertical rather than horizontal construction; I have very little space.

To put better numbers on the air delivery, since that is pretty crucial, the spec for the Jun-Air 6-25 says it has a "displacement" of 50 litres/minute and a "FAD" of 32 litres/minute at 8 bar.  The displacement number is apparently an internal thing, the number to conjure with is FAD or Free Air Displacement.  So 32 litres/minute of air ends up in the 25 litre reservoir built into the Jun-Air machine, hence it should take less than a minute to fill up.

I wanted roughly four times my existing capacity, so that implied a reservoir of 100 litres, and I'd like a slightly higher pressure, let's say 10 bar.

There are lots of videos out there on the internet of people taking any old reservoir tank, e.g. one from the air braking system on a lorry, scraping off the rust and bolting a refrigerator motor/pump to it.  The very useful MIG welding forum has an FAQ which links to a very useful article in the web archive where it says:

Remember, when you're sizing your compressor, that 1 HP [746 Watt] of compressor output generates about 4 CFM [113 litres/minute] of compressed air at about 90 PSI [6.2 bar]. This is the accepted wisdom for air compressors over 10 HP in size. For smaller compressors, those under 10 HP, you must rely on the figures provided by the manufacturer or guestimate that you're getting around 2-3 CFM [57-85 litres/minute] of compressed air per HP @ 90 PSI.

Refrigerator or freezer motor/pumps don't exceed 10 Watts, so they would take a little while to fill a 100 litre tank.  To fill a 100 litre tank in a couple of minutes (hence allowing a continuous flow in a pinch), applying the above rule of thumb, my motor/pump would need to deliver around 50 litres/minute which suggested at least a 1 HP or ~750 Watt motor/pump.  Looking at the fixed-spec Jun-Air machines the motor/pump is 400 Watt and, for larger reservoirs all they do is add more motor/pumps, the 150 litre machine having six of them.  The problem for me with these machines was that the vertically-oriented ones maxed-out at a 25 litre reservoir, going horizontal for the 40 litre and 150 litre reservoir versions, so they didn't meet my physical spec.

Having checked around the internet, the only quiet motors available were those built into the Jun-Air or Bambi hobbyist/dental compressors; these have the additional advantage that the unit is a motor and a pump, two problems solved in one. I already had one Jun-Air compressor motor/pump.  And I found someone selling another Jun-Air machine with a leaky reservoir for parts on E-bay and bought it for £26.  That was enough to get me started.

Sketch Design

A compressor consists of a motor driving an air pump to deliver air into a reservoir tank (so that the motor doesn't have to run continuously).  The reservoir tank is fitted with an adjustable pressure switch that turns the motor on when it is at less than the desired pressure (with a little hysteresis).  There is a non-return valve in the pipe going from the pump to the reservoir tank, so that the pump can be switched off without air escaping, and the reservoir tank also has a pressure relief valve to prevent accidents.  Those are all the essential components.

I searched E-bay for a vertically-oriented 100 litre air tank and found the EXTREMELY useful [if you're in the UK] Context Pneumatic Supplies.  Not only were they able to supply the tank with fittings (for ~£300), they were able to supply most of the other things I needed as well and were very patient with my voyage of discovery concerning compressor construction.  I sent them first the picture on the left.  Then I realised that I might want to amp the system up with more motors/pumps and so I sent them the one on the right, swapping the Y-connector for a manifold and adding a couple more non-return valves:

As they only had one of the 100 litre tanks in stock I ordered that immediately while we discussed the above; it was on my doorstep within 24 hours.  It was a beast: definitely a two-person lift.  Holes/fittings as supplied with it were as follows:


Basically the hole in the top takes the 1/2" BSP, 11 bar, safety valve, the hole in the bottom takes the 1/2" BSP elbow and ball valve, the holes either side take the two 3/4" BSP ball valves and, of the holes in front, one takes the pressure gauge and the other is capped off (this is where the pressure switch will go).  I found that the threads in the tank were a little rough so I bought myself some BSP taps to tidy them up.  For some reason only a single fibre washer suitable for the 3/4" BSP ball valves was supplied so I also bought a set of assorted fibre washers.

Salvage

In terms of bits that I could re-use I now had these two lovelies:



The one on the right, my old compressor, had spent its working life opening and closing a door somewhere and so had been fitted out for "continuous" operation with an automatic drain (first two pictures below), output filter (with 6 mm push-fit output) and drain bottle:


Basically the mains-driven automatic drain valve opens for 0.3 seconds every up-to-30 minutes, with the intention of letting water that has been compressed out of the wet air leave the tank, while the filter removes any spray from the oil-immersed pump/motor; both drains empty through 4 mm push-fit piping into the bottle.  In my case the reason the tank had sprung a leak was likely because the end of the pipe attached to the automatic drain had clogged and hence the tank had rusted due to the collecting water.  All I needed was appropriate adapters to re-use all of this in the new build.

The thread on the automatic drain was nice normal 1/4" BSPT male but the thread on the inlet connector of the filter was very peculiar, I couldn't find a match for it in any standard (CGA, JIC, NPT, BSP, UNF, etc.); it had far too many threads per inch (> 25) for such a large major diameter:



I asked on the ever-helpful mig-welding.co.uk compressor forum and they pointed out that there is a hex inside that bull-nosed connector and, applying some considerable force (a 7 mm allen key, a vice and a hammer) I was able to remove it, revealing a 1/4" BSP female thread inside the top of the filter; much better.

