Wednesday, September 12, 2007

The perfect reverse osmosis system

Perfect also means a good price :-)

The feed shut off valve is not enough. It takes forever to shut off the system and restart. I suggest to add a plastic ball valve right after the feed valve. Surprisingly, the ball valve carried at Lowes carry a cancer causing material warning, but it's sold along side NSF approved tubings. I would buy a valve with John Guest connectors. And since JG seemed to be patented, there's always a minute but ultra fine JG symbol somewhere, even though the manufacturer's name wasn't on the component. This is very typical. Manufacturers don't normally make all the component of an RO system. So they mainly sell to OEM's, unmarked or marked with the OEM's name. BTW, JG is also a company producing many components.

You can easily find NSF approved 1/4" tubing in hardware stores, for fridge ice makers. For the filter housings, what can go wrong? It's hard to check if those components are NSF approved, as most makers sell them to OEM's. And then the whole system is approved rather than individual components. But I bet all housings are made of FDA approved food grade plastics, and that any housing looks like any others. But I'm not so sure about the O-rings used to seal the filters. There are NSF or FDA approved O-rings on sale fit for standard housings. And there are approved silicon grease to be applied on the O-rings to seal.

There are typically up to 3 pre filters about 10" long for residential use. The main purpose is to remove chlorine which attacks the membrane. But I made the mistake of choosing too good prefilters. The worse mistake is to use a 0.2 micron ceramic filter. When I looked at the data sheet, the pressure drop across the filter is 15 psi, though at a very high flow rate which doesn't occur at residential systems. So at 50 psi input feed, you could lose a lot of pressure just across one filter alone. The filters to buy are Matrikx branded. They have a stable and clear line of products. They have a good name and a lot of systems use them (I guess). They provide datasheets. You can get the top ones at as low as $10.

The static pressure is easy to measure. A pressure gauge costs a few dollars from hardware stores. Screw it on a garden tap, 3/4", and you get the static pressure, when all other taps are closed. For 1/4" RO systems, you just unscrew the 3/4" adapter typical on gauges, and you can screw the gauge directly into a JG ball valve for RO tanks. Now you just insert any tube into the valve and you get the static pressure with the valve open. But measuring the pressure when water is flowing is tricky. If the water don't flow, there's no pressure drop across the pre filters. I've seen attaching a gauge to a T adapter. Water is flowing straight through and the pressure is seemed to measure perpendicularly to the flow.

Filtered drinking water is measured in gpd, gallon per day, while the stated flow rate for pre-filters is typcally gpm, per minute. So the stated pressure drop at the high flow rate may not apply at very low flow rates. Anyway since I cannot know this, I minimize the pressure drop by choosing suitable filters.

I'll start with a 5 micron sediment filter, if you water supply isn't that dirty. Coarser filter is associated with larger dirt capacity and less pressure drop, a few psi. Then the 20,000 gal carbon chlorine filter is a no brainer. If you don't really know how much water you drink, it guarantees you only need to change filter once a year. If you know, you save money and time because the chlorine capacity is a lot larger than other filters, you only need to change filters once a long time. I estimated that the water passing through my filter will be 5,000 to 10,000 gal, so I only need to change filters once 2 to 4 years, considering the chlorine capacity alone. It also good for filtering VOC if by chance you got it, volatile organic compounds, which are basically vapor gas that the RO membrane can't deal with. The pressure drop is less than 3.5 psi and the size is about 0.6 micron.

It just happened that I have a 3 stage housing so I have to pick one more. Another same carbon filter is OK but there are a lot of other choices probably for non RO systems. The mistake I made was picking a 0.2 um one with a psi drop of 15, making it unsuitable for RO systems. The choice is between certified lead or bacteria rejection. Since my water feed is very decent, and since the membrane will deal with that at sub micron levels, I settled on a 0.5 um carbon filter with 8 psi drop, with half the chlorine capacity. There is some cyst reduction, bacteria that isn't killed by chlorine. Other reasons are price and availability.

The most important component used to be the membrane. But as I said, pumps cost a lot more. The only membrane to get is Filmtec, certified for 96 to 98% rejection. And 75 gpd is the only capacity you want. Any larger the rejection drops to 90%, and any other brands is just as good. Even if you don't drink that much water, the tank refills fast. And since you don't have 50 psi as assumed by most components, you have more design flexibility and more choice of good filters. After all, it's just $30.

