Announcement

Collapse
No announcement yet.

Faema E98 S1 pump conversion from vibe too gear

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Faema E98 S1 pump conversion from vibe too gear

    Hi Guys,

    The other day, Covid-Lockdown-Boredom-Syndrome got the better of me, and I started thinking about more unnecessary projects - like replacing the vibe pump in my old Faema E98 S1 with a rotary pump. It shouldn’t be that hard, as a quick trawl of the internet confirmed that the machines were originally available with a 240 VAC rotary vane pump, the spare pumps and motors were available in Australia, so it looked relatively straightforward.

    But as I started reading about people’s rotary pump conversions, there were other options available. The pick of the bunch seems to be a Fluid-O-Tech FG304 gear pump and motor unit. These units are very compact for a rotary pump, with the pump attached to a brushless 24 VDC motor and variable speed drive. At 60 x 60 x 95 mm, they are not much bigger than a vibe pump, but you still need to factor in some space for a 240 VAC to 24 VDC power supply. The speed control is very simple, using a potentiometer for a 0 to 5 VDC input.

    Click image for larger version  Name:	FG304 Stock.PNG Views:	4 Size:	99.5 ** ID:	885657Click image for larger version  Name:	FG304 Dims.PNG Views:	3 Size:	43.3 ** ID:	885658

    From what I’ve read on various forums, these pumps have been factory fitted to some espresso machines. A number of people have also retrofitted machines with them, primarily to allow for control the flow and pressure, for pre-infusion, pressure profiling, and other esoteric voodoo espresso stuff. At this stage, my main interest is that it is a compact rotary pump, but who knows, maybe the other features would be an advantage as well. Anyway, I need to do a quick sanity check, look at the technical data, and see if the flow and pressure area appropriate, and let you guys know how it all goes.

    Cheers, Stavros.
    Last edited by Stavros; 8 October 2021, 05:27 PM.

  • #2
    Would be a cool project , especially if you can control pressure / flow. I believe the gear pumps are not as smooth/ quiet as a rotary pump. The E98 A1 version came with a rotary pump and could run from in built tank. Cheers

    Comment


    • #3
      Only took a random mention of ' Esoteric Voodoo Espresso Stuff' ....and I'm all in!
      Waahoooo a whole new section please Java baby.

      Pimping My Espresso Machine !

      Keep it coming Stavros.

      Comment


      • #4
        Evening Guys,

        Looking at the manufacturer’s technical data, the 304 gear pump with a 4 mm gear set has a fixed displacement of 0.3 ml/rev and a speed range of between 300 to 5000 rpm. There are some internal losses, and you can see the flow/pressure (blue lines) drop off on the below graph as the pressure increases.

        Click image for larger version  Name:	FG304 flow graph.PNG Views:	0 Size:	90.2 ** ID:	885841


        Against your typical nine bar head, (green points), the pump is capable of delivering between 85 to 625 ml during a 30 second shot. In addition, if you were playing around with pre-infusion or such at three bar (orange point), at the minimum pump speed the flow would be around 5 ml/second. ( In comparison, at nine bar an Ulka EP5 pump would deliver about 130 ml in 30 seconds.) So, there is more than enough flow for my single group machine, and you could easily pull a 40 gram shot at a fairly low speed. Actually, at a low speed, a proportion of the flow would still be going through the pressure reduction valve, (OPV), back to the tank or drain.

        OK, this is the point where the relationship between flow and pressure in a coffee machine does my head in!!! If the pressure is just the result of resistance to the flow through the fine ground coffee puck, at some point the excessive flow from the pump will just open the OPV at whatever it’s set at, and that’s now the fixed pressure you extract at. I can't see how you could possibly reduce the shot pressure below that of the OPV setting while the pump flow is greater than what you want to pass through the puck. It appears that even at low speed, the flow rate of this pump is too great.

        I will need a couple of double-shots tomorrow morning before I go any further.

        Cheers, Stavros.

        Comment


        • #5
          Hi Guys,

          Well, little progress this week, as UPS have lost the pump I ordered from Italy. The pump made its way to Australia remarkably quickly, and was apparently ‘delivered’, but without it turning up at the house. I’ve raised a case with UPS, who now seem to be dragging their heels finding it. A tad annoying - if this continues, I will probably have to buy a second one and hope that they can deliver it correctly.

          Meanwhile, I've been reading up and apparently other people who have been messing around with flow and pressure control have found that the flow from rotary pumps, (including the FG304), can still be a bit excessive. They have fitted either a needle valve or a fixed orifice to the pressure line to allow them to bleed some of the flow back to the tank, reducing the flow to the group at the lower speeds. A couple of people have also noted that at 5000 rpm, the pump is a tad noisy, and have sized their bleed so that they get what they consider to be a suitable maximum flow at approximately 2000 rpm. It's great that other people who have identified these issues - it makes my life a lot easier.


          When I finally get this pump I will need to play around with it, so I'm starting to get a OPV, needle valve, and fittings together to bodgy up a test rig.

          Cheers, Stavros.

          Comment


          • #6
            Can you not decrease flow rate by changing the driven gear (pump) to a 4.5mm or 5mm gear set?

