Tiling Over Old Screed: Checking the Substrate Step by Step

A tiled surface is rarely the problem itself. If it sounds hollow after years, shows cracks or comes loose at the edges, the cause almost always lies one layer deeper. In existing buildings in particular, with screeds that are twenty, thirty or fifty years old, checking the substrate determines whether an installation will last. This article describes what is checked and in what order. The values given are guide values from sets of rules and manufacturers' specifications – they are not binding legal statements.

Why Most Complaints Come Down to the Substrate

Anyone who looks at cases of damage to tiled surfaces rarely finds the cause in the tile. More often it is voids between the screed and the substrate, cracks that telegraph through into the covering, excessive residual moisture, a dusty surface or one covered with release agents, and missing or tiled-over movement joints. All of this is detectable before tiling. Afterwards it can only be corrected with considerable effort, because the new covering has to come off again first.

Contractually, this is not a side issue. The contractor carrying out the work has a duty to inspect and to give notice: they have to check the substrate before starting work and raise concerns if they identify defects – sensibly in writing and before the first tile. The scope of this inspection is, however, limited. Anything going beyond an inspection using simple means, such as laboratory tests or elaborate measurements, is generally regarded under the relevant contract conditions for tiling work as a special service and has to be commissioned and paid for separately. How this is governed in an individual case depends on the contract; this note does not replace legal advice.

For clients this means one thing above all: checking costs time and sometimes money, a case of damage costs more. If a quotation deals with substrate preparation in general terms or does not mention it at all, it is worth asking which checks and which preparatory works are included and what happens if the substrate turns out to be worse than expected. If you are facing this question and would like to sort out the sequence, we are happy to discuss it with you.

Visual Inspection and Tapping Test

It starts with a look at the surface, ideally in raking light. What is of interest are cracks and their course, efflorescence and discolouration as an indication of moisture, remnants of old coverings and adhesives, plaster or gypsum residues, paint and mortar splashes, as well as the question of whether perimeter joints and edge insulation strips are present and whether there are movement joints within the field. The levels matter too: how much build-up height is available up to the door frame, up to the door leaf and up to the junction with the adjoining rooms?

The tapping test follows immediately. Using a hammer or a test ball, the surface is systematically tapped. A bright, hard sound indicates a firm screed bearing across its full area; a dull, hollow sound points to detachment from the substrate or to a crack beneath the surface. Areas with voids are marked and their extent recorded before deciding whether they can be resin-injected, taken up or accepted.

Three further checks using simple means have become established in practice. The scratch or grid-scratch test shows how firm the surface is: if the screed sands away under the scratch tip, it lacks the strength for a bonded covering. The wipe test with the flat of the hand reveals dust and laitance layers that have to be ground off and vacuumed away. And the wetting test reveals release agents: if a drop of water beads instead of soaking in, oils, impregnations or adhesive residues are involved – then only mechanical removal helps, and no primer. Where greater precision is needed, for instance with thin build-ups on a doubtful substrate, an instrument-based pull-off test is a further option.

One question almost always arises in existing buildings: can the old tiled surface stay in place and be tiled over? In principle this is possible if the old covering is firm and free of voids, the substrate beneath it is load-bearing and the additional build-up height can be accommodated. The prerequisite is thorough preparation: cleaning, degreasing, grinding or roughening, and working with a primer or bonding coat suitable for smooth, non-absorbent substrates. Where the old covering sounds hollow, shows cracks or where the build-up height at doors and junctions is insufficient, removal is the more reliable solution – and often the one that pays off, because you then see the substrate anyway.

All results belong in writing, with the date, the area, the measuring points and photographs. This is not bureaucracy for its own sake. A test record documents the condition at the time of handover, makes notices of concern traceable and is the only sound basis if the cause of damage is discussed years later. It is also useful for the client, because it shows what they paid for in terms of preparation.

Moisture Measurement in Practice

Moisture in the substrate is the cause you see the least and pay for the most. Electrical and capacitive hand-held meters are fast and non-destructive, but deliver only indicative values: their readings are influenced by aggregates, reinforcement, bulk density and salts. As the sole basis for deciding whether tiling may go ahead, they are therefore unsuitable.

In Germany, the CM measurement is the recognised procedure for assessing readiness to receive a covering. Material is taken from the cross-section of the screed – not only from the surface –, crushed, weighed and made to react with calcium carbide in a pressure vessel; the resulting pressure gives the moisture content in CM per cent. The procedure is destructive, and the sampling points have to be made good afterwards. In addition there are newer approaches such as the KRL measurement via the corresponding relative humidity; in the laboratory, the oven-drying method serves as the reference procedure.