15% to 20% of water pass through the membrane, while the rest go down the drain. What makes water pass through the membrane is a flow restrictor on the brine side (waste water) of the membrane. Interestingly, almost certainly you have a wrong flow restrictor. If you have a 75 gpd membrane, waste water is 4 times that, arriving at about 800 mL per minute, which is the standard restrictor to use. But, the Filmtec membrane recovery is 15% rather than 20%, which requires a larger capacity. Also, instead of 75 gpd at 50 psi, you have 40 psi across the membrane if you are lucky. And for pumpless systems, you have just about 20 psi when the tank is almost full.

You don't have much control over pressure unless you have a booster pump to vary the input psi, and a permeate pump to maintain a constant pressure across the membrane. In my area we need a pressure regulator so I can increase that to 100 psi risking bursting pipes anytime, or decrease at the expense of weaker showers. Other than that you can buy filters with larger or smaller psi drops. However since you can't measure that reliably, I suggest to get a few flow restrictors, test the system performance with a TDS meter, pick the best one and return the rest !!

It's worthwhile to make a bypass path across the restrictor, controlled by a ball valve. It's called flushing the membrane when you by pass the restrictor, when all water goes to drain to clean the membrane. With JG connectors, it only cost you a minute to make. NSF components are not required at the brine side. Expensive systems used to have that path, and of course a timer to flush everyday for a few minutes. Membranes need flushing regularly only if your system is at 98% I guess. It's easy to achieve over 95% and never flush. At $30 it's worthwhile to change the filter every year. But if you use it for a few years for the full design life, you feel better to flush every few months or a year. The waste water are black, mostly carbon resides from the filters.

After the invention of the "affordable" permeate filter, it's clear that prior pumpless systems are toys. No matter how good your filters and membranes are, you only operates at a fraction of the pressure. The pump comes with a new hydraulic valve, which cutoff the supply when the tank pressure is 90% of the feed, and turns on the supply when the pressure drops below 70%. Since the membrane always see full pressure, the high cut off allows the tank to fill more water at higher pressure, and the on pressure prevents starting the system every time for a minute when you just pour a glass of water. The higher pressure is important if previously your RO water can't get through to the fridge to make ice. Previously standard valves cut off at 66%, leaving you with 33% pressure across the membrane. Permeate pump not only save you water, because less water flows through, you get longer filter and membrane life.

The final filter is always after the tank. I wondered why. Until one day I tasted the tank water and it smells and taste like rubber. I think it's a taste of the bladder inside for cheap tanks at least. It's also a good idea to filter the water when you drink it, not when you store it in a tank. I always use NSF certified Omnipure. Unlike Matrikx carbon block filters, which are designed for chlorine capacity and minimize carbon residues clogging the membrane, the final filters are carbon granulates design for maximum bad taste reduction. And I can vouch for it. These filters are smaller and have a much lower capacity as the drinking water to total water ratio is at least 1 to 4 or 6. But still you may need to change every 6 month instead of a year.

Cheap tanks rust easily on the outside and taste like rubber. They also lost pressure fast. There's a tire like bladder inside. When you lost pressure, you need to pump water out using a tire pump, measure the air pressure at 7 psi for typical tanks. Of course you better get a new tank when you lost pressure. The one to get is Flowmatic RO Pro tank. It's NSF listed. It's plastic so it doesn't rust, at least on the outside. The shape and color is a lot more pleasant than metal tanks. It fits well under the sink like a tall slim guy rather than a short fat guy with a beer belly. 4 gallon total is about the largest size that fit comfortably under the sink with space for other things. Practically you get only 2+ gallon for pumpless systems, or 3+ with a permeate pump. You also get a nice reserve in earthquake zones, enough for at least 6 people's maximum drinking needs for one day, but a least a few days for survival.

You have little choice about the faucet. A designer faucet cost you more than all the other components. Most people use Touch Flo faucets, look decent, cheap, and NSF listed. There was a settlement about lead many years ago. So they won't make the mistake twice. An ordinary drill and a special bit from hardware stores will make an extra hole on your sink, but depending on the material, I still have nightmares about it. But in my neighborhood, you can't find people to come in and drill a hole for you, not at a price you want to pay. You rather hire any heavy machinery you need from the hardware stores and blast everything yourself.

