2. Making a Desk Reach the Heights You Already Measured

Two numbers, one desk

Mara's seated elbow height is about 28.5 inches off the floor with her chair cushion, her standing elbow height about 41. Those illustrative figures become a requirement a stranger could shop from: one surface that reaches both. The chapter separates frame height from finished surface height and does the arithmetic in both directions, including what a thicker butcher-block top demands of the frame beneath it, and how to tell which kind of number a listing is quoting. A walking pad adds a line: pads sold now sit roughly six to eight inches off the ground, so standing elbow height measured from the room floor no longer describes where the surface goes.

The failure people find after the box is empty is the minimum, not the maximum, and it follows from column construction. Two-stage columns bottom out around 27.5 to 29 inches bare; with a top, the lowest surface lands between 28.5 and 30. The Autonomous SmartDesk Core and Pro, ApexDesk Elite Pro, FlexiSpot's EN1 and EC1 and Desktronic's HomeOne all clear 41 inches easily and miss Mara's seated number by an inch or less. Three-stage frames — the FlexiSpot E7 at 23.8 to 49.4 inches finished and UPLIFT's C-legged frames — swallow it. IKEA's IDÅSEN, two-stage but inverted, shows stage count is a mechanism rather than a law.

Lift capacity is a torque number and says nothing about steadiness. Wobble-meter measurements put a V2's side-to-side wobble at 20 at 42 inches but front-to-back rocking at 33, and a Jarvis frame's rocking at 48 at 46 inches. A steel crossbar, thick cast feet with welded gussets, and a four-leg frame fight it; carpet undoes it. Warranties are read in pieces — frame, motor, electronics, top — and the tiers differ there. Buying a bare frame plus a butcher-block slab costs around $500 against about $310 finished, and buys wood, not a better frame. The full comparison with write-ups behind every desk carries commission links that did not buy the picks.

Getting it square, level and calibrated

Face down on padding, frame assembled loose, crossbar set screws left alone until after the top is bolted — tighten early and the columns go out of parallel and the desk trips its own collision sensor. Alternating bolts on the feet, screws checked against top thickness, motor plugs clicked fully home. Two people flip it by the frame. Level both axes and confirm every glide carries weight before power. Then drive the frame to its mechanical stop to reset it — five seconds held down on the FlexiSpot, about ten on an UPLIFT — because a preset pressed before calibration just flashes. Presets come from the measurements: 28.5 saved to one, 41 to two, each checked against forearms, elbows and shoulders. Load it, run it, type at the top — your own wobble test. Fifteen to twenty-five inches of cable slack, checked mid-travel. Then legroom, and the usable depth left once the monitor stands where it will stand.

Typing height is resolved and saved to two buttons. Screen height is not: a monitor on the surface rises with it. Tonight's record is one line — the two preset heights and the date.

Mara has four numbers and one free change. With the cushion on her chair, her elbows sit about twenty-eight and a half inches off the floor when she is seated and typing, and about forty-one inches off the floor when she stands. Those are illustrative figures for our worked example, not yours. But they are the two numbers that decide what desk she can buy, so let us turn them into a requirement she could hand to a stranger.

The requirement is short. The surface she types on must be able to sit at about twenty-eight and a half inches from the floor, and it must also be able to rise to about forty-one inches. One desk, both heights. That is what a sit-stand desk is: two motorized legs, called columns, that telescope up and down, with a flat top bolted across them and a small control panel under the front edge.

Now the part that trips people. The number a desk manufacturer prints on the box is often the height of the frame, not the height of the surface you type on. The frame is the pair of columns and the metal brackets across their tops. The top is a separate slab that sits on those brackets. So the surface you actually touch is higher than the frame by the thickness of that slab.

Do the arithmetic out loud and it stops being confusing. Say the top is one inch thick, which is normal for the pressed-wood laminate desktops most desks ship with. Mara needs the surface at twenty-eight and a half inches, so the frame underneath it must be able to come down to twenty-seven and a half. And she needs the surface at forty-one inches, so the frame must be able to reach at least forty. Now say she wants a thicker top, solid butcher block at an inch and a half. Then the frame must come down to twenty-seven and reach forty. Thicker top, lower frame required. That is the whole relationship, and it runs in both directions: if a listing tells you the frame goes from twenty-two to forty-eight inches, add your top thickness to both ends and you have the finished range. Twenty-three to forty-nine with a one-inch top.

