A temperature controlled fan system for topical acquarium

[PeterMcMonty]

Hello everybody,

I have two tropical aquariums and one of the needs of them is to keep water temperature around 25°C.
During cold seasons it is very easy, since there are commercial heaters that do very well the job: they are sealed glass tubes with a heating resistor and a trimmable electromechanical thermal switch inside. Power supply comes from AC mains and they are completely waterproof.

But what to do during summer months, when ambient temperature can reach even 37-38°C (at least here in Italy)? Plants and fishes are going to boil for weeks or months and bye bye aquarium!

Yes, there are air conditioners, but I don't like them, and in any case it's very expensive to run them 24 hours a day, especially if you don't stay for a long time in that room.

So, the basic idea is to use some 12V computer fans to move the air over the water surface, so allowing a better water evaporation: when water evaporates, passing from the liquid state to the gaseous state, it takes a lot of energy in the form of heat; so the remaining water decreases its temperature. It is the same principle that our body uses to decrease its temperature: by sweat.

As temperature sensor I use an LM35 because it has a linear output of 10 mV/°C with 0V at 0°C. Its output goes to AN4 input of a PIC16F818 which Vref+ (AN3) is set at 0,5V. So the internal 10 bit ADC can range from 0°C up to 50°C. The remaining three analog inputs (AN0, AN1 and AN2 used with Vref+ at +5V) read voltages set by BCD thumbwheel switches that set the temperature over whith the fan operates. This limit can be set from 20.00°C to 29.99°C.

All the informations are displayed on a 2 row x 16 characters alphanumeric LCD display. The fans are activated when temperature is 0.1°C over the limit set by thumbwheel switches and stopped when it decreases to the limit set.

Also, the informations are sent serially with three output lines (Strobe, Clock and Data).

The measured temperature is the sum of 12,500 10 bit samples, divided by 256, rounded to the nearest integer and converted to 4 digits BCD.

The temperature range is limited to +2°C to +45°C. Out of this range the display shows "under" or "over" messages instead of temperature value.

An input pin (RA5/MCLR on real project, but changed to RB7 for simulation, because VSM model for PIC16F818 reads RA5 always zero, regardless if it is set as a digital input) when tied to ground performs temperature calibration.
The procedure is as follows:
- by using a precise thermometer measure water temperature
- set the thumbwheel switches to the measured temperature
- press the CAL button

The temperature readings are summed together until the average reaches the calibration setting, then the number of readings is stored in EEPROM, substituting the default of 12,500 readings. The new number of readings is then used from now on.

At the end of calibration the new number of readings is shown on the display.
After calibration, the maximum error over the entire range is about +/-0.1°C (at least according to simulation results), if it is done around midrange (+25°C). Of course, the calibrating thermometer error has to be added to these values.

If the difference between actual measurement and calibration value is too large, a fail message is shown and the number of readings remains unchanged.

On average (depending on number of readings) each measure repeats every 642 ms.

The firmware, by querying the LCD display for ready state, it is able to detect the presence of the display and can work even without it.

This is an old project I did in 2004 (revised in 2006) and I use it in my acquariums since then. At that time I was using Proteus v4.7 just for schematic capture and PCB design, not for simulation. To evaluate Proteus v8.1 design suite I redesigned the schematic and I created a new project with the same ASM file that works since 10 years.

It's an odd thing that I had to make some little changes to my firmware to allow it run in simulation. That is because PIC16F818 and LM016L alphamumeric LCD display are not modelled correctly.

As said before, Pic16F818 doesn't read RA5/MCLR as an input, so I make it returning MCLR.
Display LM016L responds "not busy" even when it is busy, with the result that in some cases a command or a data is ignored. I solved this issue by adding a delay where it happens.

All these firmware modifications that I've done to make possible to simulate it are commented with a line of asterisks, so to make easy to revert to the original (working in the real world) version.