You need an air gap somewhere to prevent back flow of waste water. If the law allows you can use a check valve instead. Air gap at the faucet is a bad idea unless the faucet comes with tubings attached. It's a nightmare to attach all the tubes under the sink for a compact faucet at the very corner. You can use the air gap for the dishwasher. There are air gaps designed for that with extra RO input.

A digital TDS meter is a must. It's cheap and compact anyway. It's a measure of the conductivity of the water, which reflects directly the dissolved solids in the water. And since only the membrane can remove dissolved solids, the meter tells you how well your system is doing, if put together correctly at all. Typically for hard water, the supply is 400 ppm. If the RO water is 20 ppm, you get 95% rejection. It can be addictive. If you are a Monk person, it's worthwhile to get one of those that mounted on the sink to give readings continuously.

The other parameters that indicates your system is working correctly is a flow rate meter. But it's rather not necessary. You just need a measuring cup and a stopwatch. You get a good idea of the flow rate out of the restrictor and out of the membrane (when the tank is shut off). Restrictors can be inaccurate because of deposits, which is at higher concentration at the brine side.

Now there are low cost total volume flow meters on the market made of FDA approved plastic. You can program 5 reminders to change your filters according to total volume passing through. Nobody have a good idea how much water they actually use. Filters are either changed prematurely along with all other filters, or way too late. So the saving in filter cost will make the meter virtually free. You also get peace of mind. But you need to know the design capacity of the filters. But actually you need two, since there are three flow volumes in the system, the supply, the permeate (drinking) and brine (waste). Since the permeate and brine are never exactly 1:4 or any other ratio as designed, you need two meters to get all the three flow volumes, unless you have a good idea what the measured ratio actually is.

You need to disinfect and clean the system on a regularly basis, say once a year. The membrane is tricky if you decide to keep it for a few years. It can only tolerate low concentrations of chemicals and you need to pump the solutions through the membrane constantly for limited time. The details are on their website. I would rather flow it away after a year, or just soak it. To disinfect you need H2O2 from pharmacy and dilute about 10 times (check website). I'll soak it in the filter housing instead of pumping it for the recommended period. I have no pump. Then I'll put fresh solution in the empty filter housings, put the membrane back in it's housing, and turn on the feed supply. The H2O2 gets pumped across the membrane, but further diluted.

But if you actually bother to disinfect the membrane, you might just as well buy NaOH and HCL solutions from science stores and perform the alkaline and acidic cleaning steps as recommended.

Before the membrane, you can just as well pump higher concentration of cleaning and disinfecting solutions around the system, the tank, the tubes and the filter housings. Then using the tank to pump the solutions all the way to the fridge. Of course you need to pump a lot of fresh water to get rid of the solutions, and throw away the ice.

I think that's about it. Now the cost.

JG Valve's and T's $30
Filter housing $60
Filter 1 $10
Filter 2 $10
Filter 3 $15
Filter 4 $15
Membrane $40
Pump $45
cut-off valve $15
Tank $70
Faucet $30
TDS meter $20
Flow meter $30
Total $390

This is the retail component prices. If you buy in bulk and sell whole systems, you make a decent profit without much work. The JG connectors are a snap to use.


Judy said...


We just put an RO system in. No pump on either side. The frig is not getting enough pressure for the ice maker to work. Question: Which is better - pump on the inlet side or pump on the delivery side? Also - where did you find such an inexpensive pump?

Anonymous said...

Pump at the inlet will improve water quality, but expensive to deal with main water pressure.

Delivery pump after filters and the membrane won't improve water quality. At low pressure the pump can be cheaper.

I find most of my brand-named components on eBay. Also use Google products (froogle) - some company websites may do well against eBay.

I'll check if the tank and the system is normal before buying a pump. In my 1st system, the tank pressure when full is about 30 psi I think. Enough to send water across the kitchen walls to the fridge.

Judy said...

My frig is only 4 years old and it ways that the minimum pressure needed is 30psi. There is a 25" run of tubing from the ro system to the frig, which doesn't help.
Does this make more sense to you?

Judy said...

Sorry....25ft. (of course)

Anonymous said...