Then there was the pressure switch: the one provided with a Jun-Air compressor, though set to 8 bar, was fully adjustable up to 16 bar and was rated at 240 Volts/20 Amps and 2.2 kWatts, easily within what I need for two motors @ 2.9 Amps each and likely OK for four since the effect of the additional load is to reduce the already large > 100,000 cycle operating life (and in any case I have a spare).  The connection to the tank was 1/4" BSPT male again, nice and simple.



And of course there were the two pumps/motors with their starter capacitors, very slightly different generations of the same model, with nice normal 1/8" BSP male connections:

Connecting Up

To connect everything up I ordered the following from a combination of Context Pneumatic Supplies and RS Online, total cost about the same as that of the tank (since many of the items from RS had to be bought in packs):

And here they all are.

Test Assembly And Adjustment

For an initial pressure test I pneumatically connected (a) the two pumps/motors to two of the input ports of the manifold through non-return valves (the spare output port being blanked off and the two spare input ports being blanked off using non-return valves with blanking plugs fitted just in case), (b) the manifold output to the lower-right hole of the tank through a ball valve, (c) the pressure switch to the lower hole in the front of the tank, (d) the pressure gauge to the upper hole in the front of the tank, (e) the pressure relief valve to the hole in the top of the tank and (f) ball valves to the upper-left and bottom holes of the tank to close them off.  Electrically both motors were connected to the pressure switch and the other end of the pressure switch to a mains plug, all re-using cabling salvaged from the two Jun-Air compressors.  Assembly required a heavy duty adjustable spanner (adjustable up to 30 mm in width), a 6 mm allen key, an electrical screwdriver, wire strippers and the pneumatic PTFE tape (a few turns on each thread applied in a clockwise direction).

All quite satisfactory: it took around 10 minutes to pressurise the tank to 8 bar.  Next I needed to adjust the pressure switch to give me 10 bar.  From the Jun-Air manual:

The cut-in pressure (normally 6 bar) is set by adjustment of differential screw B; turn clockwise to reduce the cut-in pressure.  The cut-out pressure is set by even adjustment of the two screws A (cut-in pressure + differential = cut-out pressure); turn clockwise to increase the cut-out pressure.  The switch is normally factory-set for operation at between 6 and 8 bar (approximately 90 to 120 psi).

I found this confusing: screw B is called the "differential screw" and apparently adjusts the "cut-in pressure", while the equation in brackets has the "cut-in pressure" plus the "differential" pressure giving the cut-out pressure, even though the cut-out pressure is stated as being adjusted by turning screws A; so do screws A adjust the "differential" (despite the name given to screw B) or the "cut-out pressure"?  I decided to see if adjusting screws A would adjust the cut-out pressure by following the three step adjustment procedure shown in the second picture below, making sure to keep my 7 mm spanner and fingers clear of the exposed mains connections.  One turn each of screws A increased the cut-out pressure by about 0.25 bar so it took a few iterations to achieve my desired 10 bar cut-out, leaving time for the motors to cool between the later iterations.  When done I checked the cut-in pressure and it was around 8 bar. I did try adjusting it upwards to exactly 8 bar but that increased the cut-out pressure without moving the cut-in pressure so I put it back as it was; good enough.  It took 3.5 minutes for the two pumps/motors to re-pressurise the tank from cut-in to cut-out and 15 minutes to pressurise it fully from empty; could really do with another pump/motor or two but, as it is, just within the operating limits for a Jun-Air motor.

Final Assembly

With all that sorted it was time for final assembly.  To avoid further clutter in my small loft workshop I fitted the compressor into half of a fitted wardrobe on the floor below.  Since this wasn't a very open space I purchased a number of 240 Volt fans, 70 mm x 70 mm x 25 mm, and 3D printed myself some stands (in PLA, 0.3 mm resolution, each took 3 hours to print; the two parts just shove together) to mount them in. These were screwed to the floor in pairs behind each pump/motor, space being available for three pumps/motors.  The tank, pumps/motors and drain bottle were free-standing; they weren't going anywhere.  I purchased a fused mains switch and some junction boxes, plus some shrouded mains terminal blocks to connect the suppressor capacitors for each of the fans in a safe manner.  The automatic drain and the fans were wired such that they would always be on, i.e. irrespective of the state of the pressure switch.  When I completed the testing above I had found that no water came out of the hole in the bottom of the tank afterwards so I set the automatic drain to the maximum periodicity of 30 minutes. I wanted the user connection at the pressure regulator to be mounted out of the way on the side of my workbench but I also wanted the pressure gauge to be easily visible so, rather than screwing the gauge into the body of the pressure regulator as intended, I mounted it on a short length of 4 mm tubing (re-using the push-fit connectors that I had originally intended to use to connect the automatic drain valve) and a length of stiff wire that allowed me to adjust its position.

I downloaded a mobile phone app (the Bosch one) which purported to measure sound level in dBA and, 0.5 metres in front of the open wardrobe door, with everything running, it measured 65 dBA.   To prevent the floor of the wardrobe resonating too much I went to my local metals merchant and had a plate of 5 mm thick mild steel cut (for £16) to sit underneath the motors/pumps, mounted on a couple of thick rubber grommets front and back to act as feet, and that reduced the noise to about 62 dBA.   I probably wasn't going to get much better than that in such an enclosed space.  I later moved it to a different wardrobe so that it wasn't against a party wall with my neighbour, just in case, so here it is in its final resting place; interestingly, same sound level measured.