A diagram would be nice. How close is the cold reservoir (CR ) to the tap ? If it is close why not have several TECs cool the water pipe feeding the tap ? In addition to one cooling the main CR. Have the TECs hot side connected to a small copper pipe ( SCP ) that runs from the cooling loop. Use a non-potable pump to move water slowly through pipe and return it to the top of the cooling reservoir. When the TECs are off they can be retracted from contact with the SCP using a small motor and detecting stall current. Temp sensors How about a MAX6575 or equivalent. built in 8 device multi-drop capability. 3 connections ( Vdd [ 2.7 to 5.5 V ] , GND, Common I/O ) $1.35 each Is that too expensive ? AGSC ( ITPOP---ITIITBS )** 000000000000000000000000000000000000000000000000000000000000000000000000 00000 On 2006-Mar 18, at 11:50 AM, Robert Ammerman wrote: Some of you may remember a thread that I started a month or two ago about finding a potable water pump for a TEC based water cooler I wanted to build. I received many useful and interesting replies. Those, of course, are two overlapping, distinct, subsets of all the replies :-) Well, I have progressed in my thinking and want to expose my current thoughts for comment. See the section after the line "******" for specific areas where I am looking for ideas. As currently designed, my system would consist of three primary components: 1: A cold water reservoir. This is fed by the output of the water filter, at house mains water pressure, and the output is to the separate "good water" tap at the sink. 2: A cooling loop. This consists of a water-to-water heat exchanger within the cold water reservoir (just a coil of copper tubing), a small pump, and a waterblock mounted to the TEC for extraction of heat from the system. 3: A heat sink. This would consist of a large container of water thermally connected to the hot side of the TEC. I expect to fabricate the cold water reservoir from a vertical section of large diameter PVC pipe. This is both pressure-safe and safe for drinking water. The reservoir will have two ports: (a) at the top, via a simple diffuser that forces the water to spread out horizontally for the incoming water, and (b) at the bottom for the cold water draw off to be sent to the tap at the sink. I need to complicate things a little bit, though, because I have to run a loop back from the tap which returns to the top port. This allows me to keep the water in the loop at the cold reservoir temperature so that it is immediately available at the tap. (Otherwise I'd have a problem similar to waiting forever for the hot water when at a distance from the heater and waste a lot of my cold water just filling the tubing) The cooling loop is a separate closed loop for several reasons: 1: Its components do not have to be safe for drinking water. 2: It can operate at atmospheric instead of house mains pressure. 3: Its cooling liquid can contain additions to inhibit bacteria, mold and corrosion as necessary. 4: Its cooling liquid can run below 0 C to provide more efficient heat transfer to the reservoir My heat sink will just be a large container of room temperature water (in the basement in my case) thermally connected to the hot side of the TEC. I can make a lot of cold water before the heat sink will get too hot, and if it does the controller can just shut the system down, or refill the heatsink with new tap water :-) To maximize efficiency, the cooling loop (not including the pump motor of course), the cold water reservoir, and the loop to the tap (about 2 m long) would all be 'hyperinsulated'. Most of it would be inside several inches of foil-faced polyisocyanurate foam (best R for given thickness, easy to work with) sealed with foil tape. Both the up and down runs of tubing for the loop to tap would be enclosed at the center of a single assembly of two or three layers of standard foam pipe insulation wrapped with foil and sealed with foil tape. With this construction, my computations show that I can limit loses to 1 or 2 Watts or so, not counting whatever loss I get by reverse conduction through the unpumped liquid in the loop from the heat sink when the TEC is off. Just a couple of numbers from my thinking: Desired dispense temperature: 5 C Recovery rate: 2L per hour assuming incoming feed at 20 C Heat flow required for recovery above: 35 Watts Reservoir capacity: 4L Reservoir size: about 50 cm of 4" PVC pipe (sorry for the mixed units but I really know nothing about metric pipe sizes). ******** So finally, questions/requests: 1: Any general comments, including criticisms please (if couched politely). 2: I need a way to get the water in the reservoir and the loop to the tap to recirculate for two reasons: (a) to keep cold water at the top of the loop near the tap, and (b) to mix the water in the reservoir so that it is all cooled by the heat exchanger. I don't know if natural convection can do it for me, but I don't think so without allowing for more heat loss to set up a bigger temperature differential and I don't want that. Any good ideas here? I don't want or need a real pump here, just something to keep the water drifting along. Also this is something that I don't want to have in action when I am in a drawdown state (i.e.: water has been used and I haven't yet recovered from it), because I am hoping that the water will remain stratified in the reservoir (warm on top where it comes in and cold on the bottom where it is drawn). Note, by the way, that a checkvalve must be included between the loop return and the top of the reservoir to prevent the tap from trying to draw from the top instead of the bottom of the reservoir. The recirculation could be driven in the tap loop, or in the reservoir itself. Perhaps something magnetically coupled. Lack of wear points would be great. 3: Any good insulation ideas. Especially those that would allow for easy access for rework, repairs, etc. 4: Any good way to thermally isolate the heatsink / TEC from the cooling loop when the TEC is not running to avoid thermal loss. (One of my prime goals here is to make the 'idle efficiency' of this thing ridiculously good - just for the challenge of it. The construction here is currently planned as follows: (a) Large container with thermal 'slug' bolted/soldered/epoxied/welded to it, (b) Slug/TEC/waterblock stack bolted together, (c) waterblock plumbed into cooling loop. For details on the waterblock I have see: http://www.polarflo.com/index.asp?PageAction=VIEWPROD&ProdID=125&HS=1 5: I need cheap, simple temperature sensors to connect to PIC. Working temperature range from 0 C to 30 C or so. Accuracy not as important as precision, although it would be nice if they 'tracked' each other very well. Response time doesn't have to be particularly fast. Linearity is not a big deal if the correction function is known. They should be easy to attach to various components of system. They will all be within the 'enclosure' of this device, at most 0.75 m or so from the controlling PIC. [Hey, is that enough PIC related to let me change the topic of this message? :-)] 6: Good ideas for diffusers for the reservoir. The goal is to keep the water stratified when in the drawdown state so that consistently cold water is available for as long as possible. 7: Good ideas for bringing the cooling loop lines, and several termperature sensor lines, out of the reservoir without compromising the pressure integrity of the reservoir. 8: A simple way to sense when water is being drawn from the system. n: Any general comments, including criticisms please (if couched politely). =============== Addendum: Expected user interface: one tricolor LED and one 'touch button' (capacitively coupled input on sink deck). Touch button operates small valves mounted directly below tap to select between output of cooler and direct output of filter (no sense in filling the tea kettle with chilled water). LED is green when valve has selected room temperature water, blue for cold, red is reserved for potential future 'instant hot water' mode. Unit reverts to cold water mode after a timeout. If unit is in recovery mode the blue LED blinks, mostly on when nearly back to full capacity, lower duty cycle when further to go. Comments on this? Thanks!!! Bob Ammerman RAm Systems -- http://www.piclist.com PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist ( ITPOP---ITIITBS )** I top post on purpose --- I think it is the better scheme -- http://www.piclist.com PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist