Recent changes to this wiki:

cleanup
diff --git a/hardware_enclosure.mdwn b/hardware_enclosure.mdwn
index fecbe77..e17d985 100644
--- a/hardware_enclosure.mdwn
+++ b/hardware_enclosure.mdwn
@@ -41,10 +41,10 @@ The computer is not located in the enclosure, it's in the house about 40
 feet away, since it needed to be close to the charge controller and
 inverter.
 
-The custom circuit board simply connects up the relays and splits an
-incoming onewire bus out to 3 temperature probes. I wish it fit more.
-Note the added resistors on the onewire signal pins, which are to try to
-dampen echos and avoid some sensor read errors. 
+The custom circuit board simply connects up the relays' control lines, and
+splits an incoming onewire bus out to 3 temperature probes. I wish it fit
+more. Note the added resistors on the onewire signal pins, which are to try
+to dampen echos and avoid some sensor read errors. 
 
 I also had to adjust the resistor size on the relay boards to make the
 relays work reliably at this distance from the computer. Before fixing

cleanup
diff --git a/hardware_enclosure.mdwn b/hardware_enclosure.mdwn
index 9045098..fecbe77 100644
--- a/hardware_enclosure.mdwn
+++ b/hardware_enclosure.mdwn
@@ -24,6 +24,10 @@ wish it had more places for screws.
 It has two relays each powering their own AC outlet; I use the second one as a
 spare, or perhaps for a freezer or AC powered dump load. 
 
+Notice that the black AC wire is connected to the relay. That is the hot
+wire. You want to interrupt the hot wire, not the neutral wire. Consult
+your local code when dealing with AC wiring.
+
 Since this is outdoors, it needs to be GFCI protected; it plugs into the
 outdoor AC outlet next to it, which contains a GFCI. It would
 actually be more efficient to have put the GFCI outlet after the relay,

cleanup
diff --git a/hardware_enclosure.mdwn b/hardware_enclosure.mdwn
index 46e2978..9045098 100644
--- a/hardware_enclosure.mdwn
+++ b/hardware_enclosure.mdwn
@@ -1,4 +1,4 @@
-Building a fridge0 probably involves switching AC power with a relay under
+Building a fridge0 involves switching AC power with a relay under
 [[computer_control]]. High voltage needs a suitable enclosure for safety.
 Other things you may want to put in the enclosure include:
 

cleanup
diff --git a/hardware_enclosure.mdwn b/hardware_enclosure.mdwn
index c09ea19..46e2978 100644
--- a/hardware_enclosure.mdwn
+++ b/hardware_enclosure.mdwn
@@ -45,6 +45,10 @@ dampen echos and avoid some sensor read errors.
 I also had to adjust the resistor size on the relay boards to make the
 relays work reliably at this distance from the computer. Before fixing
 that, the relays would sometimes chatter or not turn on all the way.
+At one point I measured AC voltage split 50-50 
+between the NO and NC pins of the relay that was on half way, which seemed
+like it could lead to ugly failure modes. So make sure the relay turns on
+crisply and fully before plugging anything into it.
 
 
 ## others

cleanup
diff --git a/hardware_enclosure.mdwn b/hardware_enclosure.mdwn
index ac21b90..c09ea19 100644
--- a/hardware_enclosure.mdwn
+++ b/hardware_enclosure.mdwn
@@ -30,7 +30,8 @@ actually be more efficient to have put the GFCI outlet after the relay,
 because a GFCI outlet has some phantom load to run the protection circuit,
 perhaps as much as 3-5 watts. (But I don't know if that would be a good
 idea as far as the ground fault protection goes.) 
-As it is, that GFCI outlet is powered whenever the [[inverter]] is turned on.
+As it is, that GFCI outlet is powered whenever the [[inverter]] is turned on,
+even when fridge0 is not running.
 
 The computer is not located in the enclosure, it's in the house about 40
 feet away, since it needed to be close to the charge controller and

cleanup
diff --git a/hardware_enclosure.mdwn b/hardware_enclosure.mdwn
index 60cb881..ac21b90 100644
--- a/hardware_enclosure.mdwn
+++ b/hardware_enclosure.mdwn
@@ -25,7 +25,7 @@ It has two relays each powering their own AC outlet; I use the second one as a
 spare, or perhaps for a freezer or AC powered dump load. 
 
 Since this is outdoors, it needs to be GFCI protected; it plugs into the
-outdoor AC power socket next to it, which contains a GFCI. It would
+outdoor AC outlet next to it, which contains a GFCI. It would
 actually be more efficient to have put the GFCI outlet after the relay,
 because a GFCI outlet has some phantom load to run the protection circuit,
 perhaps as much as 3-5 watts. (But I don't know if that would be a good

update
diff --git a/hardware_enclosure.mdwn b/hardware_enclosure.mdwn
index 23e1bd8..60cb881 100644
--- a/hardware_enclosure.mdwn
+++ b/hardware_enclosure.mdwn
@@ -16,7 +16,8 @@ find in a variety of shapes and sizes. Recommended.
 
 ## Joey Hess's fridge0
 
-I used a fairly large enclosure, and wish it had been even larger.
+I used a fairly large enclosure, and wish it had been even larger. I also
+wish it had more places for screws.
 
 [[!img joeyh_enclosure.jpeg size=640x]]
 

others
diff --git a/hardware_enclosure.mdwn b/hardware_enclosure.mdwn
index fbb9a42..23e1bd8 100644
--- a/hardware_enclosure.mdwn
+++ b/hardware_enclosure.mdwn
@@ -43,3 +43,8 @@ dampen echos and avoid some sensor read errors.
 I also had to adjust the resistor size on the relay boards to make the
 relays work reliably at this distance from the computer. Before fixing
 that, the relays would sometimes chatter or not turn on all the way.
+
+
+## others
+
+Please add your hardware enclosure information here if you do a fridge0 build.
diff --git a/is_it_sunny.mdwn b/is_it_sunny.mdwn
index 882a933..a8af5b5 100644
--- a/is_it_sunny.mdwn
+++ b/is_it_sunny.mdwn
@@ -29,3 +29,7 @@ With some tuning I got my software to work well at finding appropriate
 times to run fridge0 in all conditions. Checking a photocell and PV voltage
 are two approaches I'd like to try as they could be simpler and need less tuning
 for the specific installation
+
+## others
+
+Please share how you deal with this in your fridge0 build.

link
diff --git a/computer_control.mdwn b/computer_control.mdwn
index c882d3c..8815c92 100644
--- a/computer_control.mdwn
+++ b/computer_control.mdwn
@@ -12,6 +12,8 @@ that the computer users to control power to the fridge,
 a waterproof temperature probe or two, and an interface to the
 [[charge controller]] to keep track of the current solar power situation.
 