Some manufacturers publish the finished range instead, with a stated top already included. Both kinds of number are useful. What is not useful is guessing which kind you are looking at. If a listing says a desk goes from twenty-three point eight to forty-nine point four inches, that is a finished height with a top on it, because a bare frame does not have fractional inches like that by accident. If it says twenty-two point eight to forty-eight point four, that is the same desk with the top taken off. When you cannot tell, look for a spec sheet or manual that names the desktop thickness, and check that the two ranges differ by exactly that thickness.

There is one more thing that shifts the range, and it shifts the top end. A walking pad — a slim under-desk treadmill you stand on and walk slowly while you work — is a slab that raises the floor under your feet. Current pads are not thin. The ones being sold now sit roughly six to eight inches off the ground. If Mara ever puts one under this desk, her feet are six to eight inches higher, which means her standing elbow height measured from the room floor no longer describes where the surface needs to be. Her standing elbow height was forty-one inches from the floor. Standing on a seven-inch pad, the surface needs to be about forty-eight inches from the floor to meet the same elbow. That is not a small correction. It is the difference between a desk that comfortably clears her requirement and a desk that runs out of travel with her shoulders shrugged.

So the requirement grows one line. Surface down to twenty-eight and a half inches. Surface up to forty-one inches for standing on the floor. And, if a walking pad is a real possibility later rather than a daydream, surface up to something in the high forties. Notice that this costs nothing today. It only rules out desks whose maximum is too low, and it is far cheaper to rule them out now than to discover the ceiling later with a pad already bought.

Now name the failure, plainly, because it is the one people find after the box is empty and the screws are in. It is not the maximum. It is the minimum. A desk that cannot go low enough is unfixable. You cannot shorten a column. You can raise a chair, but Mara already raised her chair, and her elbows landed at twenty-eight and a half inches as a result. If a desk's lowest finished height is twenty-nine and a half, she is typing an inch above her elbows forever, or she is buying a different desk.

This is not a rare defect. It follows from how the columns are built. A two-stage column is one tube sliding inside another: a single extension. Its total travel is usually about eighteen to twenty inches, and because the lower tube has to be long enough to hold that travel, the whole assembly bottoms out around twenty-seven and a half to twenty-nine inches as a bare frame. Add a three-quarter-inch or one-inch top and the lowest surface you can get lands somewhere between twenty-eight and a half and thirty inches. For Mara's twenty-eight-and-a-half-inch requirement, most of them fail, and they fail at the bottom.

Real examples make that concrete. Among the desks in our own comparison and the specifications behind them, the Autonomous SmartDesk Core and Pro bottom out above twenty-nine inches finished. The ApexDesk Elite Pro runs twenty-nine to forty-eight. FlexiSpot's own budget frames, the EN1 and EC1, sit around twenty-eight point nine to twenty-nine point three at their lowest. Desktronic's HomeOne bottoms at twenty-eight point six. Every one of those reaches forty-one inches for standing without breaking a sweat. Every one of them misses her seated number by an inch or less. An inch or less is exactly the margin nobody checks.

A three-stage column is two extensions instead of one: a tube inside a tube inside a tube. That geometry gives roughly twenty-five to twenty-six inches of stroke and drops the bare frame down to somewhere between about twenty-one and a half and twenty-four inches. FlexiSpot's E7, which is the desk our table treats as the benchmark for value, is a three-stage dual-motor frame with a finished range of twenty-three point eight to forty-nine point four inches. UPLIFT's V3, a three-stage frame on a C-shaped leg, runs twenty-two point six to forty-eight point seven finished. Either one swallows Mara's twenty-eight and a half at the bottom with inches to spare, reaches forty-one easily, and still has headroom for a walking pad later.

