Trevor Bird, Australia
Figure 1. Installed Spectra Newport 400 Watermaker
Obtaining a used watermaker
I had always thought we could use a watermaker on board our yacht, mainly so that the notion of staying out indefinitely wasn’t always accompanied with the concern and anxiety about water use. When a well-known brand of watermaker was for sale on our local second hand trading site, Gumtree, I purchased it. It sold for the princely sum of AU$900 and tinkering with my newly acquired Spectra Newport 400 would start me on a course of self-education in how spectra watermakers work.
Just what have we purchased?
The previous owners had removed the watermaker from a motor cruiser and could not get the water production any higher than 40 litres per hour. The unit is rated at 64 litres per hour so they spent considerable time and effort to increase the product (fresh) water output but finally decided to simply replace it for a different type. I was the lucky recipient of that swap out, obtaining a complete Spectra Watermaker at a very reasonable price. I was relieved to see that all of the pieces were there; Clarke pump, filters, computer card with remote user panel and the feed pump and motor. I had all of the basic components of a working unit.
This type of watermaker uses a very innovative hydraulic intensifier called a Clarke pump. The Clarke pump takes medium pressure feed water from an electric feed pump, say 60 – 100 psi and increases the pressure at the reverse osmosis membrane to several times the electric pump pressure. It does that by returning the pressurized brine discharge water exiting the membrane to the Clarke pump, thereby reinforcing the pressure derived from the electrical pump. This system provides an astonishing level of efficiency. I was so pleased to have purchased a watermaker with this great feature that allows water to be produced without the energy impact of having to achieve very high pressures from a powerful electric or engine driven pump.
The Clarke pump had apparently been refurbished so I was hoping that the rebuild was done well, as that was one part of the system I was not too keen to investigate. The unit came with a Leeson 240 volt 3 phase motor and had a 115 volt single phase to 240 volt 3 phase variable speed drive controller. We have 240 volts from an inverter on our boat but not 115volts. I thought it unusual to have a 240 volt motor with a 115 volt input to the speed controller, particularly in Australia. The previous owners must have had a supply of 115 volts on their boat. I also wanted to have the automatic freshwater flushing functions working on our yacht and we always switch off the inverter when the boat in unattended so didn’t want to rely on the 240 volt system. I investigated running the entire watermaker on 12 volts.
The Leeson motor was coupled to a Procon water pump rated at 175PSI. This was a beautifully made stainless steel pump but it was not going to be used so it was set aside.
Figure 2. Spectra Clarke Pump assembly
The Clarke Pump is the secret to the high efficiency of this watermaker. It is a hydraulic intensifier allowing high water pressures to be developed using medium pressure pumps. With this device, an electric pump producing as low as 60PSI can be used to drive the membrane which requires several hundred PSI to work effectively. Figure 2 shows the Clarke pump assembly, which is the box with the “Spectra” sign on it and the shorter tubular assembly below it. The longer tubular assembly houses the Reverse Osmosis membrane. It is designed to separate fresh water out from the salt water. It is 40 inches long, 2.5 inches in diameter and is accessed by unscrewing the end cap of the long pressure vessel. The two hydraulic accumulators mounted above the long pressure vessel are simply connected to the input to the Clarke Pump and are used to reduce the pressure fluctuation when the Clarke Pump reciprocates the pistons in the shorter tubular assembly.
Figure 3. Pre Filters – note pressure sensors on the top
The pre-filters as shown in Figure 3 ensure large volume particulate matter does not reach the Clarke Pump. Poor filtering can cause the Clarke pump to wear internally, resulting in an expensive rebuild. The first filter is rated at 20 microns and the second is rated at 5 microns. A pressure sensor is located on both the input and the output of the filter housing. These pressure sensors allow the computer to measure the differential pressure across the filters. Too much pressure drop across the filters indicates the filters are blocked and need replacing or cleaning.