+[[Hardware_enclosure]] has more on the hardware side of this.
+
 ## Joey Hess's fridge0
 
 I use a cubietruck computer which uses around 5 watts but is also my

update
diff --git a/hardware_enclosure.mdwn b/hardware_enclosure.mdwn
index 9907507..fbb9a42 100644
--- a/hardware_enclosure.mdwn
+++ b/hardware_enclosure.mdwn
@@ -38,6 +38,8 @@ inverter.
 The custom circuit board simply connects up the relays and splits an
 incoming onewire bus out to 3 temperature probes. I wish it fit more.
 Note the added resistors on the onewire signal pins, which are to try to
-dampen echos and avoid some sensor read errors. I also had to adjust the
-resistors size on the relay boards to make the relays work reliably at this
-distance from the computer.
+dampen echos and avoid some sensor read errors. 
+
+I also had to adjust the resistor size on the relay boards to make the
+relays work reliably at this distance from the computer. Before fixing
+that, the relays would sometimes chatter or not turn on all the way.

expand
diff --git a/hardware_enclosure.mdwn b/hardware_enclosure.mdwn
index 30c7438..9907507 100644
--- a/hardware_enclosure.mdwn
+++ b/hardware_enclosure.mdwn
@@ -21,9 +21,15 @@ I used a fairly large enclosure, and wish it had been even larger.
 [[!img joeyh_enclosure.jpeg size=640x]]
 
 It has two relays each powering their own AC outlet; I use the second one as a
-spare, or perhaps for a freezer or AC powered dump load. Since this is
-outdoors, it needs to be GFCI protected; it plugs into the outdoor AC power
-socket next to it, which contains a GFCI.
+spare, or perhaps for a freezer or AC powered dump load. 
+
+Since this is outdoors, it needs to be GFCI protected; it plugs into the
+outdoor AC power socket next to it, which contains a GFCI. It would
+actually be more efficient to have put the GFCI outlet after the relay,
+because a GFCI outlet has some phantom load to run the protection circuit,
+perhaps as much as 3-5 watts. (But I don't know if that would be a good
+idea as far as the ground fault protection goes.) 
+As it is, that GFCI outlet is powered whenever the [[inverter]] is turned on.
 
 The computer is not located in the enclosure, it's in the house about 40
 feet away, since it needed to be close to the charge controller and

update
diff --git a/charge_controller.mdwn b/charge_controller.mdwn
index e320fca..d802977 100644
--- a/charge_controller.mdwn
+++ b/charge_controller.mdwn
@@ -4,22 +4,7 @@ run the fridge.
 It's a good idea to pick a charge controller that can be connected to the
 [[computer|computer_control]] and tell it at least the current voltage and
 how many watts the solar panels are producing. Otherwise, you would need
-additional sensors for fridge0.
-
-If there's no load, solar panels won't produce any power no matter the
-weather, so it can be surprisingly difficult to tell when is a good time
-to run fridge0. Some approaches to deal with this:
-
-* Have some other load, like a water heating element, that can load the solar panels
-  when fridge0 is not.
-* Use a photocell to detect when it's sunny.
-* Look at the battery state of charge, eg if the battery is full assume
-  it's sunny.
-* Try running fridge0 and see if the solar panels start to produce enough
-  power, otherwise turning it off to avoid using the battery for more than
-  a minute.
-* Query the charge controller for the PV voltage, to detect when it's found
-  a maximum power point that will be sufficient to run fridge0.
+additional sensors for fridge0 to determine [[is_it_sunny]].
 
 ## Joey Hess's fridge0
 
@@ -37,18 +22,6 @@ full battery with 100 watts coming in may be a fairly cloudy day, or full
 sun with the battery in the absorb charging stage. (It can tell you when the battery
 is charged in the float stage.)
 
-My software mostly looks at the state of charge of the battery to tell when
-enough PV power is available to run fridge0. It also estimates power used
-by other parts of the house, to check if the solar panels are getting good
-power. I have experimented with a switchable dump load, which is sometimes
-useful to detect the absorb charge situation and eke out an extra 30
-minutes of fridge0 runtime.
-
-With some tuning I got my software to work well at finding appropriate
-times to run fridge0 in all conditions. Checking a photocell and PV voltage
-are two approaches I'd like to try as they could be simpler and need less tuning
-for the specific installation
-
 ## others
 
 Please add your controller information here if you do a fridge0 build.
diff --git a/hardware_enclosure.mdwn b/hardware_enclosure.mdwn
new file mode 100644
index 0000000..30c7438
--- /dev/null
+++ b/hardware_enclosure.mdwn
@@ -0,0 +1,37 @@
+Building a fridge0 probably involves switching AC power with a relay under
+[[computer_control]]. High voltage needs a suitable enclosure for safety.
+Other things you may want to put in the enclosure include:
+
+* AC power outlet(s)
+* Connections for temperature sensors
+* The embedded computer itself
+* And any other small hardware that needs to be close to the fridge
+
+You may want to locate your fridge0 on a porch or other location outside to
+avoid it needing to run as much in the winter. So it's a good idea to plan
+for a water resistent enclosure.
+
+A NEMA compliant, waterproof, locking enclosure is easy and inexpensive to
+find in a variety of shapes and sizes. Recommended.
+
+## Joey Hess's fridge0
+
+I used a fairly large enclosure, and wish it had been even larger.
+
+[[!img joeyh_enclosure.jpeg size=640x]]
+
+It has two relays each powering their own AC outlet; I use the second one as a
+spare, or perhaps for a freezer or AC powered dump load. Since this is
+outdoors, it needs to be GFCI protected; it plugs into the outdoor AC power
+socket next to it, which contains a GFCI.
+
+The computer is not located in the enclosure, it's in the house about 40
+feet away, since it needed to be close to the charge controller and
+inverter.
+
+The custom circuit board simply connects up the relays and splits an
+incoming onewire bus out to 3 temperature probes. I wish it fit more.
+Note the added resistors on the onewire signal pins, which are to try to
+dampen echos and avoid some sensor read errors. I also had to adjust the
+resistors size on the relay boards to make the relays work reliably at this
+distance from the computer.
diff --git a/hardware_enclosure/joeyh_enclosure.jpeg b/hardware_enclosure/joeyh_enclosure.jpeg
new file mode 100644
index 0000000..32ab6d6
Binary files /dev/null and b/hardware_enclosure/joeyh_enclosure.jpeg differ
diff --git a/index.mdwn b/index.mdwn
index d76612b..b6794ff 100644
--- a/index.mdwn
+++ b/index.mdwn
@@ -34,6 +34,13 @@ make less sense than they used to. A kilowatt of solar panels provides
 enough power to run a conventional fridge on even most cloudy days, and
 costs less than a commercial offgrid fridge.
 
+## guides
+
+Guides for different parts of the process of building your own fridge0:
+
+* [[hardware_enclosure]]
+* [[is_it_sunny]]
+
 ## Joey Hess's fridge0
 
 [Joey Hess](https://joeyh.name/) developed the fridge0 concept and the
diff --git a/is_it_sunny.mdwn b/is_it_sunny.mdwn
new file mode 100644
index 0000000..882a933
--- /dev/null
+++ b/is_it_sunny.mdwn
@@ -0,0 +1,31 @@
+If there's no load, solar panels won't produce any power no matter the
+weather, so it can be surprisingly difficult to tell when is a good time
+to run fridge0. Some approaches to deal with this:
+
+* Have some other load, like a water heating element, that can load the solar panels
+  when fridge0 is not.
+* Use a photocell to detect when it's sunny.
+* Look at the battery state of charge, eg if the battery is full assume
+  it's sunny. Or use the battery charging as a load.
+* Try running fridge0 and see if the solar panels start to produce enough
+  power, otherwise turning it off to avoid using the battery for more than
+  a minute.
+* Query the charge controller for the PV voltage, to detect when it's found
+  a maximum power point that will be sufficient to run fridge0.
+
+## Joey Hess's fridge0
+
+My software mostly looks at the state of charge of the battery to tell when
+enough PV power is available to run fridge0. It also estimates power used
+by other parts of the house, to check if the solar panels are getting good
+power.
+
+Since my [[charge_controller]] does not report when the battery is in the
+absorb charge stage, I have experimented with a switchable dump load, which 
+is sometimes useful to detect the absorb charge situation and eke out an extra 30
+minutes of fridge0 runtime.
+
+With some tuning I got my software to work well at finding appropriate
+times to run fridge0 in all conditions. Checking a photocell and PV voltage
+are two approaches I'd like to try as they could be simpler and need less tuning
+for the specific installation

expand
diff --git a/charge_controller.mdwn b/charge_controller.mdwn
index a99f710..e320fca 100644
--- a/charge_controller.mdwn
+++ b/charge_controller.mdwn
@@ -31,11 +31,23 @@ It also has the advantage of having a data port, so the computer can
 read a lot of information from it, including the current voltage, watts
 being produced by the solar panel, etc.
 
-My software uses a combination of several of the approaches listed above to
-determine when there's enough power to run fridge0. With some tuning I got
-that to work well. Checking a photocell and PV voltage are two approaches
-I'd like to try as they could be simpler and need less tuning for the
-specific installation.
+One very annoying thing about this charge controller is it does not
+distinguish when the battery is in the absorb charging stage, so a not yet
+full battery with 100 watts coming in may be a fairly cloudy day, or full
+sun with the battery in the absorb charging stage. (It can tell you when the battery
+is charged in the float stage.)
+
+My software mostly looks at the state of charge of the battery to tell when
+enough PV power is available to run fridge0. It also estimates power used
+by other parts of the house, to check if the solar panels are getting good
+power. I have experimented with a switchable dump load, which is sometimes
+useful to detect the absorb charge situation and eke out an extra 30
+minutes of fridge0 runtime.
+
+With some tuning I got my software to work well at finding appropriate
+times to run fridge0 in all conditions. Checking a photocell and PV voltage
+are two approaches I'd like to try as they could be simpler and need less tuning
+for the specific installation
 
 ## others
 

reword
diff --git a/charge_controller.mdwn b/charge_controller.mdwn
index 4e757ff..a99f710 100644
--- a/charge_controller.mdwn
+++ b/charge_controller.mdwn
@@ -31,11 +31,11 @@ It also has the advantage of having a data port, so the computer can
 read a lot of information from it, including the current voltage, watts
 being produced by the solar panel, etc.
 
-I use a combination of most of the strategies above to determine when
-there's enough power to run it. With some tuning I got that to work well.
-Checking a photocell and PV voltage are two approaches I'd like to try
-as they could be simpler and need less tuning for the specific
-installation.
+My software uses a combination of several of the approaches listed above to
+determine when there's enough power to run fridge0. With some tuning I got
+that to work well. Checking a photocell and PV voltage are two approaches
+I'd like to try as they could be simpler and need less tuning for the
+specific installation.
 
 ## others
 

expand
diff --git a/charge_controller.mdwn b/charge_controller.mdwn
index 9f1837a..4e757ff 100644
--- a/charge_controller.mdwn
+++ b/charge_controller.mdwn
@@ -6,6 +6,21 @@ It's a good idea to pick a charge controller that can be connected to the
 how many watts the solar panels are producing. Otherwise, you would need
 additional sensors for fridge0.
 
