If you search for an invisible sliding door in aluminium, you will get three different products back, and they are not three versions of the same thing. One is a structural glass wall whose outer frame has been buried in the building. One is an interior door leaf hanging off a concealed rail, rated for the weight of a bedroom door. One is a panel that disappears into a wall cavity and sits flush with the plaster when it is closed. All three are sold under the same adjective, and the differences between them are not stylistic - they are differences in load class, in who has to make the decision, and in what can still be repaired five years later.
This article is about the first one, because it is the only one of the three that a factory drawing can actually specify, and the only one that fails expensively when it is specified late. We will come back to the other two, because a surprising number of enquiries turn out to be about them.
Before any dimension is discussed, the word has to be pinned down, because the three meanings sit in completely different weight classes. A concealed-track kit sold for interior doors carries the load of a timber leaf. A recessed-frame structural slider carries several hundred kilograms of insulated glass per panel. Nobody should be choosing between those two - they are bought by different people on different drawings.
| What the word means here | What actually gets hidden | Who has to decide it, and when | The load class it lives in | What it cannot remove |
|---|---|---|---|---|
| Recessed outer frame | Head, jambs and bottom track are set into the floor build-up, the ceiling and the reveals, so the visible aluminium shrinks to the vertical meeting stile | Architect and structural contractor, before the slab is poured | Hundreds of kilograms per panel; insulated glass is the dominant term | The interlock, the bottom track, and the drainage route |
| Concealed mechanism, no visible track | The rail is machined into the back of the leaf and the hangers are fixed to the wall, so there is no exposed top or bottom rail | Joinery and hardware selection, at fit-out stage | Interior leaves. Published ratings on retail concealed kits run around 60 to 80 kg, with maximum panel widths around 1006 mm | The weight ceiling, and the flatness of the wall it hangs on |
| Pocket or flush in-wall door | The whole leaf slides into a wall cavity; when closed it sits in the same plane as the wall, with the opening closed by a moving cover profile | Wall build-up at framing stage. Flush plasterboard versions commonly call for around 125 mm of wall thickness | Moderate, but the cavity has to be framed for it | The cavity, the access panel, and whether the hardware can come out again |
That table is the whole argument in miniature. The first row is a piece of building structure that happens to be glazed. The second and third rows are interior architecture. When a buyer asks a structural glazing factory for an invisible door, they usually mean row one - and what they are really asking for is a decision that has to be made before the concrete.
Recessing works by relocating the outer frame out of your field of view. The head goes into the ceiling, the jambs into the reveals, the bottom track into the floor build-up. What is left standing in the middle of the view is the vertical interlock - the meeting stile where two panels overlap when they are closed. It runs floor to ceiling, roughly at eye level, and it is the number you will look at every day for as long as you own the building.
Published figures for this class of system, taken from installers and system houses rather than from us: a true slimline aluminium slider puts the interlock at roughly 20 to 25 mm, a standard aluminium slider sits around 50 to 60 mm, and a uPVC patio door is commonly 100 mm or more. Those are useful reference points. The catch is in which direction they are measured.
One London installer publishes the trick plainly, and it is worth repeating because it is the most practical thing written about slim sightlines anywhere: most doors described as 20 mm only measure 20 mm head-on. Look along the glass from the side and the interlock steps out somewhere in the region of 45 to 65 mm, which is exactly the angle you view it from when you are sitting in the room. A flush interlock that reads the same from every angle is a different and more expensive piece of engineering than a stepped one that reads 20 mm in a brochure photograph. Ask which one you are being quoted.
There is a second measurement that gets sold as slimness and is not. Frame depth - the front-to-back dimension - is a structural capacity number, not a visibility number. It tells you how much glass the profile can carry and how the drainage and hardware are accommodated. It tells you nothing about what you see, because most of it is buried in the reveal anyway. If a quotation leads with frame depth, it is answering a question you did not ask.
Here is what we publish, and it is worth being precise about what each column is, because mixing them up is where most specifications go wrong.
| Series | Frame wall width (depth, buried) | Published frame width | Published sash width | Track and threshold | What it means for a recessed detail |
|---|---|---|---|---|---|
| SL155 four-track | 155 mm | 62 mm | 44.5 mm | Passive check valve drainage; integrated flush handle | Narrowest published sash figure; 20-wheel 400 kg pulleys; single sash to 2800 x 2600 mm |
| SL126/182 (75) | 126 mm two-track, 182 mm three-track | 36 mm | 75 mm | High track 43 mm, low track 8 mm | Described as a 75 mm narrow-edge central interlock with a 38 mm pressure-line-free frame; 27A glazing with inert gas |
| SL126/182 (96) | 126 mm two-track, 182 mm three-track | 36 mm | 96 mm | High track 43 mm, low track 8 mm; eave-type waterproofing; bottom track water-blocking edge | Lift-and-slide; single sash to 2.5 x 3.0 m; the drainage hardware is the reason it can carry that span |
| SL112/167 | 112 mm one and two-track, 167 mm three-track | 26 mm | 53 mm or 75 mm | 45 mm high-ｄｒｏｐ bottom track; optional 85 mm water barrier | This is a deliberately raised threshold. It is the opposite of flush, on purpose |
| TS130 side-press | 130 mm | 26 mm | 75 mm | Track height 53/31 mm | Eight-point locking and four-sided continuous sealing; 200 kg pulleys |
Now the honest part. None of the figures in that table is the 20 mm number used in slimline marketing, and they are not measured the same way. We publish sash width and frame width. We do not publish a head-on interlock sightline taken under the same definition the marketing figure uses, and we are not going to imply one by putting 20 mm in a headline. If your brief says 20 mm, put it in the first email, tell us which measurement you mean, and ask for a drawing with the sightline dimensioned at the angle you will actually be standing at. That is a reasonable request and it should be answered with a dimension, not an adjective.
One more thing worth flagging while you are reading depth figures: depth does not increase with track count. Our four-track SL155 is 155 mm deep, which is shallower than the three-track SL126/182 at 182 mm. Depth is a property of the series, not a function of how many panels it runs. For reference, the widely published depth figures for this class of slim system are around 116 mm for a dual track and 182 mm for a triple track - so our three-track matches the common triple figure while our two-track is deeper than the common dual figure. Do not let anyone on site assume that more tracks means a deeper wall.
This is the section worth reading twice, because most disappointed projects are disappointed here rather than about sightlines. Recessing the frame relocates four elements. It does not eliminate any of them.
| What people try to design out | Why it cannot go | What happens if you force it |
|---|---|---|
| The bottom track | Panel weight rides on it. Our published load figures - 200 kg, 220 kg, 400 kg - are per sash, and in every case the glass is the dominant term. Lift-and-slide lifts the panel only at the moment of moving; it drops back onto the track when you close it | Either the panel gets smaller than the architect drew, or the rollers wear out in a fraction of the design life. There is no third option |
| The interlock | Two sliding panels must overlap somewhere. That overlap carries the lock, the seals and the anti-lift engagement. It is the only place a slider can put them | Nothing. You can make it narrower and carry less glass. You cannot make it zero and still have a sliding door |
| The drainage route | Water that gets into the profile has to leave it. Once the track is below finished floor level, the water leaves below finished floor level, which means it is now a building drainage problem rather than a door problem | Water sits in the recess and finds the interior. The repair is not a door repair |
| Threshold height as a water barrier | In driving rain, height is the cheapest defence you have. Our SL112/167 offers a 45 mm high-ｄｒｏｐ bottom track with an optional 85 mm water barrier precisely because some facades need a dam | You get the flush line and you inherit the water. Pick per elevation, not per photograph |
This is the trade at the heart of every flush threshold, and it deserves to be said in plain terms. A raised threshold is visible, mildly annoying, and does its job. A flush threshold removes the thing you can see and replaces it with a thing you cannot see. Whether that is a good trade depends entirely on how much water arrives at that elevation, and on whether the drainage was designed before the floor was.
One system manufacturer states the flush-threshold requirement more clearly than anyone else: two kinds of water need two separate routes. Water running across the external floor towards the door is collected at floor level, in a channel or a gravel strip. Water that enters the profile itself - condensation, driving rain, moisture running down the glass - exits through drainage openings in the lowest part of the recessed profile. Those are two different systems doing two different jobs, and providing only one of them is the most common reason a flush threshold leaks.
That same source adds a sentence that should be printed on every site drawing: the drainage openings must remain unobstructed at all times, and must not be blocked by sealant, mortar, insulation material, or a floor or terrace finish installed at an excessive height. In practice, the person who blocks them is not the glazier. It is the floor finisher, working to a finished floor level that was set before anyone drew the recess.
Set against that, look again at the two threshold options on our SL126/182: a 43 mm high track and an 8 mm low track, on the same series. The 8 mm low track is a genuine accessibility improvement and it is the closest thing to invisible that a thermal-break slider can offer. It is not a substitute for the drainage list above - it is the reason you need it.
Every specification has a point after which a decision can no longer be reversed cheaply. For a dual-colour frame it is the crimping line, because the two shells cannot be separated once they are rolled onto the polyamide strip. For a recessed slider it is the concrete pour.
A sliding door manufacturer writing for builders says it without softening: the most expensive mistake is a slab poured flat with no recess, because the only remedies afterwards are grinding the slab or raising the interior floor. Both are expensive, and both compromise the detail you were trying to buy.