The following guide values for ceramic coverings have become established: cement screed up to about 2.0 CM per cent, and up to about 1.8 CM per cent with a heated construction. Calcium sulphate or anhydrite screed is considerably stricter at around 0.5 CM per cent, and for heated constructions some sets of rules state 0.3 CM per cent. These figures are orientation and not a legal requirement – what is decisive are the specifications of the screed manufacturer and of the manufacturer of the installation materials, as well as what was agreed contractually. Anyone wishing to deviate from this should obtain written approval.

With old screed, however, the question shifts. A construction that has been used in dry conditions for decades is as a rule thoroughly dried out. If high readings are nevertheless measured here, the construction period is rarely to blame; instead moisture is flowing in: missing or aged waterproofing against the ground, a pipe leak, rising damp in the masonry. In this case no waiting period and no building drying helps, only clarifying the cause. Tiles are dense enough to conceal the problem for a while, and not dense enough to solve it.

Heated screed forms a point of its own. If there is underfloor heating in an existing building, it is heated up and cooled down again before the covering is laid, following a defined procedure, and this procedure is recorded. The record is the proof that the screed has largely completed its movements – it does not, however, replace the moisture measurement but comes in addition to it. Without both together, no covering should be laid on a heated construction, and during the installation itself stricter requirements then apply, for instance for the movement joints and for the choice of adhesive.

Putting Flatness Tolerances Into Context

Flatness is measured as a gap dimension: a straightedge is laid on the surface and the greatest distance between the straightedge and the substrate is determined. The permissible values depend on how far apart the two bearing points are. The basis is Table 3 of DIN 18202, the German standard on dimensional tolerances in building construction. For finished floors, Row 3 usually applies, with 2 mm at 0.1 m measuring point spacing, 4 mm at 1 m, 10 mm at 4 m, 12 mm at 10 m and 15 mm at 15 m.

Row 4 of the same table contains the increased requirements: 1 mm at 0.1 m, 3 mm at 1 m, 9 mm at 4 m, 12 mm at 10 m and 15 mm at 15 m. These stricter values do not apply automatically, but only if they have been expressly agreed. Anyone expecting them has to write them into the specification – and allow for the fact that the substrate will as a rule have to be additionally levelled for this.

Two misunderstandings persist stubbornly here. First, tolerances are limit values and not target values: a floor that just about complies with Row 3 conforms to the rules and can still be visibly wavy. Second, flatness says nothing about the level or about a desired fall; those are separate characteristics with their own tolerances. The difference becomes practically relevant above all with large formats and narrow joints, because even slight unevenness shows up there as lippage – which formats suit the room and the substrate is the subject of our article on tile formats in the bathroom. The values of DIN 18202 count as generally recognised rules of technology; they are not a statutory requirement.

Levelling Measures at a Glance

The preparation follows from the test results. In order, from the simplest to the most demanding:

• Cleaning, grinding, milling or shot blasting to remove laitance layers, adhesive residues and release agents and to open up the surface
• Primer to suit the absorbency and the type of screed – calcium sulphate screeds require different products than cement screeds
• Close dormant cracks with a force-fit cast resin; cramps inserted crosswise secure them against opening up again
• Levelling and smoothing compounds in the layer thickness approved by the manufacturer; there are separate products for very thin and for thick applications
• Decoupling membranes where there is a residual risk of cracking or where the substrate is still moving
• Levelling fill with dry screed elements where there are larger differences in level or where weight matters
• New screed where strength or load-bearing capacity is insufficient – if in doubt the more reliable and sometimes also the cheaper option

Two rules apply regardless of the measure chosen. Movement and perimeter joints in the substrate have to be carried through into the covering; tiling over them means the cracks appear exactly there. And the products used should come from one system: primer, levelling compound, waterproofing, adhesive and grout are matched to one another, and only then do the data sheet specifications for layer thicknesses, drying and waiting times apply. In a dispute, these data sheets are the more reliable source than any rule of thumb.

Time is the factor most frequently underestimated in substrate preparation. Primers need their flash-off time, levelling compounds and cast resins their curing, bonded waterproofing its drying time before tiling, and the tile adhesive its time before grouting. These times depend on temperature and humidity; in a cold shell in winter everything takes longer than the data sheet states for standard conditions. Anyone planning the sequence too tightly creates precisely the pressure under which the mistakes arise that come back later as complaints.

Which tile finally goes onto this substrate is the comparatively easy question. The floor tiles category in the shop gives an overview of the range, and we have summarised the basics of choosing in our article on floor tiles for modern interiors. In the showroom in Haan, formats, surfaces and slip resistance can be compared directly side by side. Assessing the substrate on site and carrying out the work, by contrast, belong in the hands of a tiling contractor. With old buildings, further questions arise which we describe in our article on bathroom renovation in the Bergisches Land.