I include the pdf of the schematics (from PDS v8.1) and a zip file that includes PDS v8.1 project, ISIS v7.10 schematics and all MPLAB IDE v8.92 project (source file, listing, executables, etc.). It includes also the original asm file, without modification (CAL input is RA5 and no further delays are added for the LCD display).

Two words about the use of this device.
I've made a plexiglass front panel covered by a black adhesive plastic sheet and fixed to my acquarium cover.
On this panel are fixed five 40x40xh20mm 12V fans, the LCD display and behind it the breadboard with the control circuit. The power supply comes from a 12V wall adapter and the teperature probe is connected by three wires and an audio circular connector.
The probe is composed by the LM35, C1, C2 and R1, all sealed within plastic hot glue. The most difficult issue is to made it really waterproof.
The panel is fixed on the front edge of the aquarium cover, keeping it opened about 8 cm. I include some photos to give an idea of this: they are in a zip file you can download fom the link below.

I set the cooling temperature at 28.00°C as it is a good compromise between the ideal temperature (25°C) and the amount of reverse osmosis water I have to fill in order to compensate for evaporated water.

This project could be easily modified, adding for example an output to control the heater (a triac optocoupler and a triac can be driven to switch the AC mains). The three thumbswitches could be substituted with four pushbuttons (left/- right/+ escape enter) and an appropriate menu to set low limit, high limit, and calibration.

The backlight could be driven by a PWM and an appropriate command in the menu used for this purpose.

In alternative, the three thumbswitches could be used to set tens of degrees, units and tenths, in order to allow setting a desidered temperature in all the measure range (range that could be increased, say from +1°C to +49°C). In this case, the fan could start at Tset + 0.1°C and stopped at Tset, while the heater could start at Tset - 0.1°C and stopped at Tset.
Also, the hysteresis value could be reduced from actual 0.1°C to 0.05°C.

This could be useful if you need a good thermostatic system, for example to keep controlled temperature in a photographic developing bath.

Of course the firmware could be ported to a more recent and less expensive PIC devices, as PIC16F1507, PIC16F720 or PIC16F1826. Porting to PIC16F819 (2k word instead of 1k word of PIC16F818) is straigtforward: Just change the device definition at the beginning of code.

So, I hope to give somebody some little hint to start for his application.

Best regards,
PeterMcMonty

[PeterMcMonty]

december 26th, 2014

an errata corrige and some improvements in the implementation of the temperature sensor.

First of all, the RC network R1 and C2 are not connected as R-C Low pass filter, but rather as R-C damper, i.e. R1 in series with C2 and both between LM35 output and ground, as shown in LM35 datasheet.
C2 is not 10 uF but 1uF, while R1 remains 10 ohm.

Now I've built the sensor in a different way that (I hope!) is more waterproof than before: I used for Vcc bypass C1 (100 nF) and R1 (10 ohm) 0805 smd chips, while C2 (1 uF ceramic 16V) is a 0603 smd chip.
All the chips, toghether with LM35DZ in TO92 are assembled on a small 5mm x 10mm x 1,6mm thick double sided pcb board. I want to use a 0,8mm thick board, but it is not available at local stores and I don't want to buy it via internet (I'm too lazy...).

The whole assembly fits in a vial cap (a vial is a small container, typically cylindrical and made of glass, used especially for holding liquid medicines); as the cap I mean that part you break to open the vial.

First I put some thermally conductive silicon grease inside the cap, then I inserted the sensor circuit, already connected with three silicon rubber insulated wires, and finally I sealed everything with transparent silicone sealant (that used to seal windows, bathroom fixtures, aquarium glasses, etc.).

Now I'm testing the new sensor, sinked in one of my aquariums, to see if it is really waterproof.

In fact, my big problem with this project is that sensors remain waterproof for a while (sometimes several years), but at some time water fills in and the readout goes to overflow.

Take the sensor out of water, disconnect it and wait a couple of days to let it dry makes it to return functional, but it must be sealed again.

In this way I hope at least that the resealing task will be easier.

Finally, I added the PCB layout, both for the main bord and the sensor.