There is an exception worth knowing, because it shows that stage count is a mechanism and not a law. IKEA's IDÅSEN uses two-stage columns mounted inverted, so the wide tube is at the top and the narrow one goes to the floor, and it reaches a finished height of about twenty-four and three quarter inches. Two stages, very low minimum. So the rule is: use stage count to predict which desks are likely to fit, then confirm against the actual published range for the actual model.

Which brings us to the real lesson of a specification sheet. It answers a narrower question than the one you are asking it. You are asking "will this desk be steady while I type standing up." The sheet answers "how much weight can the motors lift." Those are different questions, and the second does not imply the first.

Lifting capacity is a torque number. The E7 and the V3 are both rated at three hundred fifty-five pounds. That tells you the motors will not stall under a heavy top and three monitors. It tells you nothing about whether the desk sways when you hit the space bar at forty-one inches, because swaying is not about lifting force. It is about how much the joints deflect sideways.

Here is the mechanism. At full extension, the amount of tube still overlapping inside tube is small — somewhere around four to six inches at forty-one inches of height. Inside that overlap are plastic glide pads that keep the tubes from grinding. Those pads have a tolerance, a deliberate gap, and a gap of a quarter of a millimeter down at the joint becomes a visible wobble up at the desktop, because the desktop is at the far end of a long lever. That is why a desk feels rock solid seated and nervous standing. Independent stability tests using a wobble meter, which pushes a known force at the desk and measures deflection, show commercial three-stage frames deflecting almost imperceptibly at thirty inches and noticeably more at forty-two. The same desk, the same frame, different geometry.

The tests also separate two kinds of movement people lump together. Side-to-side wobble is one. Front-to-back rocking is the other, and it is usually the worse of the two. On a standard UPLIFT V2 with a C-shaped leg, measured at forty-two inches, side-to-side wobble came in at twenty on a scale where anything under about twenty is imperceptible and the low thirties is merely fair — but front-to-back rocking scored thirty-three. On a three-stage Fully Jarvis frame, rocking reached forty-eight at forty-six inches, which is into the range you would call noticeable instability. Both are decent desks. The rocking is a consequence of shape: a C-frame deliberately pushes the column back so your knees have room, which lengthens the lever between where you press and where the desk is anchored.

So what actually fights the wobble? Three things, and only one of them appears in a headline. A steel crossbar tying the two columns together is the single most effective part for side-to-side stiffness, and its cost is that it sits near knee height, where a tall person's shins find it while seated. Thick cast feet with welded gussets resist bending far better than thin stamped base plates, which flex under a load hanging out in front of them. And a four-leg frame, with four columns in a box, arrests both directions at once; independent testing of four-leg builds under heavy loads at forty-one to forty-four inches shows almost no monitor sway. Finally, the floor. A rigid frame on high glides sitting on thick residential carpet will rock as a whole unit no matter how good the frame is.

None of that is knowable from the word "sturdy." What you can establish from a manual is the documented height range, the desktop thickness, the load rating, the column stage count, whether there is a crossbar, and the desktop's material and dimensions. What you can establish from an independent measurement is deflection at a stated height under a stated force. What a headline lift capacity leaves entirely open is stability, longevity, and how the thing feels under your hands in month eighteen.

Two more specification lines deserve a careful reading. The first is the controller and its memory presets. A preset is a button that drives the desk to a saved height, and four presets is the useful number: seated, standing, and two spare for a second person or a walking height. Both the E7 and the V3 offer four. The V3 adds a six-axis gyroscope for anti-collision, which watches for tilt as well as for a current spike, and stops and reverses an inch or so when it meets an obstacle. Useful, and also a thing that will stop your desk for reasons you do not understand if the frame is not sitting square.

The second is warranty, and it must be read in pieces. Electronics are the part that fails. A fifteen-year frame warranty with two years on the control box is not a fifteen-year desk. Read frame, motor, and electronics as three separate terms. UPLIFT publishes fifteen years across all of it, including the control box, the keypad, and the desktop. FlexiSpot publishes fifteen years on frame, motor, and electronics for the E7 tier specifically, with the desktop covered two years for chipboard and five for solid wood, and the budget series drops to two to five years overall. DeskHaus publishes twenty years on frame, motor and electronics with one year on the top, original purchaser only. Vari's ComfortEdge caps electronics at five years. Those differences are the real price difference between tiers, and they are invisible if you read the word "warranty" and stop.

