Pool Plant Room

Start the cell, after the salt charge

This is the Monday job, and it is about the CELL. The pH side is a separate fault with its own procedure below - they used to be written as one and gave different answers, which is why they are now split by job. You do not need the pH regulation working to start the cell.

1. Salt in and circulated first. The cell will show Check Salt and refuse to produce until the reading sits between 4.0 and 5.0 g/L, and that is correct behaviour rather than a fault.
2. Read the salt figure on the unit before adding the last sacks, and add the balance against that reading rather than an estimate.
3. Check the CONSEGNA ACL setpoint once it is producing. It governs how hard the cell works, and it is tuned against a measured free chlorine a few days later, not guessed at now.
4. Keep the robot out until free chlorine is under 4 mg/L.

Expect ERRORE PH to still be on the screen, and ignore it for this job. The TRi PRO runs pH and the cell as two independent halves. A fault on the pH side does not stop chlorine production, so the salt charge and everything after it proceed regardless. Correct the pH by hand with item 4 and a strip meanwhile.

Hand the TRi PRO to a technician

This is a legitimate place to stop. Running this pool and calibrating a pH electrode are different trades, and the second is a thirty-minute job for someone who does it weekly. Nothing below is urgent and none of it blocks the pool. Print this page or send the link.

First, the thing that makes it not urgent. The TRi PRO does two separate jobs: it regulates pH through the peristaltic pump, and it drives the salt cell through the ACL side. ERRORE PH is a fault on the pH half only. The cell will still make chlorine with that message on the screen, so the salt charge and everything that follows from it go ahead regardless. What is lost meanwhile is automatic pH correction, and that can be done by hand with item 4 and a test strip.

What to tell them

• Unit: Zodiac TRi salt chlorinator with the TRi PRO module. Symptom: ERRORE PH, persistent over days.
• The probe reads about a unit low. MISURA PH showed 7.0 while a tablet comparator put the water at 7.8 or above, twice, on different days. No ^ beside the setpoint, so the unit never calls for acid, so the peristaltic has been idle past the 72 hours that raises this error by itself.
• Nothing has put acid in this pool since it was refilled, and the fill was fresh hard mains, so the water cannot have reached 7.0 on its own. That is why the probe rather than the water is suspected.
• DIMENS PISC reads 3. Manual 03 bands level 2 as 60 to 90 m³ and this pool is 70 to 75, so it may want changing to 2. Worth their judgement rather than ours.
• Unverified: whether the peristaltic actually pumps. TEST ACID showed nothing moving, but the acid is colourless and the line was already primed, so the test proved nothing either way.
• On site already: a spare peristaltic tube, probe storage solution, and buffer sachets at 7.01 and 10.01, both expired. The unit's calibration wants 7.5, which is not in the house.

What to ask for

1. Calibrate the pH probe, or replace it if its slope has gone.
2. Confirm the peristaltic pump actually delivers, and fit the spare tube if it does not.
3. Set the pool volume level to whatever they judge correct, and say which and why.
4. Clear ERRORE PH and confirm the unit dispenses.
5. Leave the numbers: what the probe read before and after, and what the volume level was set to. Those go in LOG.txt.

The acid drum is 36 percent sulphuric, above the 25 percent PWTAG ceiling for pool supply, and the plant room is about 4.5 m² below ground with one door. Tell them before they open anything.

Check the pH probe against a buffer

Ten minutes, no money, and it decides what to buy. You are not calibrating. You are reading what the probe says about two liquids whose real value is printed on the packet, and writing both numbers down. Never press the confirm key at any point, because that starts a calibration.

You need: the two sachets, two clean glasses, a bottle of clean water for rinsing, a towel. Do not let the probe tip dry out at any stage, and never wipe the glass bulb at the tip; rinse and shake only.

1. Pump selector to 0, then panel off.
2. Shut G and 7 on the suction, and all four D on the returns. The pool and the tank both sit above this room, so without this the POD does not drip when opened, it runs. Towel down.
3. Unscrew the collar holding the BLUE probe in the POD. Hand tight only, no tools. Lift it out gently, cable still attached.
4. Rinse the tip in clean water and shake the drops off.
5. Empty the 7.01 sachet into a glass. Stand the probe in it. Wait two minutes without moving it.
6. Panel on so the unit can read. Menu, CALIB. pH\ACL, CALIB PH. Write down the MISURA PH number.
7. Press MENU to leave. Not select. Select starts a calibration and would tell the unit that this liquid is 7.5.
8. Rinse the tip again. Empty the 10.01 sachet into the second glass, stand the probe in it, two minutes, and read and write down the second number the same way.
9. Rinse, put the probe back in the POD, collar hand tight only.
10. Re-open G, 7 and the four D. Panel on, pump back to M.
11. Check round the POD for drips before you walk away.

