diy solar

diy solar

Dump all excess PV to air heater?

please help me understand these two voltage ranges.
Thanks for the reply.
These are just the voltages either side of the pwm CCs.
Your info has clarified a few things.
Probably use a few on the fish tanks. Heating or cooling depending on the season
Another possible use is sticking these on the back of a normal fridge to assist...are there any heat sink adhesives?

Been using some VHM-009 over/under battery voltage modules. Very reliable when switching another relay.

 
heat sink adhesives?
This is what I used between the peltier/copper/aluminum interfaces: https://www.amazon.com/Smalloo-100x100x1mm-Conductive-Resistance-Temperature/dp/B0922C83Z7

it is merely sticky think vinyl statically clinging to glass maybe a bit stronger?

thermal conductivity 8 W/(m•K) is decent.. peltier face is 30. aluminum is ~200. copper is near 400.

it’s weak adhesive though and won’t resist much pulling force and probably would eventually give with even a tiny constant force. i have wrapped the entire inedible sandwich with tape to keep the thermal pad firmly squeezed against the interfaces.

thermal epoxy exists and if you search newegg or other places it should be abundant. that sort of thing in a tube can hold much stronger and longer.

if you want to be a total crazy person you can use thin indium sheets which has 80 W/(m•K) thermal conductivity which is great for thermal interface material which usually ranges like 2-8 for PC thermal paste from my reading. it has no adhesive properties though. very easy to disassemble though. food for thought

really like those applications!

lol now i’m curious what a peltier assisted minifridge with overbuilt insulation would be like *think think*

the midea fridge i have has the condenser embedded into the left and right outer walls. using thermal pad to mate a few 240x80mm aluminum water blocks to cover the area that gets hot(condenser) and then even just pump that through to an external radiator might allow for overbuilt insulation.

anyways trying to focus on heating, but refrigeration with heat pumps has the flip side of pumped heat to be utilized

here’s the data from my recent test:

1628066757319.jpeg

1628066767611.jpeg
ambient ~25C

btw i’m using an ac resistance meter to spot check modules and will be using it to match modules going forward. most individual TEC1-12715 modules exhibit 0.8-0.9 ohms ac resistance as measured by yr1035 meter.

it’s so weird, i can feel the cycling heat very gently from the AC resistance measurement waveform. just mentioning this so people can check if they got a fake module

you can be fairly certain that your TEC1-12715 is fake if the ac resistance meter indicates in excess of 1.5 ohms.
 
Last edited:
The issue that many don't think about with peltier they are a window to the world when they aren't powered. Heat loss is an issue when they are opportunity loads.
 
I dump excess power into a grid tie inverter from midday onwards during the sunny seasons using a voltage sensing relay and a 24hr timer.
Statistically these systems work well. I favor timed systems that operate a device for a short period of time and then check voltage again instead of waiting for a drain down voltage. Again looking at panel voltage is probably a better indicator charge state.
 
Statistically these systems work well. I favor timed systems that operate a device for a short period of time and then check voltage again instead of waiting for a drain down voltage. Again looking at panel voltage is probably a better indicator charge state.

You're completely right. I'd do better if I had a system to disconnect the panels and repurpose them but I'm not that advanced yet.

That said I do use a 10w panel as a "sun sensor" to trigger automotive relays for other projects.
 
here’s the data from my recent test:

ambient ~25C

btw i’m using an ac resistance meter to spot check modules and will be using it to match modules going forward. most individual TEC1-12715 modules exhibit 0.8-0.9 ohms ac resistance as measured by yr1035 meter.

it’s so weird, i can feel the cycling heat very gently from the AC resistance measurement waveform. just mentioning this so people can check if they got a fake module

Have you measured temperature change and water volume though a heat exchanger? Or rate of temperature change for a volume of water?
That would let you determine Joules or BTU/hour for a given power input and temperature delta between hot/cold sides.
 
wow cool!!! dissipating heat directly through the hull sounds so cool! either through the hull or by drawing in sea water through a heat exchanger. i would worry about biology clogging a pumped heat exchanger. thank you for mentioning this interesting application!
Do a search for 'keel coolers'. They have been used decades.
 
