ZARM Bremen Drop Tower Facility

Jurgen Howaldt, licensed CC BY-SA, via Wikimedia Commons.

Testing a manipulator, a separation mechanism or an instrument meant to work in free fall against Earth’s own gravity means removing that gravity somewhere on the ground, and no single ground method removes it for long, cleanly and cheaply at once. A NASA survey of the alternatives lines them up by that trade: a small drag-shielded tower buys 2.2 s at package sizes to 125 kg, a deep evacuated shaft buys 5.18 s at about 10^-6 g but only one or two drops a day, and an aircraft flying parabolas buys tens of seconds but at gravity quality two orders of magnitude worse than either tower [5]. The aircraft option was formalized at NASA Lewis as the DC-9 program: about 1e-2 g for 18 to 22 s per parabola, tens of parabolas a flight, or 1e-3 g for a few seconds free-floated, with the ceiling set by g-jitter from atmospheric disturbance, piloting and airframe vibration rather than by any hardware limit [7]. A design study for a crewed microgravity technology laboratory draws the same trade from the researcher’s side: what a facility can offer is traded against iteration rate, modularity and how much of the apparatus can be operated remotely, and a ground rig that a team can rebuild and refly in an afternoon is worth more to a maturing technology than one that offers better gravity but a slower cycle [9]. ZARM Bremen sits at one corner of that trade: the deepest and slowest end, at the price of throughput, unless the campaign is run on the site’s second machine instead. One building at Am Fallturm 2 holds both weightlessness machines and the hall that feeds them: the 146 m drop tower gives 4.7 s of free fall at about 10^-6 g in a 120 m evacuated tube, or 9.3 s in catapult mode, and the GraviTower Bremen Pro standing beside it in the integration hall gives up to 2.5 s at about 10^-4 g but 20 times an hour, the only ZARM machine that produces sustained lunar and martian gravity [1]. The constraints that come with those microgravity figures govern what can fly: a 100 g structural design case in the tube, no damping elements, no cantilevers in catapult mode, and a standing instruction that mass must not move inside the capsule during the flight. Both fly the same Standard Capsule, are booked through the same operating company, and are supported by the same two assigned engineers, so an experiment can move between them without being rebuilt.

ZARM has existed since 1985, with ground broken for the tower in May 1988, the catapult cellar already allowed for, the tower inaugurated in September 1990, and the catapult, designed by ZARM engineers, following in December 2004 [1]. The GraviTower has been in service since the beginning of 2022, and its partial-gravity mode entered the operator’s user guide only in version 1.5 of 8 July 2024, revised in version 1.6 of 22 April 2025, so that capability is still being extended. No total floor area is published for the site; ZARM also operates thermal vacuum and vibration laboratories and a 12.5 m, 30 g centrifuge, which are separate facilities and are not covered here [3].

Laboratory data

Parameter	Value
Operator	ZARM Fallturm-Betriebsgesellschaft mbH, University of Bremen
Location	Am Fallturm 2, 28359 Bremen, Germany
Commissioned	Tower 1990; catapult 2004; GraviTower 2022
Type	Evacuated drop tube and driven partial-gravity tower
Floor area	Not published
Capabilities	Drop tower, GraviTower, Capsules, Hall
Simulant or terrain	Not applicable; sealed capsules only
Instrumentation	Standard Capsule Equipment, NI PXI, 1 kHz logging
Ground truth	ASC 7.050LN.075 IMU, plus or minus 50 g, mandatory
Fidelity limits	Own 10^-6 g figure not verifiable by the standard IMU
Access	ZARM FAB mbH, worldwide, 5 working days on site
Cited by	No robot entry; one accelerometer campaign [2]

Capabilities

Drop tower

Parameter	Value
Working volume	Tube 120 m by 3.5 m, 110 m of fall; capsule payload 1017 mm high
Test article limits	Capsule 500 kg in drop, 400 kg by catapult
Vacuum	10 Pa; 18 pumps clear 1700 m3 of air in 1.5 to 2 h
Temperature	Not controlled; optional -20 to +60 C liquid circuit
Illumination	Not applicable; sealed capsule in an evacuated tube
Slope	Not applicable; payload center of gravity on the vertical axis
Gravity offload	Free fall, order 10^-6 g; 4.7 s drop, 9.3 s catapult
Instrumentation	Up to 6 Phantom Miro C321, 1920 x 1080 at 1480 fps

Drop and catapult are two operating modes of one rig, not two rigs: the tower is a 146 m concrete shell 8 m in diameter around a 120 m steel tube of 3.5 m diameter, with about 600 steps to the top, and the 110 m fall ends in a container holding 15 m3 of polystyrene pellets to a depth of 8.2 m [1]. In drop mode the capsule is released at the top and falls for 4.7 s; in catapult mode a pneumatically driven, hydraulically controlled launcher in the cellar throws the same capsule up the tube at 168 km/h after a 0.25 s acceleration, so the free-fall time doubles to 9.3 s, with the acceleration force computed for each experiment so that the capsule arrives as close as possible to the top of the tube. An independent user describes it as an “8 meters high braking tank filled with fine graded polystyrol” [2], and notes that residual drag behaves differently in the two modes [2].