What the two numbers mean. About 7 and about 10: the probe is sound, nothing needs buying, and the fault is elsewhere. About 6 and about 9: a clean offset, and a 7.5 buffer will fix it. About 7 but 9 or lower on the second: the slope has collapsed, which a calibration cannot repair - buy a probe, not a buffer. Both near 7: dead, same answer. Three of those four outcomes mean the 7.5 buffer is the wrong purchase, which is why this comes before any order.

Clear ERRORE PH on the TRi PRO

The home screen shows what you asked for, not what the water is. The bottom line of the default display is the pair of setpoints: PH 7.2 on the left, ACL on the right. It is a target. It does not move, and a pH of 7.2 sitting there is not a reading of 7.2, it is the unit telling you what it is aiming at. Anyone watching that number for a change will watch forever. The measured pH lives one menu down, under CALIB. pH\ACL, and is the only figure worth comparing against a tablet test.

Two things the home screen tells you free, and both are precise.
1. Whether dosing is switched on at all. With the peristaltic disabled the bottom line reads pH --- instead of a setpoint. So a visible PH 7.2 is proof that ACID ON is already set and the eight-hour arming wait is long past.
2. Whether the unit currently wants acid. Manual 03 is exact about this: the ^ appears to the right of the pH setpoint when the pool's measured pH is ABOVE that setpoint, and injection then follows automatically. So no ^ means the unit is measuring 7.2 or less. Set that against a tablet test reading 7.8 to 8.2 and the two cannot both be right.

Diagnosed 2026-09-12: the probe reads about one unit low. MISURA PH showed 7.0 while a tablet test put the water at 7.8 to 8.2. The physical argument settles which is wrong, and it does not depend on trusting either instrument: nothing has put acid into this pool since it was refilled. The peristaltic's own idleness is what raised the error, the fill was fresh hard mains, and another 900 L of it went in with the backwash. There is no mechanism by which that water arrives at 7.0. A tired glass electrode drifts toward neutral, and 7.0 is where a dead one sits.

Note what the manual's own test does not catch. Manual 03, section 4.1, says a reading above 8.2 or below 6.4 indicates a dirty or damaged probe. 7.0 passes that test comfortably and is still a whole unit wrong. The window only finds gross failures; a probe can look healthy by it and still be the fault. The only check that finds this one is a second instrument, which is why the tablet kit earned its place the week it arrived.

That combination points at one thing. A probe reading falsely low never calls for acid, so the peristaltic never runs, so the second trigger of ERRORE PH fires on its own: the pump has not been activated for a cumulative 72 hours. No ^, no dosing, and an error that is simply telling the truth about its own idleness. The measured value settles it.

Reading the measured pH without calibrating by accident. There is no MISURA pH line of its own: the value appears inside CALIB PH, as the first thing that screen shows. Read it and back out, with the menu key twice or by simply waiting thirty seconds. Do not press the confirm key, because that begins a calibration, and a calibration performed with the probe sitting in the POD tells the unit that pool water is pH 7.5. That does not fix a lying probe, it makes it lie consistently and hides the fault. Calibration happens in the 7.5 buffer, with the probe out, and never otherwise.

What the message actually means. Manual 03, section 10.3, gives it exactly two triggers and no others: the peristaltic ran three full dosing cycles without reaching the setpoint, or it has not run for a cumulative 72 hours. The second is benign and fires on any unit that has been switched off for days. The first is not, and it means acid is going in and the pH is not moving, or acid is not going in at all. The listed causes, in the manual's own order: the pH minus container is empty, the probe is dirty or uncalibrated, the probe is dead, the pool-size level is too low, the peristaltic has lost its prime.

Why it is worth solving rather than living with. pH is what makes chlorine work. At the red end of the scale only about a quarter of your free chlorine is hypochlorous acid, so every gram you hand-dose is mostly wasted until this is fixed. It is also the thing that will keep the cell honest once the salt goes in.