Figure 4. Charcoal filter and first saltwater filter
The two filters shown in Figure 4 are used to filter fresh water for flushing and to filter the intake salt-water to stop large particulate matter reaching the feed pumps. The housing on the left contains a charcoal filter for the removal of chlorine in the fresh water supply. Chlorine is extremely detrimental to the longevity of the RO Membrane so if marina water is used to flush the system after use, any chlorine must be removed or membrane damage will occur. The right hand filter is rated at 50 microns and protects the pressure pumps. The three way valve on the front of the housing allows the input to the system to be turned off, open to the salt water inlet thru hull or switched to the service inlet to allow the system to be “pickled” with chemical for long term storage.
Figure 5. Original motor and Procon feed pump
The original motor shown in Figure 5 is a large AC Leeson motor which didn’t suit us hence the motor, pump and housing were not used in our installation.
Figure 6. MPC3000 computer card
The brains of the system can be seen in Figure 6. This custom designed computer card has the software to interpret and act upon inputs from the various sensors used in the watermaker. This is an early model card, denoted as an MPC 3000. More recent units are equipped with an MPC5000. The MPC5000 is superior to the MPC3000 as it allows owners to change preset variables from the user panel. The MPC3000 requires the user to interface via an RS232 port to a Windows Computer running Spectra Software to change the preset variables.
Figure 7. Experimental setup for testing purposes
Lets make it work for our application
Spectra have watermakers that operate with one or two 12 volt Shurflo feed pumps. These pumps have special heatsinks and small fans to dissipate the heat generated while working continuously into fairly high head pressures. The pumps are rated at a maximum of 125psi and a maximum flow rate of 6.8 litres per minute. The cost of these pumps in Australia was pretty high so I investigated alternatives that may have been able to do the same job. After some investigation, I settled on an American made Silvan pump, model DDP552a with a rated maximum pressure of 120 PSI and a maximum flow of 7 litres per minute at 30% of the cost of the Shurflo. These pumps are normally supplied for agricultural use and intended for spraying farm chemicals on crops so I reasoned that pumping salt water through them would probably be fine. The Silvan pumps were available as a special order from from Bunnings, a local hardware chain.
The pressure rating of the pumps relates to the pressure that the auto switch-off is set. A pump rated at 120 PSI does not necessarily run at 120 PSI. It has a maximum pressure of 120 PSI prior to the pressure switch turning the pump off. The flow is a maximum flow rate based on an open pipe without backpressure. The small variation between the specifications of the pumps meant they were pretty well equivalent. A major variable of course, is the quality of the pump. With the pump running continuously into a high head pressure, a poor quality pump may fail early in its life. Time would tell.
Control Panel cable
The watermaker I purchased did not come with the cable from the computer board to the operational control panel. It looked like a standard RJ12, 6 core cable as used for a computer LAN or a telephone application. Reading various manuals for the Spectra watermakers, it states very emphatically “Route the display cable from the display to the jack marked “display.” Use only a Spectra-approved cable. The cable is not a standard LAN cable or phone cord.” In various manuals for many models, Spectra went to a lot of trouble to warn against using non-proprietary cables suggesting damage may occur if the pin connections were wrong.
This led me to believe that the cable must have had an unconventional crossover of some kind so I approached this issue with great caution. Eventually I wrote to Spectra in the USA who provided excellent service, very quickly responding with a photograph of both ends of a cable showing the pin for pin connections. Despite the dire warnings from the Spectra manuals, it was a very simple, perfectly standard RJ12 6 core phone cable.
Setup for experimentation
We were travelling around Australia in our caravan when I bought the watermaker. I set it up in the annex and started to learn how it worked. Firstly I wanted to ensure that the sensors and valves were operational so the functionality of the computer could be fully utilized.
I was greeted with “Salinity Sensor Failed” on the user panel that I discovered was caused by a dirty connector on the base of the sensor. It uses a normal RJ12 connector so that was replaced and the sensor connector cleaned. The salinity sensor is simply a device that measures the conductance of the water. Too much conductivity means impure or salty water. It is an indicator only and not an absolute measure of water impurity. It can be easily calibrated against an external salinity meter if one is available.
The pressure sensors on the pre-filters did not respond as they should. The pressure sensor on the output side of the filters was fine but no pressure could be read on the input side. After much cleaning and diagnosis the input sensor was deemed to be faulty. These pressure sensors are quite expensive from Spectra so I purchased cheaper units on Ebay. These were installed and worked fine. Now I could see the pressure the pumps were providing to the input of the pre-filters and also the pressure drop across the filters, thus indicating the state of the filters. If the filters are clogged, the pressure differential across the filters increases; if they are clean not much pressure differential exists.