Tolerances matter just as much as the recess itself. Published installation guidance for flush tracks gives a levelness figure of about 3 mm over 3 m for the sill, beyond which panel operation and sealing start to suffer. A recessed track is not more forgiving than a surface-mounted one; it is less, because you cannot shim your way out afterwards. Published structural guidance for the head is a deflection limit around L/720, with the observation that a head deflecting by roughly 3 mm is enough to bind the top of the panel.
Two more that catch people out. First, protection during fit-out: construction dust, grout and sealant overspray in the track channel block the drainage ports and damage the rollers, so the track should be covered during finishing trades and cleaned and inspected before handover. Second, serviceability: published guidance on pocket systems is blunt that you should never permanently close over the hardware. If a roller, a lock or a pane of glass cannot come out without demolishing the reveal, you have not designed a door, you have designed a sealed box.
| Item | What has to be fixed in writing | If it is left open |
|---|---|---|
| Finished floor levels, inside and out | Both elevations, in millimetres, on the same datum, signed off before formwork | Tile lips at the threshold, or a floor finish that buries the drainage ports |
| Recess depth and width | Enough for the buried profile, the drainage openings, insulation and waterproofing, fixing, and future access | A recess that fits the profile and nothing else - the usual outcome when the floor build-up is fixed first |
| Sub-sill pan and its outlet | Material, full width, slope direction, and where the water physically goes | A pan with nowhere to drain, which is a reservoir |
| External drain and fall | Linear drain position, grate type, and the fall on the external surface | The profile drainage is asked to do the terrace's job, and it cannot |
| Head deflection | Engineer's figure, in writing, for the actual span and loading | Panels that slide beautifully empty and bind once the building is loaded |
| Services in the slab | Confirmation that the recess does not cross heating, water or conduit runs | A recess that cannot be cut at all |
| Access and replaceability | How each roller, lock and pane comes out, and which lining is removable | The first roller replacement becomes a demolition job |
Two honest statements here, and both matter more than they sound.
First, about the numbers we publish. The K values in our documentation - 1.3 on the AW90 passive window, 1.6 on the C120, 1.7 on the C116 - are system values, not whole-door or whole-window values, and they are not NFRC figures. Our sliding series do not carry a published K value at all. If your project needs one, ask for a calculated whole-door figure against your actual glazing build-up and your actual frame fraction, and ask which standard it is calculated under. Do not accept a system value presented as a product value.
Second, and more specific to this detail: the junction between a recessed frame and the floor slab introduces a linear thermal bridge, usually written as a psi value. That bridge belongs to the floor and wall design, not to the door, and it is not inside any door K value you will ever be shown. A beautifully insulated panel buried in a badly detailed slab edge will still produce a cold line at the floor and, in a humid or air-conditioned interior, condensation along it. This is why the recess has to be designed as one assembly with the floor build-up and the waterproofing, rather than handed to the glazier as a finished hole.
For the record on materials, our profiles are 6060-T66 at 1.8 to 2.5 mm, with PA66GF25 polyamide thermal breaks. On the C116 the insulation strip runs 34 to 44 mm. Those are the published figures. Anything about how a specific recess detail performs thermally has to come from a calculation on your detail, not from this page.
Because invisible means different things to different people, the fastest way to a useful quotation is to say which of these you are.
| If your brief says | Where to start | Why, and what to confirm first |
|---|---|---|
| A 20 mm sightline | SL155, which carries our narrowest published sash figure at 44.5 mm | State which measurement you mean. Our published figure is sash width, not a head-on interlock sightline, and the two are not interchangeable |
| Flush floor with very large panels | SL126/182 (96), lift-and-slide, single sash to 2.5 x 3.0 m | Confirm the drainage design before the floor build-up is fixed, not after. This is the brief that goes wrong most often |
| A coastal or driving-rain elevation | SL112/167 with the 45 mm high-ｄｒｏｐ bottom track and the optional 85 mm water barrier | Accept that this is not flush. On an exposed elevation the dam is worth more than the flush line |
| Casement-grade sealing without a swing arc | A sliding swing door system such as the TS130 side-press, with eight-point locking and four-sided continuous sealing at 200 kg | Confirm the track count, because it decides whether the net opening is one third or two thirds |
| The wall should genuinely disappear | A pocket configuration, or glass folding doors | Recessing a frame does not make a wall disappear. Only removing the panels from the opening does that |
| An interior door that vanishes into the wall | That is the third meaning in the table at the top - a concealed or pocket interior leaf, in a different load class entirely, of the sort people specify as a sliding door for washroom or a wardrobe entrance | Confirm the wall build-up and the panel weight. It is not the same product family as a structural slider, and should not be quoted on the same drawing |
A specification gap is not a weakness if it is declared. It becomes a weakness only when it is filled with an adjective. Everything in the right-hand column below is a question we would rather you asked at enquiry stage than discovered on site.