1. Read MISURA pH before you touch anything. Menu, CALIB. pH\ACL, then the measurement line. This is not the number on the home screen. This is the free cross-check the tablets have been waiting for: two instruments, same water, never once compared. Write it into LOG.txt next to the tablet figure.
2. Interpret that number carefully. Manual 03 says a reading above 8.2 or below 6.4 means the probe is dirty or damaged. That rule assumes the water is in a normal range, and this water is not. If the unit reads high and the tablets read high, the probe is probably fine and the pool really is out at the end of the scale. It is a probe fault only when the two disagree.
3. Run TEST ACIDO and watch the translucent tube. Thirty seconds, stops itself. This one test splits the whole problem in half: acid moves, and the delivery side is sound, so the fault is the probe, the pool-size level or the water's own buffering. Nothing moves, and you have found it. A long suction line may need the test run several times to prime.
4. Look at the drum and the line while you are there. Is there anything in the 25 kg drum. Is the ceramic weight actually under the surface. Is the peristaltic tube perished, flattened or cracked, because that is a consumable and it is the classic silent failure.
5. Check VOLUME ACQUA reads 2. The factory default is 1, so if nobody ever set it, it is on 1 and every dose since installation has been sized for a smaller pool. That alone can produce this error: the unit doses on schedule, the pool does not move, three cycles later it gives up.
6. Then, and only if the first five leave it unexplained, clean and calibrate the probe. After calibrating, the measurement must read 7.5; if it does not, repeat. What the calibration needs is explained below, because it is a thing rather than a step.
7. Total alkalinity is the last suspect and the one nothing here can measure. At a high TAC the unit injects every two hours and the pH does not move, which looks identical to every fault above. Zodiac wants 80 to 150 ppm. A fresh hard-water fill is exactly the condition that overshoots it.

Before pulling the probe out of the POD: stop the pump and switch the panel off, then shut G and 7 on the suction and all four D on the returns. The pool and the tank are both above this room, so there is head from both directions and the POD will not simply drip, it will run. Same precaution as the strainer lid, same reason. Have a towel down. Do not touch the glass bulb at the tip, rinse it in clean water and shake it, never wipe it.

Test the water (pH and free chlorine)

Why it matters: until 2026-09-10 there was no kit here and every dose went in blind. While the cell is off for want of salt, hand-dosed chlorine is the only thing keeping this water safe, and "every two or three days" is a guess dressed as a schedule. This is what turns it into a number. Weekly in normal running, every two days while the cell is off, and always before dosing rather than after.

1. Sample at elbow depth, away from the four D returns and away from the wall. Surface water is not the pool: it carries whatever is floating and has taken the most sun. Water straight off a return is what the plant just made, not what the pool holds.
2. Rinse both cells three times with the water you are about to test, then fill to the 10 mL mark. A cell rinsed under the garden tap reads the garden tap.
3. One tablet per cell, dropped straight from the foil. Do not touch them. They are hygroscopic, and skin carries both oil and chloride. DPD No.1 for free chlorine, phenol red for pH.
4. Cap, and shake until dissolved, crushing the tablet against the wall of the cell if it is slow.
5. Read the chlorine within a minute. DPD starts fading from the moment the colour forms, so a cell left on the side while you do something else reads low.
6. Read against daylight but not direct sun, card behind the cell, something white behind the card.
7. Write both numbers and the time into LOG.txt as a TEST entry. One reading is a number; a column of them is the only thing that shows a drift before it becomes a problem. And keep a running total of the cyanuric acid in the same file, because nothing in this house can measure it and it only ever rises: the granular chlorine carries its own stabiliser at roughly 0.9 mg/L for every 1 mg/L of free chlorine it delivers. Zodiac wants it under 30 and it locks the chlorine up at 50. Arithmetic is the only visibility you have.

If you cannot match the colour, that is itself a reading. A sample whose value sits inside the scale matches easily: one swatch is plainly closest. A sample past the end of the scale matches nothing, because it is more saturated than the top patch and every comparison looks slightly wrong, which is what being unable to decide actually feels like. [2026-09-12: reported exactly this on the pH side, twice, by someone who had never had trouble with this kit in previous years. It is not the eye and it is not the light. The water has gone off the top of the card.] So record it as "at least 8.2" rather than as a failed test, and treat the difficulty as the information it is. The same logic works at the bottom: a chlorine sample that will not match anything because it is paler than the 0.1 patch is a reading of "less than 0.1", not a failure.