Season greetings to everybody,
PeterMcMonty

Links:
Schematics in pdf (Fan818dsp3_revised_v2.pdf):
https://app.box.com/s/pgueapu18di1qe8ndom3

Proteus 7.10 and Proteus 8.1 projects, with firmware (FAN818dsp3_revised_v2.zip):
https://app.box.com/s/xzs58ogn8foq5s3is9fb

Gerber files, generated by Proteus 8.1 (Fan818dsp3_revised_v2 - CADCAM.zip):
https://app.box.com/s/m5bc9s8b465b5m8rm4xs

CAM350 v10.0.0.279 pcb (Fan818dsp3_revised_v2.cam):
https://app.box.com/s/jpi8obgpg7v6hnoioqio

Temperature sensor (new implementation) photos (2014_12_22_temp_sensor_in_a_vial_cap.rar):
https://app.box.com/s/pudmfn1ayk7x9llz0qbb

Photos (2014_10_23_ventole_acquario_80lt.zip):
https://app.box.com/s/7d7afrcmoxnvfvrjc4jy

[Gallymimu]

you should use a fully encapsulated thermistor instead of the LM35 and you won't have the problems with water in your sensor.

Also, there is no reason to use thermal grease since you don't have a thermal load on your sensor.  It will be at the surrounding temperature regardless as long as the thermal gradients to not change rapidly (which they won't in a tank of water).

Otherwise neat project!

[PeterMcMonty]

you should use a fully encapsulated thermistor instead of the LM35 and you won't have the problems with water in your sensor.

Yes, or a fully encapsulated PTC.
But NTC thermistors are non linear devices and even if you have an R to T curve or even an equation, it's more difficult to calibrate them with just one calibration point.
PTC are more linear (but not exactly: there is a four or five order equation for them, even if the most weighting term is that of first order) and precise (depending on the grade... and cost), but a very small change over temperature: a PT100 tipically has 100 ohm at 0°C and 137,5 ohm at 100°C: you should compare the voltage drop of PT100 to the voltage drop of a reference 100 ohm resistor, with a wien bridge or a current mirror and then use a differential amplifier to amplify the difference voltage, introducing more error sources.
LM35 is straigforward: 10 mV/°C starting from 0V at 0°C. No amplifier, directly in the ADC input and just one point calibration for the slope, neglecting the offset.

Also, there is no reason to use thermal grease since you don't have a thermal load on your sensor.  It will be at the surrounding temperature regardless as long as the thermal gradients to not change rapidly (which they won't in a tank of water).

Right! In fact, up to now I simply encapsulated the sensor in silicone sealant (I tried also hot plastic rubber, that you find in rods that melts with a dedicated gun), but putting it in a glass cap leaves air between the glass and the sensor, so I though that thermal grease could improve the thermal contact in between, even if thermal gradients in a thank of water don't change rapidly, as you correctly said.

Otherwise neat project!

thank you for your appreciation!

[bobcat1]

@PeterMcMonty

You can get encapsulated and sealed low cost thermistor in eBay about 1-3 USD low quantity
Also as an add on to your project - you can add PH and Pressure sensor (also found in eBay for low cost) to your project
to have a fully automatic aquarium controller.

All the best

Bobi

[PeterMcMonty]

@bobcat1

NTC thermistors are not linear and therefore difficult to calibrate (two calibration points are needed, at least). Platinum PTC (PT100, PT500 or PT1000) are better, since they are almost linear, but either for NTC and PTC the cables resistance must be keeped into account, and copper resistance varies with temperature too!
Also, Platinum PTC has a little resistance change with temperature and must be feeded with as low current as possible (max. 1 mA, but 100 uA are better), giving an even smaller dV/dT output, compared with the offset voltage.
For example, a PT100 has 100 ohm at 0°C and 138,51 ohm at 100°C. With 100 uA you get 10 mV at 0°C and 13,851 mV at 100°C.
PT1000 is not so much better: you get 1000 ohm at 0°C and 1385,1 ohm at 100°C that means 100 mV at 0°C and 138,51 mV at 100°C.
Here are some tables for Platinum PTCs:
PT100: http://www.hayashidenko.co.jp/en/info12.html
PT500: http://www.hayashidenko.co.jp/en/info13.html
PT1000: http://www.hayashidenko.co.jp/en/info14.html

As I said before, LM35 is simple and straightforward: 10 mV/°C from 0°C = 0 mV up to 100°C (1 V) for LM35DZ. I'm interested in the 0 to 50 °C range, especially temperatures from 20°C to 30°C, so I feed it directly to the ADC input, setting its Vref at 0,5V to get the better resolution with 10 bit A/D conversion.

Since every sensor may have a +/-1,5°C accuracy error at 25°C, I've implemented a firmware calibration routine, that corrects the number of readings according to the "real" temperature measured with a more accurate thermometer. This is a very simple one-point calibration.

About pH measurements, I have a portable device (HI98129 by Hanna Instruments: http://www.hanna.it/tascabili-e-indicatori/un-unico-strumento-tascabile-per-phectdstemperatura-scala-ectds-bassa/#.VKBab_8kA - sorry it is in Italian, but you can have an idea by the pictures) that measures pH, EC (electroconductivity, in uS), TDS and temperature.
Since I don't have a system to fertilize plants by adding CO2, I don't need a constant monitoring of pH, just periodical measurements, for the moment, but maybe in the future...

[pickit2]

A long time ago we had some kit in the field, one of our engineers tried this to bring in to calibration, he made the power supply to the LM35 variable. by adjusting the voltage he set the LM35 to read the same as the test kit used by the service engineers.
It saved replacing or returning the kit to the workshop, when a LM35 sensor had to be replaced.

[mtmcndrw]

An excellent well documented project that can easily be implemented for other uses.

In my case I am going to use the project for the temperature control of a propagator used in the germination of chilli seeds.

When growing under artificial light a lot of heat has to be removed to keep the soil/air at optimum temperature.

One added bonus of this project is the use of airflow. This achieves what I require quickly but the airflow also replicates the seeds growing in the open in windy conditions. This means that they don't grow tall and spindly as they normally do in a propagator but short with thick stems.

[PeterMcMonty]

A long time ago we had some kit in the field, one of our engineers tried this to bring in to calibration, he made the power supply to the LM35 variable. by adjusting the voltage he set the LM35 to read the same as the test kit used by the service engineers.
It saved replacing or returning the kit to the workshop, when a LM35 sensor had to be replaced.

Thank you Pickit2: I wasn't aware that LM35 readout is power supply dependent. This may be useful to calibrate it, as in the case you describe, but also a warning to have a well regulated power supply, to avoid unwanted readout changes.

An excellent well documented project that can easily be implemented for other uses.

In my case I am going to use the project for the temperature control of a propagator used in the germination of chilli seeds.

When growing under artificial light a lot of heat has to be removed to keep the soil/air at optimum temperature.

One added bonus of this project is the use of airflow. This achieves what I require quickly but the airflow also replicates the seeds growing in the open in windy conditions. This means that they don't grow tall and spindly as they normally do in a propagator but short with thick stems.

Thank you mtmcndrw I'm happy you appreciate this project.

Effectively I started by simply using some fan to create an airflow that helps water to evaporate and subtract heat from the tank, but suddendly I realized that the system was too much efficient, since it cooled the tank below 25°C making the heater resistance to turn on: that was very silly... cool the water so much that it needs to be heated...

So I though to a temperature control system. I started with an analogic circuit with a double op-amp: the first one as a Sallen and Key low pass active filter with a cutoff around 1 Hz and a gain of ten, the second one as a comparator with a small histeresys (the equivalent of about 0,2°C). The threshold of the comparator was set with a 20 turns trimmer.

It worked perfectly, but the adjustement had to be done when the water temperature was going to rise to the maximum allowed level, i.e. during summer.
So I thought that a digital approach could be more friendly to manage, and that's all.

Just an hint: if you are going to drive loads that need more than 500 mA, or that are not supplied by 12 Vdc you need to redesign the output part of my circuit. The easier solution (if you are too lazy   ) is to use the circuit as it is to drive the 12 Vdc coil of a relay (don't forget an 1N4004 or similar in parallel to the coil, cathode to +12V, anode to transistor collector). But there are more elegant solutions, depending on the load you need to drive.

Another hint: keep the sensor far away from the air flow.

Please, keep us informed about your application: personally I'm really interested in.

Season greetings to everybody
PeterMcMonty

[Gallymimu]

I have to disagree with your perceived complexity of a thermistor.

I don't know how "accurate" you think you need to be but you certainly don't need to code any high order polynomials.  Simply use a lookup table generated in excel and you are good to go.

As for calibration, as is the case with most things you can do a simple one point (offset) or two point (gain and offset) calibration just like with any other sensor.

There is no real advantage these days to use a sensor that is more linear off the shelf since we typically have lots of processing power for calculations or memory for lookup tables (again assuming you are not shaving pennies off the cost of a high volume product).

just FYI we are using these in a project. http://www.digikey.com/product-detail/en/CWF1B103F3380/317-1310-ND/1191085

That said of course you should use what you are comfortable with.  LM35 if dead simple.  I just find them a pain to mount in any kind of usable fashion.

[PeterMcMonty]

I used thermistors in several designs and I'm not saying that they are complex. Also a lookup table is the easiest and practical solution for this kind of detector. It is even possible to "linearize" their response in some range, by using a calculated resistor network (two resistors and the thermistor).

But as you said, LM35 is dead simple and I did my project around it.
I did also several probes that last immersed in water several years, sealed either with hot plastic glue or silicone sealant.

But sometimes they fail and let water fill in. No problem: I make another probe and do recalibration. It's easier than redesign the whole circuit around a specific NTC.

Now, using the glass cap of medicine vials (I use them for one of my cats, that has problems to his kidneys) and smd components ona a tiny PCB, I think I've improved waterproof performance and in the case of failure it's easier to repair the probe.

[Gallymimu]

what failed when water entered your previous designs?  I am curious what the challenges were with the water seal.

[PeterMcMonty]

what failed when water entered your previous designs?  I am curious what the challenges were with the water seal.

very simple: readout goes in overflow, as if the measured temperature is higher than maximum. In other words, LM35 output tends to go toward the +5V supply (I never measured with a DVM if it goes to +Vcc or simply goes over +500 mV).
After you remove the probe out of water and disconnect from the circuit and you let the probe enough time to dry (a couple of days, tipically), it returns to work as before without any electrical damage. Of course it has to be better sealed, before dip it in the tank again.

The main problem is where the wires go out of the seal: it is the preferred way for water to filter inside. Fortunately I have some amount of silicone insulated wire, on which the silicone sealant can make good seal. But it is sufficient a small gap and everithing has to be reworked.

[ehnonymouse]

I have to disagree with your perceived complexity of a thermistor.

I don't know how "accurate" you think you need to be but you certainly don't need to code any high order polynomials. 

Agreed, if you are controlling to one temperature only (25C) then you only have to control to meet the corresponding resistance. No complexity or look-up table.

[PeterMcMonty]

Agreed, if you are controlling to one temperature only (25C) then you only have to control to meet the corresponding resistance. No complexity or look-up table.

Right, but I want to use it as a digital thermometer too, at least in the range 20 to 30 °C, that is what interests me.

Tipically water temperature in tropical aquariums should stay at 25°C, and this is controlled by a commercial heater with a trimmable thermal switch: it is good during cold season, when you need to heat water. Typically those heaters have around 0,5°C of hysteresis.

During hot season, the need is opposite: you need to cool water. Do exists some commercial devices that operates as a refrigerator or an air conditioner: they exchange heat by circulating water in a serpentine in contact with another circuit with some gas a s Freon or similar. They are expensive and require a more complex installation.

The idea of cooling water by evaporation, using fans to move the air over the surface, is cheap and simpler to install and manage, but the diadvantage is that ...water evaporates! And I have to fill the tanks with RO (reverse osmothic) water to keep the same level of water in the tank.

If I pretend to keep temperature at 25°C also during hot season then I have to add several liters of water each day. The 28°C temperature is a compromise that allows me to add some 1 to 2 liters every couple of days. Of course, depending on ambient temperature: in Italy in july it isn't uncommon to have a week or two over 36-37°C: in that case I have to fill tanks even with 2 or 3 liters a day.

Moreover, if i want to set another threshold or to set calibration, a linear sensor keeps life simpler.

This new sensor is in a tank since dec 24th and is still perfectly working. Moreover, it has a very fast response, as compared with my previous ones. It's easier to assemble and easier to seal and, I think, easier to repair. Usually a bad sealed sensor go crazy in a couple of days, otherwise it lasts for several years.

[Signal]

Forgive me this slight off topic. When I research the discussed problem some years ago I met the solution that I consider as the most elegant while not the cheapest - air conditioner in room where aquarium tank is located.

[CocaCola]

During hot season, the need is opposite: you need to cool water. Do exists some commercial devices that operates as a refrigerator or an air conditioner: they exchange heat by circulating water in a serpentine in contact with another circuit with some gas a s Freon or similar. They are expensive and require a more complex installation.

You can do this on the cheap if you want and without a complex system...  For example take a small dorm or apartment refrigerator (I can find these used for about $20, new they are about $70), a small pump and either a coil of hose or a water tight canister...  Then just use a temperture circuit just like you have already to turn the pump on and off that circulated the water through the coil of tubing in the refrigerator or the sealed canister in the refrigerator when needed... With a little trial and error you can easily find a cycle that doesn't shock the water temp, something like 2 seconds on wait 5 minutes re-take temp and cycle again as needed type of pattern...  I knew a small pet shop owner that did this for his 'pet' lobster at the store vs getting a fancy lobster tank, although for his system he didn't need temperature circuit as the lobsters like it cold so he was able to just pump 24/7 after dialing in the fridge down to the proper temp...

[Vineyards]

Are you sure you can get that much delta t from your fan based cooling system. You must have a thermal load to start with.

If I understand you correctly you want to place fans on top of the tank. In my opinion,  the evaporation area would be too small compared to the total mass of water in the tank. Up to a certain air flow rate the water surface will act like a wall limiting the heat exchange to the surface due to the zeta potential.  You should also consider a practical limit on the fan RPM. If it goes too high the increased air flow rate will overcome the zeta potential but it will lead to excessive turbulance which will distorb the fish and cause spillage.

In industrial environments cooling towers are used to obtain delta t's in the vicinity of 9 degrees Celcius. These towers have wooden or fiber fins on which water is spilled in small drops from the ceiling of the tower. When the water drops hit these fins their exposure surface and exposure duration increase which results in better thermal exchange. If the ambient temperature is too high a giant fan is switched on, if the temperature is excessive like in your case chiller units using ammonia as a coolant are also added. I suspect you can get that much performance from the mechanism you laid out in your message.  

You could use a circulation pump and an aluminum (preferably copper) radiator and mount the fan on this unit. You probably don't need temperature control since with these delta t's you probably can't overcool anyway (unless this is an unattended system). Actually, since you still don't have an appreciable thermal load you might think of adding peltiers to this system. It may not be very energy efficient but that seems to be the easiest way to go. There are many CPU cooling systems which you can modify for this purpose.

Furthermore, evaporating tank water inevitably causes an increase in concentration hence a constant decrease in water quality. Salts, ammonia or chloramines will become more concentrated in an evaporation cooled sytem and you will have to think of a blow down/make up method to maintain the concentration which further complicates the matters.  This point is probably at least as important as temperature stability for fish health.

