Stumbled across this thread, thought I would add my own experience setting up a daily auto-change system about 2 years ago.
On e bay, I found a used variable-speed Masterflex pump (7553-20) for $185 and a dual pump head for it (7018-20) for $49. I purchased some 7/16" OD x 5/16" ID silicone tubing for the pump on amazon for $31 (I had to buy 25' but I use the excess for other projects
). For a SW reservoir, I purchased a 165g vertical poly tank that I keep in the room behind my 270g display tank. I placed two float switches in the reservoir that connect to my Apex. One triggers it to automatically refill with RODI when close to empty, the other stops the RODI refill when enough has been added to perfectly mix with one box of salt to 1.026. Two MJ mod 1200 pumps take care of the mixing.
Initially I plumbed the reservoir to the pump and sump using 3/8" poly tubing, with the removal line "upstream" of the addition line. Because the tubing run from the reservoir was longer than the run to the drain, and because there was significant back-pressure on the addition line when the reservoir was close to full, I was skeptical that each head on the pump would move identical amounts of water. I tested this by having the pump fill two 5g buckets "in situ", and indeed there was a substantial difference between the amounts removed/added. To address this, I put a partially-closed John Guest 3/8" ball valve on the suction side of the line that was moving the larger amount of water (the drain line). At first this didn't work properly, presumably because John Guest fittings are designed to work in pressure applications vs. suction, and some air was slipping past the fitting's o-rings. I solved this by simply putting the valve on the intake end of the drain line, submerged in the sump. Through trial and error, I was then able to "tune" the two lines to move identical amounts of water. (For accuracy, I decided to use two identical buckets and weigh them with a Berkley digital fish scale that weighed to the 1/10th lb, or a little less than 2 fluid ounces.)
A second concern was that as the water level dropped in the SW reservoir over time, the back pressure on the addition line would be constantly changing, meaning it might add more water when the reservoir was full vs. when it was close to empty, all the while with the drain line constantly drawing from the same water level in the sump. This meant, while I had "tuned" everything with the ball valve, it would gradually go out of calibration as the reservoir drained a little each day. I tested this using the bucket comparison method again, and saw that was indeed the case.
My solution to this was to make a small tank that was connected to the reservoir though 3/4" vinyl tubing, with a high-volume float valve in this small tank. The small tank sits on the floor next to the reservoir (which is about 10" off the floor on a stand with casters). The float valve sets a constant water level in the small tank, and the water-adding line of the pump draws its water from this small tank. This arrangement works very well.
Occasionally I will test to make sure everything is still tuned properly, because it seems to slowly drift out of calibration over time. To make it easier to adjust, I recently swapped out the ball valve with an all-plastic needle valve. Very tiny adjustments make a significant difference, and it was just too sensitive to the coarse adjustments presented by a ball valve.
I have the system set to run for 15 minutes daily, which amounts to about 4 gallons changed, or between 1-2% system volume.