The trap that reads as zero. Well above the top of this scale, DPD is bleached straight back to colourless. A cell that flushes pink and then goes clear, or never colours at all in the hour after a dose, is telling you the chlorine is very high, not that there is none. The test for it costs nothing: half sample, half tap water, test that, and double the answer.

This kit stops at 3.0 mg/L, and two figures in this document are above it. The commissioning shock is about 8 mg/L and the robot's ceiling is 4. Neither can be read directly, so use the same half-and-half dilution and double. Test chlorine before pH as a habit: above roughly 3 mg/L free chlorine, phenol red reads falsely high and can go purple outright, so a pH taken alongside an over-scale chlorine is not a pH at all.

Backwash the filter (controlavaggio)

Pausing mid-backwash is allowed, and on this tank it is often necessary. If the column reaches your floor before the sight glass runs clear, stop: pump to 0, reintegro to M, and let the tank recover before carrying on. Nothing is harmed by the interruption and the multiport can sit on 2 as long as you like with the pump off. Never rinse from near mark 1. A rinse costs about 5 cm, so from 64 it finishes around 59, which is inside the 35 to 64 window the interlock is bracketed to, and the pump can cut out with the drain still open. Recover to at least mark 3 first. This happened on 2026-09-12 and the pause was the right call.

Backwashing to a water budget, which is how to do it when the tank is tight. Forget the clock: flow through a fluidising bed is not something this installation has ever measured, and the clear column is a litre meter that does not care. 27.3 L per centimetre. So decide the spend first, convert it to centimetres, and read the column down. Turn the reintegro to 0 before you start, or the top-up quietly replaces what you are spending and the column stops telling you the truth. Back to A afterwards. Reserve the rinse before you spend anything else. Twenty seconds, about 5 cm, and it is the part that stops the vessel's dirty water reaching the pool. Everything left over is the wash. A 450 L budget is 16.5 cm: roughly 11.5 for the wash and 5 for the rinse. That is about one vessel volume, and because the dirt release is front-loaded it recovers most of what a full wash would. Expect a partial result and a partial pressure drop, landing part of the way back toward the clean baseline rather than on it. Stop early and gladly if the sight glass runs clear first.

A backwash is free until there is salt in the water, and then it never is again. Every wash and rinse takes 900 to 1,500 L out of the system permanently, and once the pool is charged that water carries 3.6 to 6.0 kg of salt with it, roughly a quarter of a sack each time. So a bed that is loaded and a treatment that has converged should be washed before the charge goes in, not after. Re-dose about 72 g of flocculant afterwards if the water is not yet fully clear, because the wash strips the layer along with the dirt.

Leave the pump selector on M for the whole procedure and do not touch it. Every "panel off" and "panel on" below means the main switch. The selector must be on M because M means "runs whenever the panel is live": on A the time clock decides, and if it happens to be in an off block you will switch the panel on mid-backwash and get nothing.

Two numbering systems meet in this procedure, so read the labels not the digits. 1, 3 and 4 are components on the plan: the filter drain valve, the sight glass and the pressure gauge. 1 to 6 after the words "lever M" are positions on the Kripsol. The same digit means two different things, and 3 is the worst of them: the sight glass, and Rinse.

1. Panel off.
2. Open filter drain valve 1 (vertical).
3. Lever M to 2, Lavage.
4. Panel on. Run until the water in sight glass 3 runs clear. The glass decides, not the clock; two to three minutes is typical.
5. Panel off.
6. Lever M to 3, Rinse. Leave drain valve 1 open. The rinse is still routed to waste and that is the whole point of it.
7. Panel on for 20 to 40 seconds.
8. Panel off.
9. Lever M back to 1, Filtration.
10. Now return drain valve 1 to horizontal (closed).
11. Panel on.
12. Read gauge 4 and write it into LOG.txt. That reading is the new clean baseline, and it is the number every later one is judged against. On 2026-09-09 it was 1.18.
13. If a flocculant treatment is running, put about 72 g back in. The backwash strips the layer off the sand along with the dirt, which is why the label asks for a smaller dose after every wash.

The order of the last steps was wrong here until 2026-09-10, and it is worth saying why rather than quietly fixing it. This document used to close the drain and return the lever to Filtration, and only then say "do a rinse", with no position given. Followed literally that sends the vessel's remaining dirty water straight into the pool through a bed that fluidising has just left loose and unpacked, which is precisely what the rinse exists to prevent. Rinse before filtration, with the drain still open.

Note from the sheet: the balance tank level drops during this and will stop the pump. Close valve G temporarily. Gauge 4 reads about 1.18 bar on this installation when the sand is clean. This document does not follow the sheet on G. G is the pump's entire water supply, and closing it while the pump runs starves it, which is the one failure that destroys a water-lubricated seal in minutes. If the column falls toward the interlock, switch the panel off instead: same result, no risk.

Clean the pump strainer basket

Why it matters: that basket is the last thing between the pool and the impeller. Let it block and the pump is starved; let it crack and debris goes straight into the impeller, which is the expensive part. Weekly in season, and always after vacuuming, which is what the sheet's handwritten note is about.

1. Pump control to 0, panel off.
2. Shut 7 and G on the suction, and all four D on the returns. The pool and the tank are both above the pump, so there is head from both directions. Have a towel down.
3. Unscrew the two black knobs by hand. No tools - they are moulded knobs on a polycarbonate lid.
4. Lift the lid, lift the basket out, rinse it with water and a soft brush.
5. Check the basket rim for cracks. A split basket passes everything it should be catching.
6. Clean the O-ring and the face it seats on. Grit there is the classic reason a pump draws air. Look for flats, nicks and hardening, and replace it if it has gone stiff rather than finding out on a dry start.
7. Silicone grease only, never petroleum or Vaseline: they swell the elastomer and can craze the lid.
8. Refit hand-tight only. Over-tightening distorts the lid, cracks the plastic around the knob bosses, and makes the seal worse rather than better.
9. Re-open 7 and G. With the tank full the pump body fills by gravity - watch it come up through the lid before starting. Only prime by hand if the system is drained.

The lid is polycarbonate, which is attacked by solvents, alkalis and ammonia. Water and a soft cloth only. For limescale, a dilute citric or acetic solution is safe, then rinse well. Never acetone, alcohol sprays, ammonia glass cleaner or an abrasive pad - all of them craze or permanently fog it, and then you have lost the only window into the pump.

Vacuum the pool floor (aspirazione fanghi)

1. Attach the pole to the vacuum head.
2. Fill the hose with water as fully as you can, so it does not draw air.
3. Unscrew the cover and put the hose into outside point 5.
4. Filtration must be running, pump on.
5. Open vacuum valve 6 (vertical).
6. Close balance tank outlet 7 (transverse). It does not have to be fully closed: the more you close it the harder it sucks. If the head sticks to the floor, open it a little.
7. Reverse the process to finish.

Handwritten on the sheet: the basket under the clear lid after the pump has to be cleaned, by unscrewing the two black knobs. And: with the Kripsol lever on position 4, the dirt is discharged to drain instead of passing through the filter, but then the water you suck out is thrown away.