Now, the other route. You can buy a bare frame and a separate top instead of a finished desk. It is worth pricing honestly, because the frame alone does not include as much as it looks like it does.

Take the FlexiSpot E7 both ways, at prices recorded in early to mid 2026. The complete desk with a factory forty-eight by twenty-four inch laminate top ran about three hundred ten dollars, with the mounting holes already drilled and the desktop under warranty. The frame alone ran about two hundred ninety-nine. So the frame is nearly the whole price of the finished desk, and then you start adding. A pre-finished butcher block slab around fifty by twenty-five inches runs about a hundred sixty dollars. Threaded inserts and bolts, or wood screws, run fifteen to twenty-five. A cable tray or raceway runs twenty-five to forty-five. If the slab is unfinished you are also buying polyurethane or hardwax oil and sandpaper, another thirty-five to fifty, plus the hours. Total, about five hundred dollars. That is roughly a hundred fifty to two hundred more than the complete desk, and what the money buys is solid wood instead of laminate over particle board, plus a top you chose the size of. It does not buy a better frame. It is the same frame.

You also need a drill and a brad-point or Forstner bit with a stop collar to set threaded inserts without going through the surface, which is free if you own them and thirty to sixty dollars if you do not.

For Mara, a finished desk is the better answer and the reason is her space, not her taste. Her working depth is twenty-four inches. A factory forty-eight by twenty-four top is exactly the footprint she has, already drilled, already covered. Buying a fifty-inch slab would mean an overhang she has nowhere to put.

It is also worth saying that some people do not need this step at all. If your existing desk already sits at your seated elbow height and you never intend to stand, the surface step is already satisfied, and the money belongs somewhere else in the build. If a perfectly good desk has one wobbling leg or a stripped bracket, repair it and move on. Mara does not get that answer, and the reason is arithmetic: her kitchen table is a fixed thirty inches, an inch and a half above her seated elbows and eleven inches below her standing ones. No cushion, footrest or book stack changes a fixed leg. This is the step in the build where a purchase is genuinely the constraint being removed, and that is precisely why we spent two chapters measuring before we said so.

Our own comparison of these desks lives at ocdevel dot com slash ergo slash desks, and it is the deepest of our tables — every desk on that page has a full write-up behind it, not just a row of specifications, so the rankings there are ours and we will stand behind them. Those links earn a commission, and a commission has never bought a pick; the E7 sits where it sits because of its range, its stage count and its warranty split, not because of a referral rate. Prices move, so treat every dollar figure here as dated rather than current, and check the page for what a desk costs the week you buy.

And one thing no desk earns you: a promise about how your body will feel. A surface at the right height removes a measurable geometric problem. That is a real thing to fix. It is not treatment, and if discomfort persists after the geometry is right, that is a question for a qualified clinician, not for a more expensive frame.

Getting it square, level and calibrated

Say Mara orders the complete E7 in forty-eight by twenty-four. Two boxes arrive, one heavy enough that she should not carry it up stairs alone. Here is the work, in the order it has to happen.

Lay the desktop face down on the carpet or on the flattened packaging. Everything gets built upside down, on the underside of the top, and the padding is what keeps the finished face unscratched.

Assemble the frame loosely first. The columns get bolted to the crossbar assembly, and the crossbar on this style of frame telescopes so it can match the width of the top. Leave the telescoping set screws loose. This is the step that people tighten too early and it causes real trouble: if you lock the crossbar's length before the frame is bolted to the desktop, you force the two columns out of parallel. The desk will then bind as it lifts, groan on the way up, and trip its own collision sensor for no visible reason. Loose crossbar, then bolt to the top, then tighten.

Fasten the side brackets to the tops of the columns, and the feet to the bottoms. Tighten the foot bolts in an alternating pattern, opposite corners, the way you would tighten a wheel, so the plate seats flat instead of cocking to one side.