I had a basic set a couple of years ago. I ran more than 50 ft of tube from under the sink to the corner of the kitchen, then up the ceiling, pass the door, down to the floor again, then to the fridge. That's the only way to hide the tubes. :-) They are all still here.

You can always try connecting the full tank directly to the fridge using a short tube. (If there's not too much trouble of course.)

As it's a new system, there's no reason believe there's something wrong. But you should have a TDS meter anyway to see if everything is working fine for years to come.

Another thing to try without buying a pump outright is to use a bicycle pump or car tire pump to increase air pressure of the tank via the air valve. The pressure should be calibrated to a few PSI written on the tank, 7 (?). There is a slight chance that the calibration or the tank is faulty. Indeed I do that periodically to pump water out of my old tank, before retiring it.

If you want to check the water pressure out of the tank, the gauge cost $5 (?) from Hardware stores. Unscrew the 1/2 adapter for normal sized tubes you get the size for RO tubes. The valve/tap on your tank also screws perfectly to the pressure gauge. But you need a spare valve/tap to test the tank pressure.

For systems without pumps, the tank pressure is roughly 2/3 of the inlet pressure. If you have a weak water pressure less than the standard 50 psi, your tank pressure will drop with that too.

Hope that helps.

Anonymous said...

Almost forgot, other than expensive booster pump, or deliver pump, you have the option of a permeate pump too.

They are inexpensive working under lower pressure, all over eBay. They improve water quality.

To increase your output pressure into the tank and hence into your fridge, you need to change the cut-off valve too. With pumpless systems it's 2/3 of input pressure. With the so called 90% cut-off valve, you can have above 80% of the input pressure. For standard 50 psi, your pressure increases from 32 to over 40 psi.

I doubt if any fridge is designed to require pressure over 40 psi (or 30) to work. But the gain in 10 psi will help you get through tubes and gravity. Also, even if the problem isn't the pressure, you aren't wasting anything.

Anonymous said...


I just installed a new Filmtec 50 gpd membrane. Tested the ratio and it's 200:600 product:brine. What should it be for this membrane? Can I cut the capillary tube to increase the flow of brine?


Anonymous said...

> Can I cut the capillary tube to increase the flow of brine?

Tried cutting it; didn't change the recovery ratio at all.

Anonymous said...

Filmtec up to 75 gpd should be 15% recovery, or 1:5

By capillary you mean the flow restrictor? It should be about 800 mL for 75 gpd at 1:4.

If you replace the flow restrictor (or take it off?) with a larger flow rate you should see increase from 1:3 to 1:4 etc.

It's not uncommon to have system designed for 1:3, using lower cost membranes, and to reduce water use. I had one with a TDS of about 90% rejection, as compared to over 95% with Filmtec and 1:5 recovery.

The player

Anonymous said...

OK, so for Filmtec 50 gpd, what should it be? 550 ml/min?

BTW, I put a "420 ml/min" restrictor on it, and measured (using a measuring cup) 100 ml permeate and only 275 ml brine per minute.

Please advise!

Anonymous said...

I have some exact calculations somewhere, but charts are all over the net:

50 gpd 500 ml/min 1:3.5 recovery
So 420 will give you around the figure you get.

For filmtec, you can use up to 1:5 for 95+% rejection rate, measurable with an inexpensive TDS meter.

The player

Judy said...

Could someone recommend an inexpensive TDS meter that works well? Thank you!

Anonymous said...

I had this one, a couple of years ago now, still working fine, after replacing battery. HANNA the company seems well established and serious. It was about $40 then so I'm not surprised that the old model is dropped by half now.

I would go for anything even simpler and cheaper, as long as it's digital, easy to read and use. Though I had seen some toys. HANNA comes with full specifications, and calibration liquids. Sound serious enough.

As for the measurement, it's relative. Incoming hard water is up to 400 ppm, if you have 95% rejection, your drinking water reading will be 20 ppm. If you have soft water, your reading will only be a few ppm. The Hanna have enough resolution of 1 ppm to handle that.

Don't worry too much about absolute calibration. As long as the incoming water and drinking water are about the same temperature, it should be OK.

Be careful if you want to attach any measurement device permanently to the drinking water side. It had better be NSF approved, and it will be a lot more expensive.

The player