            Upon reflection the unrestricted flow rate of an ulka pump / E61 GH is around 5-8ml sec which equals 150/240ml per 30sec.

            Ill wait to hear what your early tests / observations are.
            Q/ What pump is the Rocket R91 running?

            Comment


            • #7
              Originally posted by EspressoAdventurer View Post
              Can you not decrease flow rate by changing the driven gear (pump) to a 4.5mm or 5mm gear set?

              Q/ What pump is the Rocket R91 running?
              From what I can see, the 4mm gear set is the smallest available.

              I had a quick look online, but can't find any information on the R91 pump. Mayne people who spend $9000 on the machine typically don't pull them apart to see how they work, ,

              Comment


              • #8
                Good afternoon guys,

                Why is it a good afternoon? Because someone found a lonely little box bouncing around in the back of a UPS van, and it was , , , MY PUMP! Yes, nine days after they said they delivered it, it finally arrived.

                This is a solid little pump, quite heavy for something the size of an Ulka pump. But it's so compact you could fit it in just about any machine, possibly even into a Silvia if you were crazy. I'm certainly going to have no trouble installing it into the Faema E98 - there will be loads of room.

                Click image for larger version

Name:	IMG_0671.jpg
Views:	401
Size:	26.7 KB
ID:	887345

                I had a fiddle with it this afternoon after connecting it up to a 24VDC power supply and my little brass test manifold. I'm using an aftermarket OPV , a pressure gauge, a check valve on the discharge, and a modified valve to bleed off some flow back to the tank. The quick test I've done looks very promising - the pump runs well, and the speed/flow is easy to control with the potentiometer.

                Click image for larger version

Name:	IMG_0673.jpg
Views:	403
Size:	31.1 KB
ID:	887347
                However, there are a few things I've noticed.
                • The pump is a not completely quiet. While it is way, way, quieter than an Ulka, it does still buzz a bit. And at high speed it also has a bit of a whistle. Where the pump sat on the bench, and where the hard teflon discharge hose touched things, there was a buzz and vibration. So it looks like it would be a good idea to mount the pump on some rubber isolation feet, and route the pipework well away from the frame or other hard points. (Note that as I've never heard a rotary vane pump working in a coffee machine, I can't compare it with one.)
                • The other thing was that the pump installation manual is correct when it said it needs a 3.5 amp supply. When I ran the pump at full speed against a 9 bar pressure, my 2 amp power supply tripped out on over current. I'll have to get a larger one before I go much further.
                • And finally, the $24 aftermarket OPV valve I bought is a bargain. It is way quieter than the factory one fitted to the E98, which has a annoying habit of randomly squealing and chattering during a shot. I'm now considering plumbing the new OPV into the the machine when I install the pump, and adjusting the Faema one to 10 or so bar so it never opens during a shot again.
                Anyway, things are looking good, so I will go back down to the workshop and have another play with the pump.

                Cheers, Stavros.
                Attached Files

                Comment


                • TailorCoffee
                  TailorCoffee commented
                  Editing a comment
                  This looks amazing! Its small enough to fit into most machines from the look of it. Makes me think you could turn an ECM Classika into an almost perfect machine for an espresso only enthusiast. I always wished i could have had a rotary pump with the classika. What potentiometer are you using and how are you rigging that up to the machine? Also, maybe I missed it in your earlier posts, how much was the pump?

              • #9
                For a cheap 24V high current constant voltage PS, check out LED power supplies (just make sure you get a constant voltage model not constant current).

                An example is the Meanwell LPV 24-100 which will supply 4.2A at 24V and will cost you about $35.
                Edit:removed incorrect comment re speed control.

                From the curves in your post above the total power drawn from the supply will reduce when the speed is reduced.
                Last edited by Lyrebird; 21 October 2021, 06:31 PM.

                Comment


                • Stavros
                  Stavros commented
                  Editing a comment
                  Thanks for the tip re the Meanwell power supply Lyrebird - I just got a new LPV 24-100 off FleaBay for $29

                • Lyrebird
                  Lyrebird commented
                  Editing a comment
                  Glad to be of help.

                  FWIW I went on a different tack with my pump conversion and used a rotary vane pump which I adapted to take a DC motor. I'm using the big brother of that PS: I think it's the 10 amp version.

              • #10
                TailorCoffee - Speed control is via a 47K pot wired up as below. The installation manual, including other options for the wiring the speed control, can be found here. I really haven't thought about how this all fits together. I will likely continue to use the original shot switch to operate the solenoid valve, and also provide 240 VAC to supply power to the 24 VDC power supply. The pot will probably ends up being installed on the bottom panel next to all the other electrical control switches.

                Click image for larger version

Name:	Pot Wiring.PNG
Views:	418
Size:	39.5 KB
ID:	887363

                Being a cheap bastard at the best of times, I purchased it from a seller on eBay. Including UPS random delivery, it cost me AUS$400. This was cheaper than I could find a full size 240 VAC motor and rotary vane pump typically used in these machines, so it was an attractive deal for me.