+If there's no load, solar panels won't produce any power no matter the
+weather, so it can be surprisingly difficult to tell when is a good time
+to run fridge0. Some approaches to deal with this:
+
+* Have some other load, like a water heating element, that can load the solar panels
+  when fridge0 is not.
+* Use a photocell to detect when it's sunny.
+* Look at the battery state of charge, eg if the battery is full assume
+  it's sunny.
+* Try running fridge0 and see if the solar panels start to produce enough
+  power, otherwise turning it off to avoid using the battery for more than
+  a minute.
+* Query the charge controller for the PV voltage, to detect when it's found
+  a maximum power point that will be sufficient to run fridge0.
+
 ## Joey Hess's fridge0
 
 I use an Epsolar Tracer BN.
@@ -16,6 +31,12 @@ It also has the advantage of having a data port, so the computer can
 read a lot of information from it, including the current voltage, watts
 being produced by the solar panel, etc.
 
+I use a combination of most of the strategies above to determine when
+there's enough power to run it. With some tuning I got that to work well.
+Checking a photocell and PV voltage are two approaches I'd like to try
+as they could be simpler and need less tuning for the specific
+installation.
+
 ## others
 
 Please add your controller information here if you do a fridge0 build.

Added a comment: Water additives
diff --git a/thermal_mass/comment_1_9c4733c93bad2e89a812b719b833713b._comment b/thermal_mass/comment_1_9c4733c93bad2e89a812b719b833713b._comment
new file mode 100644
index 0000000..c14c94e
--- /dev/null
+++ b/thermal_mass/comment_1_9c4733c93bad2e89a812b719b833713b._comment
@@ -0,0 +1,13 @@
+[[!comment format=mdwn
+ username="erice@56ee96f0816ee15978e942b692f40099fc6a0a7f"
+ nickname="erice"
+ avatar="http://cdn.libravatar.org/avatar/0403e6ce9731e540eb0c470011d5726e"
+ subject="Water additives"
+ date="2018-12-27T21:13:56Z"
+ content="""
+Water additives can dramatically alter the amount of thermal capacity water has.  I'm a huge fan of the Cooler Shock ice packs - they ship them dry and you add water to them.  The 0c ones really are a replacement for dry ice if you freeze them in a sufficiently cold freezer for at least 48 hours.
+
+https://coolershock.com
+
+I dunno if they would be as effective at refrigerator temperatures since they are aimed for frozen use, but it would be interesting to see if there are different additives that might help enhance the inherit capabilities of water.
+"""]]

expand
diff --git a/battery_bank.mdwn b/battery_bank.mdwn
index 31b9396..17946a4 100644
--- a/battery_bank.mdwn
+++ b/battery_bank.mdwn
@@ -30,6 +30,12 @@ bank any for fridge0. It consists of 4 deep cycle golf cart batteries
 I'm pretty sure all my fridge0 needs is a single small deep cycle battery,
 around 100-150 AH capacity.
 
+My control software for fridge0 prioritizes charging the battery bank over
+running the fridge, unless the fridge is getting too warm. This way
+battery power is available for my other needs. Balancing these needs was
+one of the harder parts of the project; it involves understanding the state
+of charge of the battery and how much potential solar power is available.
+
 ## others
 
 Please add your information here if you do a fridge0 build.

typo
diff --git a/battery_bank.mdwn b/battery_bank.mdwn
index 8fd0021..31b9396 100644
--- a/battery_bank.mdwn
+++ b/battery_bank.mdwn
@@ -22,7 +22,7 @@ power that needs to be stored is relatively small.
 
 ## Joey Hess's fridge0
 
-My house has a battery bank used for other things that fridge0 (lights,
+My house has a battery bank used for other things than fridge0 (lights,
 laptops, internet), so I use that. I didn't need to increase the battery
 bank any for fridge0. It consists of 4 deep cycle golf cart batteries
 (model #SLIGC110), 860AH capacity. Total cost was ~$500.

remove unverified bit
diff --git a/inverter.mdwn b/inverter.mdwn
index 0a59c16..0ee11ee 100644
--- a/inverter.mdwn
+++ b/inverter.mdwn
@@ -16,8 +16,7 @@ the hours that the fridge is not running.
 
 One option is to use a relay to cut DC power to the inverter when fridge0 is
 not running, but some inverters may need a button pressed to power back on,
-preventing doing that. Also, the surge of DC power when the relay is turned
-on can cause arcing in some circumstances.
+preventing doing that.
 
 ## Joey Hess's fridge0
 

expand
diff --git a/computer_control.mdwn b/computer_control.mdwn
index 6a8479f..c882d3c 100644
--- a/computer_control.mdwn
+++ b/computer_control.mdwn
@@ -7,7 +7,8 @@ Warning when the temperature goes outside of the
 [[safe temperature range|temperature_range]] is also a good idea.
 
 This could be done in many many ways; most will probably involve a small
-embedded computer such as a raspberry pi, that runs all the time, a relay,
+embedded computer such as a raspberry pi, that runs all the time, a relay
+that the computer users to control power to the fridge,
 a waterproof temperature probe or two, and an interface to the
 [[charge controller]] to keep track of the current solar power situation.
 

typo
diff --git a/photovoltaic_panels.mdwn b/photovoltaic_panels.mdwn
index 2d81a29..8de59a9 100644
--- a/photovoltaic_panels.mdwn
+++ b/photovoltaic_panels.mdwn
@@ -20,7 +20,7 @@ fuse boxes, and the [[charge controller]].
 
 Often on a cloudy day, the panels will still produce 150-250 watts,
 so still enough power to run the fridge.
-When the sun in high in spring/summer, it can run even during
+When the sun is high in spring/summer, it can run even during
 light rains.
 
 ## others

fix number
diff --git a/photovoltaic_panels.mdwn b/photovoltaic_panels.mdwn
index 3c74164..2d81a29 100644
--- a/photovoltaic_panels.mdwn
+++ b/photovoltaic_panels.mdwn
@@ -13,7 +13,7 @@ possible.
 
 ## Joey Hess's fridge0
 
-I have a 1500 watt PV array. It's used to power my whole house, not only fridge0.
+I have a 1040 watt PV array. It's used to power my whole house, not only fridge0.
 