Refill and restart, after the system has been drained

1. Pump control to 0 before anything else. Not A: on automatic the 24 hour time clock will start it on its own schedule, and the pump's seal is water-lubricated. Dry running destroys it in minutes. There is a low-level interlock that should also hold it off, but it is a backstop, not a plan.
2. Stop the cell producing. Do not switch the TRi PRO off. Fresh water has no salt, and electrolysis at low salt drives the cell voltage up and damages it, so it must not make chlorine; it will show Check Salt until the salt is back, which is correct rather than a fault. But that box does three jobs, and only one of them is chlorine. It also carries the pH probe and drives the peristaltic pump on the acid drum. Kill the whole unit and pH regulation stops with it, on a pool that has just been filled with fresh hard mains water, which is exactly when pH climbs fastest and nothing else in the room will tell you. [learned the hard way, 2026-09-10: the unit was switched off after the refill, nobody separated the two jobs, and the first test four days later put pH at the red end of the scale.]
3. Reintegro on A, all four D valves open. The top-up discharges into the return manifold, so mains water reaches the pool through the four wall outlets. It does not fill the tank. On A it stops itself when the tank probes are satisfied. What M does is unresolved: this document long said it forced the solenoid open with no cut-out, and on 2026-09-10 it was observed stopping by itself on M. Until that is settled, treat M as unsupervised and watch it.
4. Leave the Kripsol lever alone unless the gauge says otherwise. Kripsol's manual says to fill from the mains on 6, closed, so the water bypasses the filter. In this installation that is not needed: with the lever on 1 and the mains flowing, the gauge reads zero, so nothing is being over-pressurised. Check the gauge; if it shows pressure with the pump off, go to 6. Otherwise save the lever movement.
5. Expect a day or more. This line delivers roughly 1.5 to 2 m³/h, so 72 m³ takes 36 to 48 hours. To measure your own rate: one centimetre of pool level is about 400 to 500 litres, so time the rise over an hour. If it is slower than it should be, the manual bypass loop past the solenoid is the biggest single improvement available, and the strainer on the top-up line is the most likely restriction after a drain. Close the bypass before the pool crests, because unlike the reintegro it will not stop itself.
6. While it fills, hold the standing water. It has no sanitiser and no circulation. Pre-dissolve 2 to 2.5 g per m³ of water actually in the pool of the granular chlorine, so about 150 g at 60 m³, and pour it round the perimeter. Repeat every two or three days. That costs about 1 mg/L of cyanuric acid, against 8 for a full shock. Write it in LOG.txt, with the amount.
7. Watch the clear column, not the pool. The instant it starts to rise, the pool is spilling and the tank is filling. From there it is about 45 minutes to an hour to reach 0.55 m, your start level, and roughly another hour before the reintegro satisfies and shuts off.
8. Kripsol's pre-start checks. Turn the motor's fan blades by hand from the back to confirm it is not seized after standing dry. Check the thermal overload has not tripped. Open the suction and discharge valves and fill the pump body with water by hand through the clear strainer lid, and check that lid's O-ring, which is what stops it sucking air. The valve Kripsol means is SUCTION, the red handle immediately before the strainer; no valve is visible on the discharge, so here there is one to open, not two.
9. Valves: G open, which is the same valve as 7, all four D open, B and C open, A almost closed, L and H shut, filter drain 1 shut, and the level column's ISOLATOR open so the column reads the tank rather than a level it was left at.
10. First start on 5, circulation, not 1. Kripsol's instruction, and it is not the obvious choice: circulation bypasses the filter, so you are not driving air and disturbed debris through the sand on the first run. Panel off before the lever moves, handle pressed down as a clutch. Watch the strainer lid clear of air within a minute or two. If it is still churning, stop.
11. Then 2 backwash until the sight glass runs clear, two to three minutes, then 3 rinse briefly, then 1 filtration. Panel off for every movement. Write the gauge reading into LOG.txt after that last backwash: it is the new clean baseline. On 2026-09-09 that was 1.18 bar.
12. Salt: about 290 kg, twelve 25 kg sacks, added with the pump running so it dissolves and circulates. Twelve sacks lands 72.5 m³ at 4.1 g/L. Do not exceed 5.0 g/L, the robot's ceiling. Only then switch the chlorinator on.
13. Calibrate the pH probe in pH 7.5 buffer. You have replaced the entire body of water, so whatever it was trimmed to no longer applies. Check the acid drum has something in it before the TRi PRO starts dosing.
14. Keep the robot out until free chlorine is under 4 mg/L, which rules it out during any commissioning shock.

Nothing here can flood. The pool's level is capped by the deck channel, the channel drains to the tank, and the tank has a passive high-level overflow to waste. A stuck probe or valve sends mains water quietly to the drain rather than onto a floor. [added] The one exception is the manual bypass, which has no cut-out at all.

Add salt to the water

The two dosing routes in this room are opposites, and that is worth holding still for a moment. Flocculant goes in the BALANCE TANK, because the tank is where the pump draws from, so the dose is carried straight to the sand where its whole job is done. Salt goes in the POOL and never the tank, because salt arrives as a solid and undissolved grains drawn off the tank floor go through the pump suction and then the cell. Same room, same pump, opposite answers. The test is simply whether the product needs to reach the filter or needs to dissolve first.

The rule that governs everything here: salt cannot be taken out. Nothing consumes it and nothing removes it except dilution, which means draining and refilling. So under-add and top up, exactly as with a flocculant. The window is narrow at both ends: below 4.0 g/L the cell runs at rising voltage and damages itself, and above 5.0 g/L the robot cleaner is out of specification.