                Oh, and further testing this afternoon, at the minimum pump speed, (while using a small bypass flow back to the tank), I measured a flow of 25 ml in 60 seconds against a head of 1 bar. This seems like it would be slow enough for some sort of pre-infusion? ? Maybe? ?

                Cheers, Stavros.

                Comment


                • TailorCoffee
                  TailorCoffee commented
                  Editing a comment
                  Oh nice! this all seems way over my head, but will follow along eagerly!

                • Lyrebird
                  Lyrebird commented
                  Editing a comment
                  Just note that that pot is 4k7 not 47k. Standard electrical notation is to replace the decimal point with the unit, so 4k7 = 4.7k not 47k.

                  Reality check: the control voltage is 0 - 5V which is 0 - 1 / 4.8 * 24V. Divide by 4.8 with a 15k pass resistor requires 3.95k, 4.7k is the nearest standard value above this (can't go below or the motor won't reach full speed).

              • #11
                Stavros full of admiration for you patience and competence.

                Comment


                • #12
                  Originally posted by Stavros View Post
                  I really haven't thought about how this all fits together. I will likely continue to use the original shot switch to operate the solenoid valve, and also provide 240 VAC to supply power to the 24 VDC power supply. The pot will probably ends up being installed on the bottom panel next to all the other electrical control switches.
                  You'll have a better time if you use the original 240V pump supply to switch a 240V coil of a contractor or chunky relay, and use that to switch the 24VDC output from the power supply. Other option is a solid state relay but they don't tend to get on well with inrush current (if that pump is unregulated on inrush current).

                  If you just switch the supply to the power supply your pump will be delayed turning on and delayed turning off as the output of the power supply builds and drops voltage like a compressor would air pressure without a closed valve.

                  Comment


                  • Lyrebird
                    Lyrebird commented
                    Editing a comment
                    You could avoid switching and inrush issues altogether by placing a switch in series with a 100 ohm resistor across the potentiometer in the circuit shown (eg between the orange and black lines). The switch would pull the speed_in line down to < 0.2V at which voltage the motor will not run.

                    Place an appropriately sized capacitor across the switch and when opened the voltage on the output will rise as the cap charges through the 15k resistor. A 33 uF cap will give you a time constant of ~500 ms which means the pump would get to 63% of full speed in half a second.

                    The 100 ohm resistor stops the capacitor discharging in a flash as the switch is closed. 24 x 100 / (15k + 100) = 160 mV so you are still below the 200 mV threshold when the switch is closed.
                    Last edited by Lyrebird; 21 October 2021, 11:10 PM.

                • #13
                  Since the description above is confusing even to me, here's the circuit:



                  The 100R resistor could equally be in series with the cap.

                  It should be noted that the capacitor will also slow the response to pot changes if you used the pot for flow profile. When I played with an R9 ONE it had a fairly slow response to control inputs so I'm not sure it's a problem

                  BTW I got the time constant wrong, the net resistance at the input to the cap is 4k7II15k so about 3k5, thus the TC will be about 120ms. If you want longer increase the cap as needed.

                  I also forgot to draft the wiper connection on the pot, it's as per the original diagram.

                  Comment


                  • #14
                    Guys,

                    I mentioned earlier that I would probably need to mount the pump on some vibration isolating mounts to keep it as quiet as possible. I‘ve done something similar in the past with an Ulka vibe pump, where I knocked up a mount from a 20 mm thick foam gym mat. It wasn’t very scientific, but it worked well enough. At the time, Artman posted a link to a guy who used a phone app' to measure vibrations, and help select suitable mounts. So, that was how I did it this time.

                    This wasn’t that straightforward, as the pump is variable speed, and the frequency of the vibrations varies with the rpm of the pump - doh! It would have been easier to select mounts for a fixed frequency 50 Hz vibe pump. Anyway, the best thing I could do was measure it at the pump speed required to produce a nine bar pressure.

                    Click image for larger version  Name:	Nine Bar.JPG Views:	0 Size:	8.1 ** ID:	887790

                    In the end, I selected some nice gel anti-vibration mounts made by a Japanese company, Taica. Unlike the yum cha ones on eBay, they are backed up with technical information, including recommended loads and frequencies, making selection a mount for the pump weight and frequency relatively simple - provided it's running at the same speed you test it at, , ,

                    And no post in complete without some pictures of unnecessary disassembly of electro-mechanical items. The wet end has a couple of silicone O rings to seal it, and no shaft seal as it is mag' drive. And both ends have alignment dowels, so you can't put it back together the wrong way.

                    Click image for larger version  Name:	IMG_0676.jpg Views:	0 Size:	15.3 ** ID:	887791Click image for larger version  Name:	IMG_0678.jpg Views:	0 Size:	16.1 ** ID:	887792


                    Cheers, Stavros.


                    PS. Taica's summary of vibration dampening can be found here.

                    PPS. Oh and by the way, I like the pump so much, I've ordered a second one.
                    Last edited by Stavros; 23 October 2021, 04:19 PM.

                    Comment


                    • #15
                      Keep 'em coming Stav !

                      Comment

                      Working...
                      X