Have you measured temperature change and water volume though a heat exchanger? Or rate of temperature change for a volume of water?
That would let you determine Joules or BTU/hour for a given power input and temperature delta between hot/cold sides.
good idea ? i’ll try running the pump continuously and measure the weight of the water it pumps over one minute or more and note the grams and seconds elapsed. this should give a decent approximation of the grams per second flow rate.

measure mass of aluminum water block empty vs full of water to evaluate volume in block.. somewhat annoying to measure total volume in loop…hmmm i could use known flow rate from above and then time how long a lone air bubble takes to transit that side of the loop.. hmm…. thanks

attached is the csv log of this test run that lasted about a half an hour real time. had to change to txt to upload, but it's csv.
  1. cool down the secondary side to ~8C
  2. deactivated, backflow mode
  3. heated secondary side ~54C
  4. deactivated, backflow mode
  5. cooled secondary side to ~8C again
  6. force both side temperature to converge to ~25C ambient at the end
easy to see how fast the heat backflow is when deactivated. These are measured at the Water Block. The process volume is separate from the Water Block Heat Exchanger.
1628098998597.jpeg
1628099129548.jpeg
there are three thermometers mounted right now. one touching the back side of the big water block “primary side” and two on the “secondary side” one back of water block and one right on the copper surface that interfaces with the water block. the arduino writes all three temperature to serial usb every 200ms for computer log

the thermometer touching the plate has quickest response of all three of course. i used thermal pad to affix each thermometer to maximize heat flow into them. don’t want to measure wire temperature. the two on back of water block have delayed response speed relative to the plate one.

The issue that many don't think about with peltier they are a window to the world when they aren't powered. Heat loss is an issue when they are opportunity loads.
effectively not an issue if you have them insulated from the body you’re cooling or heating and use a water pump to transport the heat. you’re right that heat back flows across the peltier module when it’s turned off. the graph above reinforces that you’re right. camper coolers allow this back flow heat to reach the food, that’s a design flaw as you say. silicone is not very conductive.

the tiny water pumps i’m talking use ~5W each at 12V so it costs 10W to gain ability to modulate the back flow of heat. never let the peltier directly touch the process volume unless you’re okay with instant back flow of heat when deactivated. check out my graph and you’ll be able to even observe how quickly the back flow occurs :)

heat will very slowly back flow along the water and the silicone tubing, but water has thermal conductivity of 0.6W/(m•K) and high specific heat capacity and the inner diameter of the tube is ~7mm so the back flow will be at least one order of magnitude less than a 40x40mm peltier plate which has 30W/(m•K) thermal conductivity for the alumina surface but only ~2W/(m•K) for the bismuth telluride inside

(pi*(7/2)^2)*0.6 vs (40*40)*30 ? is what we are talking about

1628100722517.png

thus, the coleman peltier cooler type should be about 2000x worse at managing heat backflow than the solution i’m talking about with pumps
 

Attachments

  • TEC_test.txt
    162 KB · Views: 1
Last edited:
Have you measured temperature change and water volume though a heat exchanger? Or rate of temperature change for a volume of water?
That would let you determine Joules or BTU/hour for a given power input and temperature delta between hot/cold sides.
Math...

The secondary side has four main components other than water.
  1. aluminum heat block touching peltier stack
  2. aluminum heat block touching process volume
  3. silicone tubing
  4. water pump
Rough numbers..
  1. 176 grams. 200x40x12mm. aluminum 2.7g/cm^3. solid block would be 259.2 grams. implies 30.8 cm^3 volume inside this block.
  2. 75 grams. 80x40x12mm. aluminum 2.7g/cm^3. solid block would be 103.7 grams. implies 10.6 cm^3 volume inside this block.
  3. ~40 cm length. inner diameter 7mm. 0.38 cm^2 area inside. implies ~15 cm^3 volume inside tubing.
  4. just going to guess and say 3 cm^3.
Ok.. so I'm going to estimate 30.8+10.6+15+3 = 59.4 cubic centimeters of volume in this loop.

Water is ~1 gram per cubic centimeter. 4,184 Joules to increase 1 kilogram of H2O temperature by 1 degree celsius.

4.184 Joules to increase 1 gram of water by 1 degree celsius. ~60 grams of water.

So if I haven't gone totally off the rails with my calculations and estimates... approximately 250 Joules per degree celsius changed in the water loop.

This is purely an estimate of the water volume, and ignores the specific heat capacity of the aluminum blocks.

I wrote a script to take the previous 25 log entries and tabulate the average temperature change over that period. That worked out to about 21 seconds of past data being taken into account. Attached is that printout of the derived data showing delta temperature of the secondary aluminum water block over time.

During the period of heating the secondary side, a steady 0.25-0.35 degrees celsius per second was averaged. Since this thread is talking about air heating and water can service that..

That means... 75 Joules per second... 75 W thermal transfer? 255 BTU/hr? Doesn't really seem like every element is running at good efficiency.

edit: mass of aluminum in loop is ~250g and at 0.9 joule / gram Kelvin it should account for an additional 225 Joules per degree change. ? any thermodynamics knowledge advice here would be greatly appreciated.

edit2: there are copper heat spreaders inside the stack.. 250x63x1mm -> ~15.75 cubic cm.. it's 140g... 0.38 J/g K for copper.. there are 5 of these copper plates... that implies an additional 266 Joules per degree Celsius change... probably better to delete those, not clear if they actually provide a benefit.. i included them in the hopes of ensuring the adjacent peltier modules at each layer would stay close in temperature.