A failure of the filter input sensor reading high could falsely indicate to the computer that the pre-filters are very clogged and shut the system down. If this happens the input sensor can simply be disconnected and that error message and system shutdown will not be repeated. That is because the computer senses that the input pressure is lower than the output pressure (actually an impossible physical scenario) but the differential pressure does not indicate clogged filters.
I initially set the system up with one feed pump. The previous owners had not pickled the system for storage so I had a fair guess the membrane would need replacing but decided to test the system and not replace it unless it was obviously damaged. It worked and passed clean rainwater successfully so it was a good start. I am still using the same membrane and it is producing good quality water.
The electronics mounting boxes that the system came in would not be suitable for my application. The motor housing would not be required and the MPC3000 control computer board was housed in a metal box attached to the motor assembly. I resolved to re-install the electronics into a separate plastic box with the manual control switches installed on the side. I also resolved that all high current delivery to motors would be through relays so heavy wiring was not required within the box housing the electronics.
The inputs to the MPC3000 board I would be using are as follows:
1. Pressure sensor on input to pre filters (denoted by a RED band of heat-shrink)
2. Pressure sensor on the output side of the Pre Filters (denoted by a GREEN band of heat-shrink)
3. Flow sensor. This is a spinning rotor with a hall sensor device to indicate that product water is being produced.
4. Salinity Sensor in the product line from the membrane output to the diversion valve. This senses when the water quality is good enough to be diverted into the yachts fresh water tank.
5. Vacuum sensor on the input to the feed pump. This allows the computer to indicate if the inlet strainer or the thru hull pickup is clogged.
The outputs I would use from the MPC 3000 board are:
1. PMP1. This is the signal to tell pump 1 to start.
2. PMP2. This is the signal to tell pump 2 to start
3. FVLV. Fresh water valve allowing the fresh water from the yachts fresh water tanks to be used to flush the system after water has been made.
4. DVLV. Diversion Valve allowing fresh water to be sent to the yacht fresh water tank after the salinity sensor has indicated that good quality water has been produced for 1 minute.
5. Connection to the user display.
The computer card also has some other input/output connections that at this stage I will not be using.
Figure 8. Final Installation
The process of selecting the position of various components in the yacht did take some time as the watermaker components are quite large. I decided to install the thu hull myself and chose a Forespar Marelon thru hull and seacock. I also installed a 50 mesh strainer (hole size 0.28mm) immediately after the seacock to keep out large sized debris from the coarse filters.
Figure 9. The moment of truth - drilling the hole for the thru hull
The Clark pump was mounted under the floor just behind the bilge sump and the two Silvan pumps were mounted opposite the Clarke pump. The inlet manifold and high-pressure output manifold were made using common hardware plumbing pressure pipes. These manifolds allowed the two pumps to be operated in parallel, increasing the feed flow and pressure. The software does have a high efficiency mode if desired, powering only one of the pumps to reduce battery drain and still produce a smaller quantity of water.
Figure 10. Pumps joined by a plastic pressure pipe manifold
The fresh water carbon filter, used to remove residual chlorine from the flush water, and the first salt water filter were mounted as shown in Figure 11 under the settee to port and holes drilled through the bulkhead to route pipes and cables. Fresh water for flushing after each use from the yacht tank was provided by the cold water supply running down the port side behind the settee. All wiring was run to the computer, also under the floor near the pumps and Clarke pump.
Figure 11. Fresh water and salt water filters
Brine is sent to a “T” piece on the galley sink drain and the product water was passed through a manual flow gauge and then onto the spare stop cock on the three input fresh water manifold, previously only used to select forward or aft water tanks. From now on, all watermaker water would be sent to the forward tank, and the aft tank would carry marina water.