 It cost $2000 self-installed, including panels, mounting rails, cables,
 fuse boxes, and the [[charge controller]].

update
diff --git a/temperature_range.mdwn b/temperature_range.mdwn
index 8999f0d..0bb130e 100644
--- a/temperature_range.mdwn
+++ b/temperature_range.mdwn
@@ -40,3 +40,8 @@ a little above 4C are not unusual; it seems to run between 4 and 5.8C.
 [[!img merlins.png size=640x]]
 
 [[!img merlins_year.png size=640x]]
+
+## others
+
+Please add your fridge's temperature data here if you have any.
+Including from conventional fridges.

oopd
diff --git a/temperature_range.mdwn b/temperature_range.mdwn
index e05986c..8999f0d 100644
--- a/temperature_range.mdwn
+++ b/temperature_range.mdwn
@@ -39,4 +39,4 @@ a little above 4C are not unusual; it seems to run between 4 and 5.8C.
 
 [[!img merlins.png size=640x]]
 
-[[!img merlinsyear.png size=640x]]
+[[!img merlins_year.png size=640x]]

update
diff --git a/temperature_range.mdwn b/temperature_range.mdwn
index 3ac8fd3..e05986c 100644
--- a/temperature_range.mdwn
+++ b/temperature_range.mdwn
@@ -7,16 +7,9 @@ parts of a fridge, so while it claims to keep the temperature below 4C, it
 might run warmer on a top shelf or in shelves on the door. And fridge's
 have a temperature control, which also varies the temperature in some way.
 
-[This data](http://graphs.merlins.org/graphs/graph_view.php) 
-from Marc Merlin's conventional fridge suggests that temperatures
-a little above 4C are not unusual; it seems to run between 5.2 and 5.8C.
-
-[[!img merlins.png size=640x]]
-
-Your conventional fridge may not really be below 4C either. In fact, it
-doesn't matter a great deal; the 4C value is not a magic safe range despite
-what the USDA will tell you; it's just a compromise point that slows down food
-spoilage to a reasonable extent. There is some sort of a curve involved.
+In fact, the 4C value is not a magic safe range despite what the USDA will
+tell you; it's just a compromise point that slows down food spoilage to a
+reasonable extent. There is some sort of a curve involved.
 
 Fridge0 aims for 0.5C to 5C, but it may sometimes heat up to 8C on a rainy
 week, and that's no real cause for alarm. Milk typically still 
@@ -37,3 +30,13 @@ Here is temperature data from the first 8 months of operation.
 The water temp in those graphs is the temperature of the [[thermal mass]];
 different parts of the fridge range in temperature between that and the
 main temp.
+
+## Marc Merlin's conventional fridge
+
+[This data](http://graphs.merlins.org/graphs/graph_view.php) 
+from Marc Merlin's conventional fridge suggests that temperatures
+a little above 4C are not unusual; it seems to run between 4 and 5.8C.
+
+[[!img merlins.png size=640x]]
+
+[[!img merlinsyear.png size=640x]]
diff --git a/temperature_range/merlins_year.png b/temperature_range/merlins_year.png
new file mode 100644
index 0000000..3043b8b
Binary files /dev/null and b/temperature_range/merlins_year.png differ

update
diff --git a/index.mdwn b/index.mdwn
index e66e196..d76612b 100644
--- a/index.mdwn
+++ b/index.mdwn
@@ -6,7 +6,8 @@ Fridge0 consists of a [[standard chest freezer|chest_freezer]],
 an added [[thermal mass]], an [[inverter]], and [[computer control]].
 It ties into the typical offfgrid system of a 
 [[solar charge controller|charge controller]],
-[[battery bank]], and [[photovoltaic panels]].
+[[battery bank]], and [[photovoltaic panels]] to maintain a safe
+[[temperature range]].
 
 The battery bank is a large part of the cost of a typical offgrid fridge
 installation. It needs to be sized to run the fridge overnight, as well as
diff --git a/temperature_range.mdwn b/temperature_range.mdwn
index 236f8c9..3ac8fd3 100644
--- a/temperature_range.mdwn
+++ b/temperature_range.mdwn
@@ -32,5 +32,8 @@ solar panels.
 Here is temperature data from the first 8 months of operation.
 (Leaving out a period of data collection failures.)
 
-[[!img joeyh_dataset1.png]] [[!img joeyh_dataset2.png]]
+[[!img joeyh_dataset1.png size=640x]] [[!img joeyh_dataset2.png size=640x]]
 
+The water temp in those graphs is the temperature of the [[thermal mass]];
+different parts of the fridge range in temperature between that and the
+main temp.

data
diff --git a/temperature_range.mdwn b/temperature_range.mdwn
index 93144f8..236f8c9 100644
--- a/temperature_range.mdwn
+++ b/temperature_range.mdwn
@@ -26,3 +26,11 @@ If this seems scary, you can always add more solar panels if your fridge0
 build doesn't run cool enough for you. Though note that your
 [[charge controller]] has a limit on the amount of power it can handle from
 solar panels.
+
+## Joey Hess's fridge0
+
+Here is temperature data from the first 8 months of operation.
+(Leaving out a period of data collection failures.)
+
+[[!img joeyh_dataset1.png]] [[!img joeyh_dataset2.png]]
+
diff --git a/temperature_range/joeyh_dataset1.png b/temperature_range/joeyh_dataset1.png
new file mode 100644
index 0000000..fae706c
Binary files /dev/null and b/temperature_range/joeyh_dataset1.png differ
diff --git a/temperature_range/joeyh_dataset2.png b/temperature_range/joeyh_dataset2.png
new file mode 100644
index 0000000..35e0568
Binary files /dev/null and b/temperature_range/joeyh_dataset2.png differ

oopd
diff --git a/thermal_mass.mdwn b/thermal_mass.mdwn
index 1c405ad..034a222 100644
--- a/thermal_mass.mdwn
+++ b/thermal_mass.mdwn
@@ -73,7 +73,7 @@ The thermal mass is 17 gallons in size.
 It uses around 1/3rd of the space in my 8.7 cubic foot
 [[chest_freezer]].
 