| Item | What we publish | What you still have to ask for |
|---|---|---|
| Sash and frame width | Sash width and frame width per series, in millimetres | The interlock sightline dimensioned on a section, at the viewing angle, for your configuration |
| Load capacity | 200 kg, 220 kg and 400 kg per sash, by series | The maximum panel size for your specific glass build-up. Published maxima assume a glass thickness; a heavier unit reduces the maximum panel |
| Threshold | High track 43 mm, low track 8 mm, 45 mm high-ｄｒｏｐ, optional 85 mm barrier | Which of these is available on the series and configuration you are actually buying |
| Thermal | K 1.3, 1.6 and 1.7 on the AW90, C120 and C116. System values, not whole-product values | A calculated whole-door value for your sliding configuration and glazing, with the standard named. No sliding series carries a published K value |
| Corrosion and structure | 240-hour salt spray and 5,000 N shear resistance - both published for the AW90 only | Figures for the sliding series you are specifying. These results do not transfer between series |
| Hardware cycles | 30,000 cycles on CMECH hardware, tested on the C113 | The cycle figure for the sliding hardware you are buying, plus the test method. Rig cycles are a bench result, not a service life |
| Capacity | Two production centres in Foshan and Zhaoqing | Which line your order runs on, and what that line's current schedule looks like |
| The recess detail | Profile sections, drainage opening positions, track heights, fixing zones | The slab recess, the sub-sill pan, the external drain, the waterproofing and the structural head. We do not design or pour those, and we will not imply that we do |
| Ask | Why it decides something | What an unhelpful answer looks like |
|---|---|---|
| Dimension the interlock on a section, at the viewing angle | It is the only number left after the frame is buried | A round brochure number with no drawing behind it |
| Maximum panel size at my glass build-up | Published maxima assume a glass thickness | A maximum quoted without naming the glass it assumes |
| Roller rating, and how many wheels | Rating is per sash and glass-dominated; wheel count spreads the load along the track | "Heavy duty" |
| Which threshold heights exist on this series | Flush and water-holding cannot both be maximised | A flush promise with no drainage drawing attached |
| Drainage opening positions on the recessed section | Your floor finisher has to avoid them | "Standard weep system" |
| Minimum recess depth and width, in millimetres | It drives the slab detail, and it is irreversible | "Your contractor will work it out on site" |
| How a roller and a pane come out after installation | Decides whether maintenance is a service call or a demolition | Silence, or "it will not need replacing" |
| Whole-door thermal value, with the standard named | System values and product values are not the same number | A single K value with no standard and no glazing build-up |
| Salt spray and shear figures for this series specifically | Published results do not transfer between series | A test result quoted from a different product |
| What the factory will not do | Tells you who is carrying the risk on the recess | "We handle everything" |
No, and the word should not be used loosely. What is achievable is a recessed outer frame, which removes the frame from view and leaves the interlock. What is not achievable is a sliding panel with no meeting stile, because the overlap is where the lock, the seals and the anti-lift engagement live. If a quotation promises frameless, ask what is holding the panels in the opening under wind suction.
A narrow interlock is achievable within a weight ceiling, and that is the trade. Less aluminium in section carries less glass, so a very narrow interlock usually means a smaller maximum panel or a lighter glazing unit. There is a second, less obvious cost: an interlock that measures narrow head-on but steps out substantially when seen from the side has not bought you the sightline you think you paid for. Ask for the section, not the number.
Only by cutting the slab or raising the interior floor, and both are invasive. Published guidance from sliding door suppliers calls a slab poured without the recess the most expensive mistake in this category, because the remedies are grinding or rebuilding the floor. On a retrofit, a low-profile surface track is usually the better answer, and an 8 mm low track will get you most of the accessibility benefit with none of the structural work.
Usually several parties, which is exactly why it should be decided on paper first. The panel and its drainage openings are a product question. The sub-sill pan, the external drain, the fall on the terrace and the waterproofing continuity are building questions. The most frequent single cause is not a product failure at all - it is a floor finish or sealant blocking the drainage openings. Get the drainage route drawn, and get the finished floor level signed off against it before the finisher arrives.
It changes the performance of the assembly, and not always favourably. The frame-to-slab junction adds a linear thermal bridge that is not included in any door thermal value, so a well-insulated panel in a poorly detailed recess can still produce a cold line at the floor. Our sliding series do not carry a published K value, so ask for a calculated whole-door figure against your glazing and frame fraction, and treat the slab junction as a separate item for your designer.
Seven things, all in millimetres or as named standards: opening width and height; panel configuration and which panels move; target interlock or sash width with the measurement defined; threshold preference, flush or raised; glazing build-up; design wind load if you have it; and the finished floor levels inside and out. If you also tell us whether the slab is already poured, we can tell you quickly whether the recessed option is still available to you. Reach us through the ALPES contact page.