1. Work out the charge. At 70 to 75 m³, 4.0 g/L needs 280 to 300 kg, and one 25 kg sack raises this pool by about 0.345 g/L.
2. Pump running, Kripsol on 1, filtration, for the whole operation and for hours afterwards.
3. Put it in the pool, never in the balance tank. Undissolved salt in the tank is drawn straight into the pump suction and then the cell.
4. Shallow end, spread out rather than piled, and brush it along to help it go. Granular dissolves in minutes to hours.
5. Tablets go in a mesh bag or a clean plastic basket, weighted, in the shallow end. Nothing solid can then travel, nothing sits as a concentrated heap against the resin, and you can lift the bag to see what is left. Allow a day or two: slow dissolution is what the format is designed for.
6. Add most of it, not all of it. Hold back two or three sacks, let the water circulate for several hours, then read the salt figure on the TRi PRO and add the balance against a measurement rather than an estimate.
7. Only when the reading sits between 4.0 and 5.0 does the chlorinator go on. Until then Check Salt is the correct display, not a fault.
8. Top up after backwashes. Each backwash and rinse loses 900 to 1,500 L and with it 3.6 to 6.0 kg of salt, so one sack replaces four to seven of them.

Empty the pool

1. Pump off. Kripsol lever M to 4 (Waste). Valve 1 vertical (open).
2. Valve 7 horizontal (closed). If you leave it vertical (open) the balance tank empties as well.
3. Valve L open (pool floor suction). With M on 4 the pump draws off the bottom and sends it to waste, which is what empties the pool.
4. Pump control to M (manual).
5. Top-up control to 0.

Habits the sheets never mentioned

1. Calibrate the pH and ACL probes at least every two months during the season, in pH 7.5 buffer. Probes drift, so the pH the unit believes it is holding may not be the pH in the pool.
2. Press the test button on the 30 mA RCD every six months. It is the life-safety device in a wet underground room and nothing in the records says it has ever been tested.
3. Write down the gauge reading straight after every backwash. That is this filter's clean baseline, and the climb away from it is the real signal that the sand is loading.
4. Move the valve as Kripsol says: motor stopped, press the handle down firmly first because it acts as a clutch, then move it.

Note: the level column's ISOLATOR, the red-handled valve at its foot, cuts the column off from the balance tank. Closed, the column freezes at whatever level it was showing and the automatic top-up quietly stops working, with nothing else in the room looking wrong. Open it and watch the column move at the start of each season, which is the only check that proves the level this room is reading is the level in the tank.

Procedures this document still owes you

Audited 2026-09-09 against everything that had been explained in conversation but never written down. Since written up: the backwash and what the filter is doing, cleaning the pump strainer, and adding salt. Still only a sentence or a dose rather than a procedure: winterising, four to eight weeks away and the reason the Svernante is on the shelf; calibrating the pH probe, which the TRi PRO manual covers and which falls due once the cell runs; changing the acid drum, 36% sulfuric and deserving steps rather than a caution; cleaning the robot's cartridges. Absent altogether: descaling the salt cell, which is core maintenance on a salt pool; opening the pool in spring; and changing the filter sand, years off since the bed was renewed about September 2025. None is urgent today, and all are the same defect the backwash had: a task you can follow but not reason about.

Five small things, none of them expensive

1. Move some soda ash into this room. You own 25 kg of it as the pH plus on the shelf. It is the correct neutraliser for an acid spill and it is currently on the wrong side of the building.
2. Keep the chlorine out of this room, permanently. Its own label carries EUH031: contact with acids liberates a toxic gas.
3. Clip the overflow pipe's open end in the balance tank. It lies loose on the floor and will move when the pool spills hard.
4. Check the intercapedine is still ventilated. That corridor is a cavity around the below-ground structure, so it is damp by design and the staining and blown plaster below the hatch are expected rather than alarming. It only stays harmless while the cavity can breathe.
5. Buy an unstabilised chlorine for shock dosing, calcium hypochlorite or similar. Both the Recupero and the Svernante labels call for one and there is none on the shelf. This one is a purchase and not a job, so its detail and the words to search for live in the "What to buy" section. It stays listed here because this is where it was noticed.

Note: everything above that needs the balance tank wants doing while it is empty and open. That will not stay true.