[CocaCola]

Furthermore, evaporating tank water inevitably causes an increase in concentration hence a constant decrease in water quality. Salts, ammonia or chloramines will become more concentrated in an evaporation cooled sytem and you will have to think of a blow down/make up method to maintain the concentration which further complicates the matters.  This point is probably at least as important as temperature stability for fish health.

I have to wonder how much additional evaporation one would expect, over just the norm?  From experience you can expect about a 1-5% water loss a day normally on an aquarium due to evaporation, the rate will vary obviously depending on setup and circumstances...  Either way, the varying concentration of dissolved solids is something that aquarium hobbyist deal with every day, especially the salt water guys so constant replenishing of the water level is a given anyway in a tank where a specific level of dissolved solids like salts is an immediate concern...  This is why most salt water guys now use a RO/DI water constant drip system to maintain proper water volume at all times..  In a properly cycled and maintained fresh water tank it's a very minimal concern as the 'new way' recommended maintenance procedures for aquariums call for weekly if not daily water exchanges to keep the water fresh and avoid high concentrations of dissolved solids that the filter system will not remove or convert...

Also on that subject fan coolers like this have and continue to be used as a cheaper way to moderately cool aquariums, so although not a perfect cooling system it has a proven track record of working...

[Gallymimu]

There is a calculator at the bottom of this page for surface area, air flow, heat load, and water loss.  It's for swimming pools and doesn't work at very low values but you should be able to linearly scale whatever your situation is for a fish tank.

http://www.engineeringtoolbox.com/evaporation-water-surface-d_690.html

[PeterMcMonty]

I have two tropical aquariums: one is about 60 liters and the other is about 80 liters. No marine salt water (it's even more complicated!). I have several plants, that produce oxygen and consumes carbon dioxide, as any vegetable, during daylight period and the opposite during dark period.
As fishes I have guppies in the 80 lt and platys in the 60 lt, as well as corydoras aeneus (a small catfish from South America) and some caridinia japonica (a small transparent shrimp from japanese rivers).
All fishes and plants are considered "not difficult", so what is important is that water parameter are stable, but they can range since in my case they are not critical.

Periodically I make a check of some parameters of water, by chemical reagents and by a multi purpose instrument.
Tipically the most important parameters are:
pH (the content of H+ ions, that tells if water is acid or basic): tipically it can range from 6.5 to 7.5, in my aquariums is always around 7.5, since I don't use a carbon dioxide fertilizing system and therefore CO2 tend to be absorbed by plants.
KH (the carbonatic "hardness", i.e. the content of carbonate salts): when CO2 is lacking, some plants are able to get carbonium from these salts, and this parameter tends to drop to zero. A minimum value should be 2 (german degrees), under this value pH could become instable. Tipically I add some preparate (calcium carbonate, I think, but it's commercial and there are no indications) to keep KH at 4.
GH (total hardness) Ihave it around 7 to 8 german degrees.
Contents of ammonia (NH3) or ammonium (NH4) should be zero: ammonia is a poison for all animals living in the tank, ammonium is less dangerous but if pH becomes basic (more than 7), it becames ammonia.
Contents of nitrite (NO2) should be zero as well: less dangerous than NH3, but a poison too.
Contents of nitrate (NO3): it's not harmful for fishes and is somewhat absorbed by plants as a nutrient, but in quantities of more than 50 mg/lt begins to pollute the tank.
Contents of phospates (PO4): same as NO3, but should be kept less than 0,5 mg/lt

A global indication of content of salts ions is the EC (Electro Conductivity, measured in uS/cm): tipically I have around 650 to 700 uS/cm before changing some water and 600 to 700 uS/cm after water change.

I change about 10 liters of water for each tank every week (that is around 10 to 15%): better few and frequently than more but once a month, even if in some rare cases I miss one or two weeks.

The water that I prepare is 70% RO (Reverse Osmothic) water, that has EC less than 10 uS/cm, added to 30% of tap water supplied in my house.

Evaporation is around 1.5 to 2 liter for week for each tank (less for the 60 liters tank) during cold season and about the same amount every two days or even daily (depends on ambient temperature) during hot season.