Now center the frame on the top. A factory top has pre-drilled holes, so this is alignment rather than measurement. Drive the screws, checking that their length does not exceed the thickness of the top — a screw a quarter inch too long comes through the working surface and there is no undoing that. Then tighten the crossbar set screws firmly.

Wiring is short. The control box mounts near the center under the top. The keypad goes under the front edge where a hand actually falls; for Mara that is the right front corner, because the left side is where her laptop lands. Each column has a motor cable, and each goes into its labeled motor socket until the retention clip clicks. Not almost. A partly seated plug blocks the sensor feedback the controller uses to keep the legs in step, and the result is one leg driving, a tilted top, and an error code. The handset cable goes into its own socket, and the power cord last.

Then flip it. Two people, both gripping the metal frame and legs. Never pivot a sit-stand desk by the overhanging edge of the top; you are levering the brackets against the screws you just set.

Level it before you power it. Put a spirit level across the crossbar for side to side, and across a side bracket for front to back. Each foot has a threaded glide you turn to raise or lower that corner. Turn until the bubble centers on both axes, and then press down on each corner in turn to confirm all four glides are actually carrying weight. A glide hanging a hair above the floor is not a cosmetic problem. It lets the foot rock as the motor takes up torque, the controller reads that as tilt, and the desk simply stops.

Now calibrate, and do this before touching a preset button. A dual-motor controller keeps the two legs in step by counting pulses, and it needs to find bottom once to know where it is. On the FlexiSpot, hold the down arrow until the desk reaches the bottom, then keep holding another five seconds until the display flashes its reset code; the frame will dip, rebound slightly, and beep or show a height. On an UPLIFT, hold the down arrow about ten seconds until the reset code appears, release, then hold down again while the desk drives to its mechanical stop, bounces back a millimeter or two, and settles on a numeric height. If you press a preset before this, the controller sits there flashing at you and you will assume the desk is broken.

Presets come next, and they come from the measurements, not from a chart. Drive the desk until the surface is at Mara's seated elbow height, twenty-eight and a half inches, then check the actual thing being fixed: forearms roughly parallel to the surface, elbows a little over ninety degrees, shoulders down. Save that to preset one. Drive up to forty-one, check the same three things standing, save that to preset two. On either desk the sequence is the same shape: press the memory button, then the number, within a couple of seconds.

Then load the desk and drive it once. Put the monitor on it, the laptop, whatever else lives there, and run it from seated preset to standing preset and back. You are watching for three things. Does it reach both saved heights carrying the load, or does it stop short and complain? Does the top stay level as it goes, or does one end lead? And at the top of the travel, put a hand on the desk and type a line — that is your own wobble test, at your own height, with your own load, which is the only test that settles the question for your desk on your floor.

Watch the cables through the whole travel, not at the ends. A desk that rises twenty inches needs at least that much slack on anything plugged into the wall, and fifteen to twenty-five inches of loose cord is the working figure. Mara's outlet is behind her, which is the good case. Run the cords along the crossbar in a tray or a sleeve rather than letting them hang free, because a loose cord near a telescoping column can be drawn into the seam between tubes on the way down. Then drive to standing and look for a cord gone taut, and drive to seated and look for a loop resting on the floor where a chair castor will find it.

Last, sit down and check the legroom, because a twenty-four-inch depth is not generous. Lower the desk to the seated preset and look for at least a couple of inches between your thighs and anything underneath: the crossbar, the control box, the cable tray. On a frame with a crossbar at knee height this is a real conflict for taller people, and it is better to find it now than after the monitor is mounted. Then measure the usable depth left on top once the monitor is standing where it will stand. That number matters more than the twenty-four inches on the box, and it is the number the next step of this build starts from.

Mara's installed cost was nothing going in and is now the price of one desk. Her typing height is resolved, seated and standing, and saved to two buttons. Her screen height is not: a monitor on the surface follows the surface up, so the height that worked seated has to be checked again at forty-one inches. Her keyboard and pointing device are still a laptop's own. Both of those come out of the numbers she already has.

Tonight's job is one line in the record: write down the two preset heights, twenty-eight and a half and forty-one, next to the date. When the desk drifts, or someone else touches the keypad, that line is how you get it back.