-[[!img joeyh_fridge_front.jpg size=640x]]
+[[!img joeyh_fridge_front.jpeg size=640x]]
 
 The water tubs don't cover the whole bottom of the fridge, so there's a "cold
 well" next to them. I put milk, down there, and cold drinks. Meats sit on

oopd
diff --git a/thermal_mass.mdwn b/thermal_mass.mdwn
index 710296a..1c405ad 100644
--- a/thermal_mass.mdwn
+++ b/thermal_mass.mdwn
@@ -73,7 +73,7 @@ The thermal mass is 17 gallons in size.
 It uses around 1/3rd of the space in my 8.7 cubic foot
 [[chest_freezer]].
 
-[[joeyh_fridge_front.jpg size=640x]]
+[[!img joeyh_fridge_front.jpg size=640x]]
 
 The water tubs don't cover the whole bottom of the fridge, so there's a "cold
 well" next to them. I put milk, down there, and cold drinks. Meats sit on

add
diff --git a/thermal_mass.mdwn b/thermal_mass.mdwn
index cc5e7b5..710296a 100644
--- a/thermal_mass.mdwn
+++ b/thermal_mass.mdwn
@@ -73,6 +73,8 @@ The thermal mass is 17 gallons in size.
 It uses around 1/3rd of the space in my 8.7 cubic foot
 [[chest_freezer]].
 
+[[joeyh_fridge_front.jpg size=640x]]
+
 The water tubs don't cover the whole bottom of the fridge, so there's a "cold
 well" next to them. I put milk, down there, and cold drinks. Meats sit on
 the shelf above the compressor, below the top of the thermal mass, where
diff --git a/thermal_mass/joeyh_fridge_front.jpeg b/thermal_mass/joeyh_fridge_front.jpeg
new file mode 100644
index 0000000..dc40498
Binary files /dev/null and b/thermal_mass/joeyh_fridge_front.jpeg differ

reorg
diff --git a/index.mdwn b/index.mdwn
index 93b2751..e66e196 100644
--- a/index.mdwn
+++ b/index.mdwn
@@ -2,6 +2,12 @@ Fridge0 is a design for an offgrid, solar powered fridge, with no battery bank.
 Using an inexpensive chest freezer with a few modifications, the fridge
 retains cold overnight and through rainy periods.
 
+Fridge0 consists of a [[standard chest freezer|chest_freezer]],
+an added [[thermal mass]], an [[inverter]], and [[computer control]].
+It ties into the typical offfgrid system of a 
+[[solar charge controller|charge controller]],
+[[battery bank]], and [[photovoltaic panels]].
+
 The battery bank is a large part of the cost of a typical offgrid fridge
 installation. It needs to be sized to run the fridge overnight, as well as
 for several days of poor weather. Cheaper batteries only last 3-5 years,
@@ -27,12 +33,6 @@ make less sense than they used to. A kilowatt of solar panels provides
 enough power to run a conventional fridge on even most cloudy days, and
 costs less than a commercial offgrid fridge.
 
-Fridge0 consists of a [[standard chest freezer|chest_freezer]],
-an added [[thermal mass]], an [[inverter]], and [[computer control]].
-It ties into the typical offfgrid system of a 
-[[solar charge controller|charge controller]],
-[[battery bank]], and [[photovoltaic panels]].
-
 ## Joey Hess's fridge0
 
 [Joey Hess](https://joeyh.name/) developed the fridge0 concept and the

wording
diff --git a/computer_control.mdwn b/computer_control.mdwn
index 83fe879..6a8479f 100644
--- a/computer_control.mdwn
+++ b/computer_control.mdwn
@@ -23,7 +23,7 @@ which use the 1-wire bus.
 to get the cubietruck to enable the 1-wire bus.)
 
 I have one temperature probe in the air near the top of the fridge, so I
-get near to the highest temperature inside it, and another inside the
+measure the highest temperature inside it, and another inside the
 [[thermal mass]] of water.
 
 For the relay, I use 

scale
diff --git a/temperature_range.mdwn b/temperature_range.mdwn
index 6bd0cfe..93144f8 100644
--- a/temperature_range.mdwn
+++ b/temperature_range.mdwn
@@ -11,7 +11,7 @@ have a temperature control, which also varies the temperature in some way.
 from Marc Merlin's conventional fridge suggests that temperatures
 a little above 4C are not unusual; it seems to run between 5.2 and 5.8C.
 
-[[!img merlins.png]]
+[[!img merlins.png size=640x]]
 
 Your conventional fridge may not really be below 4C either. In fact, it
 doesn't matter a great deal; the 4C value is not a magic safe range despite

add
diff --git a/temperature_range.mdwn b/temperature_range.mdwn
new file mode 100644
index 0000000..6bd0cfe
--- /dev/null
+++ b/temperature_range.mdwn
@@ -0,0 +1,28 @@
+The safe temperature range for a fridge is typically considered to be below
+4C (40F). Of course, it also needs to stay above freezing to avoid damage
+to food.
+
+There is often a considerable variation in temperature between different
+parts of a fridge, so while it claims to keep the temperature below 4C, it
+might run warmer on a top shelf or in shelves on the door. And fridge's
+have a temperature control, which also varies the temperature in some way.
+
+[This data](http://graphs.merlins.org/graphs/graph_view.php) 
+from Marc Merlin's conventional fridge suggests that temperatures
+a little above 4C are not unusual; it seems to run between 5.2 and 5.8C.
+
+[[!img merlins.png]]
+
+Your conventional fridge may not really be below 4C either. In fact, it
+doesn't matter a great deal; the 4C value is not a magic safe range despite
+what the USDA will tell you; it's just a compromise point that slows down food
+spoilage to a reasonable extent. There is some sort of a curve involved.
+
+Fridge0 aims for 0.5C to 5C, but it may sometimes heat up to 8C on a rainy
+week, and that's no real cause for alarm. Milk typically still 
+lasts for a week in an 8C fridge.
+
+If this seems scary, you can always add more solar panels if your fridge0
+build doesn't run cool enough for you. Though note that your
+[[charge controller]] has a limit on the amount of power it can handle from
+solar panels.

temperature range
diff --git a/computer_control.mdwn b/computer_control.mdwn
index af7133f..83fe879 100644
--- a/computer_control.mdwn
+++ b/computer_control.mdwn
@@ -3,6 +3,9 @@ determine when there's enough solar power to run the fridge, and turn it
 on, powering it off when power is lost. And it should avoid freezing the
 food in the fridge, so it needs to sense the temperature inside it as well.
 