I fill the tanks with pure RO water, to restore the water levels, for the very reason pointed out by Vineyard: I want to keep the same concentration of salts: I keep monitored it by measuring EC.

So, sorry to be gone out of topic, but aquariums are alive and a bit complicate to understand and keep in equilibrium. I learned with trials and errors since I started, in autumn 2002 and I'm not an expert, but I reached an equilibrium also on load of effort to dedicate to them.
Since summer 2003 I started with this system of cooling by using fans, firstly with a simple analogic circuit, then by that digital controller described here. I had up to four tanks at the same time, all cooled that way during summer and after 12 summer times, where temperature has reached even 40°C same days, I can count over the fingers of my hands the days that my thanks surpassed 28°C: these were the days with maximum relative humidity.

There is a huge amount of technical topics around aquariums: heat control is one of them, but also lighting (intensity, spectra, timing...), chemistry, bacteria biology, water movement and filtering, plants fertilizing with CO2, fishes habits, reproduction, diseases... And a lot of space for electronics.

[Vineyards]

You detailed it quite right. The only objection I may have is the pH value. I think many fish species can adapt to pH variations within two pH units 6 being the minimum and 8 being the maximum.

I build ph, conductivity, orp meters professionally and somewhat knowledgeable about water chemistry. I know how hard it is to keep everything in equilibrium.

Even using your method of 1/3 tap water and 2/3 RO water your make up water contains up to 0.3 ppm chlorine, 0.1 ppm chloramine, 350 micro Siemens conductivity. Chlorine will immediately attack organic compounds forming chloramines. Though not as poisonous as ammonia, chloramines must also never exceed 0.4-0.5 ppm. If you don't want chlorine  in make up water you should add a tiny amount of sodium bisulphate (or sodium metabistulphite) in it. Alternatively, you could filter this water with an activated carbon filter. Chloramines can be removed using zeolite filter. You could let a very small amount of chlorine pass through which will convert ammonia into chloramines which can be removed using zeolite filter as there is not a simple method of removing ammonia from water.

[CocaCola]

The only objection I may have is the pH value. I think many fish species can adapt to pH variations within two pH units 6 being the minimum and 8 being the maximum.

Yes, most fish can adjust to a steady near neutral PH range as long as it's held steady, it's the rapid shifts in PH (even small) that cause the most problems for fish...  There are of course exceptions to the rule...

Also I fully agree with Vineyards if you are using municipal tap water, even diluted it will disrupt your bio filters and system...  Best to hit it with something like Seachem Prime, Pond Prime is exponentially cheaper then regular Prime same stuff or even cheaper is straight up sodium thiosulfate...  But Prime has some nice additional buffers in it that neutralize the ammonia and nitrite as well as the chloramines/chlorines, and $ for $ a bottle of Pond Prime will last most hobbyist a long time for it's simplicity and piece of mind...

You may want to consider more frequent or larger water changes as well, multiple changes totaling 50-100% per week or even more frequent like 50-100% each day for sensitive fish like Discus is not uncommon nowadays, based on 'new school' care techniques...

[Vineyards]

Oops! Sorry, I absent mindedly wrote sodium bisulphate instead of sodium thiosulphate. Sodium bisulphite is used for reducing pH. So never use it for chlorine removal. It will not decrease chlorine instead it will increase the proportion of HOCl (the more active component of free chlorine).

As CocoCola said, rapid pH changes can be very harmful and a good way to ensure it doesn't go up and down every second would be keeping water alkalinity between something like 80-120 ppm. (by adding small amounts of sodium bicarbonate).

[Signal]

... If you don't want chlorine  in make up water you should add a tiny amount of sodium bisulphate (or sodium metabistulphite) in it. Alternatively, you could filter this water with an activated carbon filter. Chloramines can be removed using zeolite filter.

I always used the simplest method to remove chlorine that I consider as classic. Just leave bucket of water for a day before pouring it to fish tank. Not a problem for mentioned tank volumes.