My end goal is to individually power each layer, and individually monitor the Seeback Voltage of each string/layer. This would allow for realtime feedback of the delta T per layer... still working on that.. As I move on to 7-10 layer designs, individual power control and monitoring of Seeback Voltage will be essential to efficient operation under a variety of conditions. This test was with 4 layers because i got lazy and got tired of applying thermal pads on the way to 7 layers prototype. It's so tedious to apply the thermal paste that I want to design 3D printed brackets with edge space carved out so I can just add a bunch of thermal paste and put it in a vice and have the excess go into the seam.

I ignored the thermal mass of the water blocks, so I am probably very off the mark by now. But I tried.

this makes me want to add more layers and lower the voltage. in my tests, 1.0V per module resulted in twice as many degrees change per watt as 2.0V per module.......

at this point just writing a script to model the individual surface area, mass, specific heat capacity, instantaneous temperature, of each material in the stack seems like my only real chance of being rigorous.

Anyways, feedback welcome.
 

Attachments

  • tec_derived_data.txt
    234.5 KB · Views: 1
Last edited:
My end goal is to individually power each layer, and individually monitor the Seeback Voltage of each string/layer. This would allow for realtime feedback of the delta T per layer... still working on that..

May I propose a name for that, "Peltier Monitoring System"?
The acronym should be obvious.


So what's the final figure, watts consumed vs. watts pumped across degrees temperature differential?
Or some other units to compare to conventional commercial refrigeration? Should be be quoted as SEER?
 
So what's the final figure, watts consumed vs. watts pumped across degrees temperature differential?
Or some other units to compare to conventional commercial refrigeration? Should be be quoted as SEER?
If I include the water and the aluminum on one side, during heating that side,
((225J/°C alu+250J/°C water)*0.3°C/sec) -> 142 Watt thermal pumped
142 Watt thermal / ( 80W power input ) = 1.78 CoP is my estimate :D

It did 0.2-0.34 °C per second on heating task. Across 16-27°C differential.

That suggests it was pumping between 95-161 Watt thermal.. at 80-100W. It's barely SEER 3 at worst.

If it can do 1.4-1.8 CoP that's 40-80% extra heat for free, right? Heheh. SEER 5-6. Not good enough yet!

If this is true, then I have some work to do on my PMS to get closer to 3-4x joules pumped as joules input.

Cooling is way slower and seemingly much less efficient than I want it to be. Still need to do more experiments.
 
Last edited:
Use the PV panel as solar thermal collector. Draw heat out of the back of the PV panel using water circulation loop.

Pump heat out of that water loop. This will reduce the temperature of the solar panel cells and increase energy yield.

This is how to get the most efficient water heater afaik. Dramatically more efficient than resistor element. Up to 11.3x.

The results show that attractive electrical and thermal performance can be achieved with a maximum annual heating COP of 11.3.

 
Finally - something simple that looks like it'll work:
Using assistants to switch the AC2 output on and off, according to State of Charge.
 
Sounds a lot like San Diego...a good place to get sunburned on a cloudy day.
Santa Cruz county...coastal...pea soup fog in the spring, Im right on the coast......The San Francisco bay area draws in a lot of coastal air to replace the hot air that they generate, resulting in some real pea soup here.
 
Santa Cruz county...coastal...pea soup fog in the spring, Im right on the coast......The San Francisco bay area draws in a lot of coastal air to replace the hot air that they generate, resulting in some real pea soup here.
I think San Diego experiences that because the Imperial Valley to the east, sucks. Which is also why the Imperial Valley gets L.A. smog.
 
Finally - something simple that looks like it'll work:
Using assistants to switch the AC2 output on and off, according to State of Charge.

Thanks for posting that. I wish I had seen that 3 months ago. I have a set up that uses that programming to turn on a space heater for air heating. It was a pain, but I figured out the assistants thing and have my AC2 output configured with voltage - it goes on when my voltage is >27.1V for longer than 60 seconds and goes off when it is < 26.5V for longer than 3 minutes (so that a short microwave burst won't shut it off). It varies with conditions, but this means it turns on when SOC is over 65% with full sun. It basically turns off (voltage <26.5) when the sun goes away. I would like it linked to SOC but haven't figured that out yet.

I use it basically as a signal and use some Hubitat smart home equipment to recognize when there is power on AC2 and then turn on the plug for a space heater in another building 75 feet away from my inverter.

It is working well so far - the heater is in a small cabin and when it is sunny and the heater is on, I don't need to keep a wood fire going all day even when it is really cold (and it is usually sunny when really cold). Its great to forget about the fire at 9 and get it going again at 4 or 5.
 
Back
Top