+Warning when the temperature goes outside of the 
+[[safe temperature range|temperature_range]] is also a good idea.
+
 This could be done in many many ways; most will probably involve a small
 embedded computer such as a raspberry pi, that runs all the time, a relay,
 a waterproof temperature probe or two, and an interface to the
diff --git a/temperature_range/merlins.png b/temperature_range/merlins.png
new file mode 100644
index 0000000..7726758
Binary files /dev/null and b/temperature_range/merlins.png differ
diff --git a/thermal_mass.mdwn b/thermal_mass.mdwn
index 912bdfe..cc5e7b5 100644
--- a/thermal_mass.mdwn
+++ b/thermal_mass.mdwn
@@ -13,6 +13,7 @@ Fridge0 often stops running before the last meal of the day, and so it will
 often be opened several times and perhaps some warm leftovers put into it.
 This causes the air in the fridge to spike up, but its temperature
 will then trail back off as the thermal mass cools it back down.
+Such spikes and recovery can be seen in the graphs above and below.
 
 [[!img example2.png size=640x]]
 
@@ -20,7 +21,8 @@ When the fridge is unable to run for a period of several days, the thermal
 mass slowly warms up, moderating the temperature of the air and groceries
 in the fridge. In the above example, the fridge didn't run for 4 days,
 and the thermal mass warmed up by 2C over that time, and kept the food
-safely cool. (Temperatures outside the fridge ranged from 4-13C.)
+[[safely cool|temperature_range]]. 
+(Temperatures outside the fridge ranged from 4-13C.)
 
 The other thing thermal mass can do, during the winter, is keep the food
 from freezing when the fridge is located outdoors, by helping to average

scal
diff --git a/thermal_mass.mdwn b/thermal_mass.mdwn
index 75aa938..912bdfe 100644
--- a/thermal_mass.mdwn
+++ b/thermal_mass.mdwn
@@ -7,14 +7,14 @@ pulls down the temperature of the thermal mass. In this example, the fridge
 ran for 2 hours over the course of the day, and the thermal mass was cooled
 down by 1.3C.
 
-[[!img example1.png]]
+[[!img example1.png size=640x]]
 
 Fridge0 often stops running before the last meal of the day, and so it will
 often be opened several times and perhaps some warm leftovers put into it.
 This causes the air in the fridge to spike up, but its temperature
 will then trail back off as the thermal mass cools it back down.
 
-[[!img example2.png]]
+[[!img example2.png size=640x]]
 
 When the fridge is unable to run for a period of several days, the thermal
 mass slowly warms up, moderating the temperature of the air and groceries

wordin
diff --git a/index.mdwn b/index.mdwn
index 493af00..93b2751 100644
--- a/index.mdwn
+++ b/index.mdwn
@@ -36,8 +36,6 @@ It ties into the typical offfgrid system of a
 ## Joey Hess's fridge0
 
 [Joey Hess](https://joeyh.name/) developed the fridge0 concept and the
-first fridge0.
+first fridge0. It's been working well for almost a full year now. 
 
-I have been using such a fridge for almost a full year now, and
-it's worked well for me. This site is a wiki to document 
-my fridge and similar builds.
+This site is a wiki to document that fridge and similar builds.

typo
diff --git a/index.mdwn b/index.mdwn
index ba987fc..493af00 100644
--- a/index.mdwn
+++ b/index.mdwn
@@ -29,7 +29,7 @@ costs less than a commercial offgrid fridge.
 
 Fridge0 consists of a [[standard chest freezer|chest_freezer]],
 an added [[thermal mass]], an [[inverter]], and [[computer control]].
-It this ties into the typical offfgrid system of a 
+It ties into the typical offfgrid system of a 
 [[solar charge controller|charge controller]],
 [[battery bank]], and [[photovoltaic panels]].
 

initial commit
diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..eecda60
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1 @@
+/.ikiwiki

