It has been an interesting learning experience and I thought I'd jot down some notes in case anybody else is looking for inspiration.

I have photos of the first generation here: http://photos.wemm.org/autochanger

I change 5% of each tank's water every 6 hours.  That makes a 20% change per day.  The fish essentially live in fresh dechlorinated tap water as there is no time for contaminants to build up.  In fact, when filters fail, there is usually no ammonia buildup either.  At least, not in any but the most heavily stocked tanks.

The main problem was that I had to run water pipes all over the place to various fish tank locations, and when the system was turned off, the pipes drained the water and filled with air.  Turning it on made quite a noiseas the air bled out.

The second problem was that we prefered the taste of dechlorinated water for drinking, coffee, etc.    My wife wanted to use dechlorinated water but didn't want to have to fight her way into the fish area to turn it on.  Dechlorinated water on-tap would have been much nicer.

Also, I was a little concerned that I might not be getting enough contact time between the water and the carbon.  It was very high grade carbon, but with the flow rate I was using, it was only getting about 90 seconds of contact time instead of the recommened 10-15 minutes.

Finally, when I changed the carbon in the filter after about 5-6 months I discovered that the process was quite a nightmare.  We got water and carbon grit everywhere, including on the carpets etc.  I didn't want to have to go through that again.

So, I modified it to a second generation system.  I kept the overflow system on the tanks unchanged, but moved the production system outside.  The carbon filter now used an Aquatec tank level controller to keep a 44 gallon barrel full of dechlorinated water.  The flow rate was much slower, giving about 8-10 minutes of contact time.  I used an Aquatec delivery pump to pressurize a 14 gallon bladder tank and fed water inside to the three locations that needed dechlorinated water.  (Kitchen and the two fish areas).  This time I ran the water tubing under the house and up through the floors to the locations that needed the water.

Like before, my computer controlled the water system using UPB appliance modules and Aquatec solenoid valves to turn on water to 3/8" tubing, non-return valves and 20 psi pressure regulators.  The 20psi output fed the dripper system and filled the tanks to the siphon overflow point.

This worked out very well.  Except for one major problem.  Our tap water is too soft.  Most of the water in the San Francisco bay area is in the 150-200ppm TDS (total dissolved solids) area.  Our water (near Walnut Creek) is 45ppm and very alkaline (pH 9.2) and very very poorly buffered (<14ppm KH, and <17ppm GH).  It is mostly snow melt from the Sierra Nevada mountains. They add raw sodium hydroxide to raise the pH at the sterilization plants.  ie: the pH is high because of the NaOH rather than buffers.  The water is great for the Killi folks, but lousy for hard water fish like guppies.  With this low level of buffering, it easily had massive pH swings if a dripper blocked up.

I was having periodic problems with dropsy like symptoms until I started adding aquarium salt by hand.  I figure the guppies just were just under constant osmotic stress due to the very soft water.  My guppies were not particularly happy campers until then.

Adding salt by hand was getting really old, really fast.  I'd forget to check for a few days and was having large TDS swings.

So, I finally got annoyed enough and went ahead and built revision 3 of the processing system.  I wanted to automatically add additives to raise the TDS to around 120ppm for the guppies.  What I did was the following.

Stage 1: dechlorination.
Instead of the 6 foot long carbon filter, I picked up some 10" filter bodies and extruded carbon block filters from http://www.wateranywhere.com/ - I have a sediment filter in front of 5 carbon block filters in series, with a tap before the last three.  That way I can periodically test for chlorine leakage past the first two carbon filters which would indicate time to change them.  I use a 100gpd RO waste flow restrictor to control the flow through.  (Note that it doesn't let 100 gallons per day through. On a RO unit, it is the flow restrictor for the waste water so that the RO unit gives 100 gallons of product water per day.  I figure it lets between 500-1000 gallons per day.)

I have a Hach low range free/total chlorine test kit and have confirmed that just one carbon block filter at that slow flow rate produces a zero reading for total chlorine, even on tap water with 2ppm of chloramine in it.  I'm using the H-F2510AC cartridge from wateranywhere (who are a retail division of "applied membranes").  5 of them in series is excessive, but I figure it will be a year or so before I exhaust the first two.  2 x $7 for a year's supply of dechlorinated water is a bargain compared to dechlorination chemicals.  Having three afterwards should give me plenty of time to be lazy about noticing the leakage past the first stage.

A tank level controller keeps a 44 gallon barrel full of dechlorinated water.  An aquatec delivery pump feeds a 14 gallon pressure bladder tank.  This water feeds the tetras inside and stage2a and stage 2b.

Stage 2a: a tank level controller takes water from stage 1 and keeps a 44 gallon barrel full.  But, I have a submerged pond fountain pump with a TDS sensor attached in the bottom of the barrel.  This is attached to a Hanna TDS controller, which I got from http://www.automatedaquariums.com/  This controller drives a peristatic pump that feeds Kent's "RO Right" additive to the barrel.  The submerged pump keeps the barrel stirred and measures the TDS as the additive is going in.  I have a larger Aquatec delivery pump and a larger 14 gallon pressure tank and a 5/8" tube run to the guppies inside.

Stage 2b: just a regular RO unit for kitchen drinking water.  The RO water passed the taste test and since everything else was in turmoil, I set that up too.

So far, everything has been working very nicely.  One of my indoor tanks that has had a particularly troubled group of fish has suddenly come to life.

Kent "RO right" might not be the best additive, but it is convenient in liquid form and our tap water isn't far from what would be RO water in some other hard-water areas.  It is deficient in many trace elements.  And so far, the reaction in a couple of tanks has been encouraging.  I do plan to switch from adding pure liquid RO right to a watered down mixture with some aquarium salt added as well.  I'm hoping that will reduce the overshoot that happens because the mixing of the highly concentrated RO Right isn't quite fast enough and there is a few minutes of time delay between the drips reaching the mixing barrel and it registering on the TDS controller.

Now all I need to to is figure out how to make an automatic gravel vac system and an automated algae scraper.  That would really make things nice and easy. :-)