Figure 12. Flow meter mounted in front of fresh water manifold
The theory is that any problem with either water is isolated and we would normally be using the forward tank while at sea. In the marina we would use the aft tank and fill it with marina water. It is important that chlorinated water not be passed through the reverse osmosis membrane hence the charcoal filter. We prefer not to totally rely on the charcoal filter however, hence the need for a disciplined approach to water segregation as our system does automatic fresh water flushing every 5 days to reduce the growth of bio fouling that causes membrane and filter degradation.
The system works well producing 38 litres per hour with two pumps running and 26 litres per hour with one pump running. We have been using the cheaper Silvan pumps for several months now without issue. I have not yet installed any heatsinks or cooling fans so they do run quite hot. Mostly the run duration is one or two hours but we have run it for 4 hours without the pumps overheating.
Sometimes water maker manufacturers state efficiency as watts per litre of product water made per hour. In our case with two pumps running we draw 14 amps for 38 litres of water produced so the efficiency is 14 amps x 12 volts = 168 watts. 168 / 38 = 4.4 watts per litre.
In high efficiency mode using only one pump we draw 7 amps for 26 litres of water produced so similarly 7amps x 12 volts = 84 watts. 84/26 = 3.2 watts per litre.
Now to calibrate the computer
The greatest satisfaction came from getting the setup software working allowing calibration of the older model MPC 3000 controller computer. This software was written for Windows XP and can be freely downloaded from the Spectra watermaker site. Once downloaded and unzipped, configuration of serial ports is required prior to launching the software. A couple of initialization file problems were resolved and it was running on my Windows 7 machine. From here I could set all of the various parameters that I wanted to see and calibrate. A parameter I wanted to investigate was product flow rate. I couldn’t get the user readout to make sense and wondered why the flow rotor was not signaling properly to the computer.
Figure 13. Electronics mounted in a plastic box under the floor
Figure 14. Computer setup screen
The Spectra machines of this vintage had a flow rotor that provided tach pulses from a hall sensor to indicate flow rate. The software had a check box to indicate a rotor flow sensor was fitted. I checked that box and immediately the flow rate was shown on the display, as it should. I was extremely happy for this to be the case but it did occur to me that the previous owners had never actually had a sensible reading from their user readout for the life of the machine! The computer had simply never been set up properly to display it.
Another parameter that I wanted to set was fresh water flush times. The unit was set to flush for 5 minutes after every run in auto mode. I know this is a controversial subject but I thought a 5 minutes flush was too long. My reason was that the flush is to dump salt water out of the system so as to reduce the chance of bio fouling. Salt water passing through the membrane produces a higher feed pump pressure than does fresh water passing through the membrane. I ran the fresh water flush until I saw the feed water pressure drop and then I knew the membrane was passing fresh water. This happened at about 1 minutes and 35 seconds. On that basis I set the fresh water flush timer to 2 minutes. There is nothing more disheartening than making water for an hour, knowing that half of it will be sacrificed for the fresh water flush. On my 12 volt unit making less water than the big powerful feed pump units, the 5 minute fresh water flush time was wasting a lot of our hard earned product!
We have now been making our own water for several months and the ability to do so has changed our cruising water use. A watermaker does make life more complex when leaving the yacht as water left in the system can cause bio fouling to occur. We use the auto flush feature which pumps new fresh water through the system every 5 days. To leave for an extended time the system can be “pickled” with a storage compound obtained from Spectra. It is important to use the Spectra product as it seems the seals in the Clarke pump can deteriorate if the correct product is not used.
Our system came with a Z-Brane option that establishes an electrostatic charge across the RO Membrane. This apparently reduces the chance of bio fouling in the membrane. At this stage I have chosen to not hook up the Z-Brane and keep the system simple. The Z-Brane system is now discontinued. These days they recommend a “Z-Ion” system, which according to Spectra “introduces a stream of metallic ions that kill the organisms in your watermaker and create an environment that prohibits them from growing and going anaerobic. The result is that your system will be kept ready to operate without any additional flushing, external power sources, pickling chemicals, or complex procedures.“ I may hook up the Z-Brane system in the future but for now we will just periodically flush the system or pickle it for long-term storage.
Purchasing the used watermaker has provided a nice addition to our yacht at a modest cost.
Figure 15. Watermaker user panel at the nav station
Figure 16. Schematic drawing of Watermaker installation
Trevor Bird, Australia