intial
diff --git a/battery_bank.mdwn b/battery_bank.mdwn
new file mode 100644
index 0000000..8fd0021
--- /dev/null
+++ b/battery_bank.mdwn
@@ -0,0 +1,35 @@
+Fridge0 does not need a big battery bank, since it stores cold
+in its [[thermal_mass]] and runs only when solar power is available.
+
+But, at least a mimimal battery is needed to run the [[computer_control]].
+Probably on the order of 5 watts for a computer like a raspberry pi; this
+could be reduced more with a more embedded computer like an arduino.
+
+Enough battery to run the fridge for a minute or two is also needed.
+Consider what happens when the fridge is running, and the sun goes behind a
+rain cloud. Suddenly, there's not enough power to run it, but it's still turned
+on. The [[inverter]] will try to run with whatever small solar power is still
+coming in, but it's not enough. Some inverters might manage a clean power
+off, probably accompanied with loud beeping. Other inverters might fail in
+more interesting ways.
+
+To deal with this situation, the computer needs to check the incoming solar power
+frequently, and power off the fridge if there's not enough. The battery is 
+needed to keep the inverter and fridge running until that happens.
+
+A supercapacitor might also work, instead of a battery, since the total
+power that needs to be stored is relatively small.
+
+## Joey Hess's fridge0
+
+My house has a battery bank used for other things that fridge0 (lights,
+laptops, internet), so I use that. I didn't need to increase the battery
+bank any for fridge0. It consists of 4 deep cycle golf cart batteries
+(model #SLIGC110), 860AH capacity. Total cost was ~$500.
+
+I'm pretty sure all my fridge0 needs is a single small deep cycle battery,
+around 100-150 AH capacity.
+
+## others
+
+Please add your information here if you do a fridge0 build.
diff --git a/charge_controller.mdwn b/charge_controller.mdwn
new file mode 100644
index 0000000..9f1837a
--- /dev/null
+++ b/charge_controller.mdwn
@@ -0,0 +1,21 @@
+Fridge0 could be used with any solar charge controller that's big enough to
+run the fridge.
+
+It's a good idea to pick a charge controller that can be connected to the
+[[computer|computer_control]] and tell it at least the current voltage and
+how many watts the solar panels are producing. Otherwise, you would need
+additional sensors for fridge0.
+
+## Joey Hess's fridge0
+
+I use an Epsolar Tracer BN.
+
+This is an inexpensive MPPT charge controller.
+
+It also has the advantage of having a data port, so the computer can
+read a lot of information from it, including the current voltage, watts
+being produced by the solar panel, etc.
+
+## others
+
+Please add your controller information here if you do a fridge0 build.
diff --git a/chest_freezer.mdwn b/chest_freezer.mdwn
new file mode 100644
index 0000000..b54296a
--- /dev/null
+++ b/chest_freezer.mdwn
@@ -0,0 +1,41 @@
+A chest freezer is used for fridge0 for the same reason that
+freezer conversions are often used for offgrid fridges: Efficiency.
+Not dumping cold on the floor when the door is opened is the usual
+explanation.
+
+There are many articles on the web about chest freezer conversions,
+how to organize groceries for easy access, etc.
+
+Any standard AC powered chest freezer can be used for fridge0. 
+It could even be a used freezer, but that could be less efficient.
+
+Bear in mind that the
+added [[thermal_mass]] used by fridge0 will reduce the storage space.
+
+## location
+
+It's a good idea to put the fridge outside or in an unheated building
+to reduce power usage in the winter. But check if the manufacturer says it's
+safe to do this. Some manufacturers do not warrantee chest freezers used in
+a below freezing environment, for reasons that are unclear.
+
+## Joey Hess's fridge0
+
+I bought a new Frigidare model #FFFC09M1RW, 8.7 cubic feet interior size.
+
+It's rated to consume 138 watts (1.2A at 115V). I measured it to use 120
+watts.
+
+It's been working for me very well. I keep it on the porch, which is
+not ideal in summer, but there's plenty of power then, and has proven to
+be a good idea in winter since it often doesn't need to run at all then.
+
+One initially surprising thing is that if the fridge is running when power
+is disconnected and then reconnected, the fridge won't immediately start
+back up. Seems that it detects the motor/compressor is too warm, and
+they need to cool down. This doesn't cause any problems in the fridge0
+build, though.
+
+## others
+
+Please add your chest freezer information here if you do a fridge0 build.
diff --git a/computer_control.mdwn b/computer_control.mdwn
new file mode 100644
index 0000000..af7133f
--- /dev/null
+++ b/computer_control.mdwn
@@ -0,0 +1,43 @@
+The brains of fridge0 is what really makes it possible. It needs to
+determine when there's enough solar power to run the fridge, and turn it
+on, powering it off when power is lost. And it should avoid freezing the
+food in the fridge, so it needs to sense the temperature inside it as well.
+
+This could be done in many many ways; most will probably involve a small
+embedded computer such as a raspberry pi, that runs all the time, a relay,
+a waterproof temperature probe or two, and an interface to the
+[[charge controller]] to keep track of the current solar power situation.
+
+## Joey Hess's fridge0
+
+I use a cubietruck computer which uses around 5 watts but is also my
+house's router and server.
+
+I've had good luck with waterproof DS18B20 temperature sensors,
+which use the 1-wire bus. 
+[This one from Sparkfun is good](https://www.sparkfun.com/products/11050)
+(I created <http://joeyh.name/blog/entry/easy-peasy-devicetree-squeezy/>
+to get the cubietruck to enable the 1-wire bus.)
+
+I have one temperature probe in the air near the top of the fridge, so I
+get near to the highest temperature inside it, and another inside the
+[[thermal mass]] of water.
+
+For the relay, I use 
+[this kit from Sparkfun](https://www.sparkfun.com/products/13815), which
+switches the AC power from the [[inverter]] to the fridge on and off.
+
+The computer talks to my [[charge controller]] using
+the software at <https://github.com/kasbert/epsolar-tracer>
+
+The software to integrate all this together in fridge0 is in my git repository at
+<https://git.joeyh.name/index.cgi/joey/homepower.git/>. As well as
+controlling fridge0 it generates graphs and also automates other parts of
+my house. If you want to adapt it to your build, I can probably help out.
+It uses haskell and FRP, as explained in 
+<https://joeyh.name/blog/entry/my_haskell_controlled_offgrid_fridge/>
+
+## others
+
+Please add your setup here if you do a fridge0 build.
+
diff --git a/index.mdwn b/index.mdwn
new file mode 100644
index 0000000..ba987fc
--- /dev/null
+++ b/index.mdwn
@@ -0,0 +1,43 @@
+Fridge0 is a design for an offgrid, solar powered fridge, with no battery bank.
+Using an inexpensive chest freezer with a few modifications, the fridge
+retains cold overnight and through rainy periods.
+
+The battery bank is a large part of the cost of a typical offgrid fridge
+installation. It needs to be sized to run the fridge overnight, as well as
+for several days of poor weather. Cheaper batteries only last 3-5 years,
+and longer lasting batteries are correspondingly expensive; either way a
+battery bank for an offgrid fridge is extremely expensive over the lifetime
+of the friddge.
+
+By storing solar power in the form of cold, fridge0 avoids the battery bank
+expense and environmental footprint. The only battery power fridge0 needs
+is enough to turn it off cleanly when the solar panels stop producing -- a few
+minutes of power instead of days -- and a small amount for its computer
+control.
+
+There are offgrid fridges produced by several manufacturers, which are
+designed for 12v solar power, but these still may need a battery bank,
+and are much more expensive than conventional equipment due to being very
+well insulated using features like vacuum bottles. These offgrid fridges
+were designed when solar panels were much more expensive than they are
+today.
+
+With modern cheap and efficient solar panels, commercial offgrid fridges
+make less sense than they used to. A kilowatt of solar panels provides
+enough power to run a conventional fridge on even most cloudy days, and
+costs less than a commercial offgrid fridge.
+
+Fridge0 consists of a [[standard chest freezer|chest_freezer]],

(Diff truncated)