The tube is held at 10 Pa, about 0.075 torr, by eighteen pumps that remove 1700 m3 of air in 1.5 to 2 hours and is reflooded with air after every flight, so the pumpdown recurs each time and sets the cadence: up to three flights per working day, each cycle four hours from handover to hand-back, with handover slots at 08:00, 12:00 and 16:00 [2]. The tube is not a thermal facility, either: capsule temperature is nominally room temperature but shifts by a few kelvin seasonally, falling in winter to a few degrees Celsius inside the tube at 120 m, and an optional closed-loop glycol and water circuit runs -20 to +60 C with 2 kW of heating and 2.3 kW of cooling at +20 C, disconnected about 1 minute before a drop and about 10 minutes before a catapult launch.

Mass limits differ by mode and capsule: payload mass is 225 kg with the Standard Capsule, 265 kg with the Long Drop Capsule, and 165 kg by catapult [1]. Catapult operation adds a balancing requirement that has no equivalent in drop mode: the capsule center of gravity must sit inside a 1 mm diameter circle about the vertical geometric center line, taring takes about four hours and is done by ZARM staff after the qualification review, and any disassembly and reassembly during a campaign must leave the center of gravity inside that circle or the capsule is rebalanced. Mechanically weak experiments that cannot be reinforced on site, and liquid-electrolyte batteries, are refused [2].

GraviTower Bremen Pro

Parameter	Value
Working volume	16 m tower; Standard Capsule envelope, 1017 mm high
Test article limits	Capsule 500 kg, payload 265 kg; 5 g design case
Temperature	Not controlled; optional -20 to +60 C liquid circuit
Gravity offload	10^-4 g to 2.5 s; lunar 2.5 s, martian 3.0 s at 10^-2 g
Instrumentation	Li-Fi telemetry to 250 Mbit/s; Phantom Miro C321

The GraviTower trades microgravity quality for repetition rate: it gives up to 2.5 s at the order of 10^-4 g, two orders of magnitude coarser than the tube, but it does it 20 times an hour instead of three times a day, the capsule opened in about ten seconds between runs so that power, fluids or components can be exchanged, and booked as a GraviTower half-day, “the minimal counting unit”, where four hours of use counts as one flight starting at handover at 08:00 or 12:00 and “it does not matter how many actual runs would have been realized in total” [1]. Against the tower’s four-hour flight cycle, that repetition rate is what makes parameter sweeps and statistical campaigns possible.

The machine is a winch, a rail and a shield: a commercial hydraulic winch system of more than 4000 hp drives a rail-guided slider up and down a 16 m tower, and the slider encloses the experiment capsule and acts as its drag shield, which is what removes the need for an evacuated tube [1]. It is also the only ZARM machine that produces sustained partial gravity, giving lunar conditions up to 2.5 s and martian conditions up to 3.0 s at a quality the operators describe as of the order of 10^-2 g and explicitly as preliminary. In microgravity mode the Release-Caging-Mechanism, developed and patented by ZARM, decouples the capsule inside the moving slider and recouples it at the end of the run; the operators describe it as capable of controlling heavy payloads “in a very smooth and precise manner”. Partial gravity instead runs with the capsule mechanically connected to the slider, without the Release-Caging-Mechanism, and drive vibration is the reason the number is what it is; see below.

The profile is a user parameter rather than a fixed characteristic: amplitudes of the acceleration and deceleration profile are set through a graphical interface up to 5 g in total, and the system computes the microgravity duration that results, so a gentler profile buys a shorter weightless window [1]. The structural design case is far gentler than the tower’s: “In GraviTower operation, max. accelerations of only about 5 g in total occur. Thus, exclusively GraviTower payloads may adapt to that level”. A simplified Matlab or Python trajectory script can be supplied before a campaign so that the kinematics can be checked against the payload’s own limits. Switching between microgravity and partial-gravity mode takes about one day.

The optional thermal liquid circuit is available through the capsule docking system but must be disconnected before each launch command and is reconnected immediately after each run. Operation is from a dedicated control desk beside the tower rather than the drop tower control room.

Capsules

Axel Hindemith, via Wikimedia Commons. Licensed CC BY-SA.

Parameter	Value
Working volume	Standard Capsule 1017 mm high; Long Drop Capsule 1780 mm
Test article limits	100 kg per platform including the platform; 50 kg point load
Instrumentation	NI PXI, 40 analog inputs at 16 bit, 1 kHz TDMS logging
Ground truth	ASC 7.050LN.075, plus or minus 75 deg/s and plus or minus 50 g

The capsule is the piece the two towers share: the same experiment can be flown in drop, catapult or GraviTower mode without rebuilding, with a payload envelope 1017 mm high in the Standard Capsule, 599 mm in diameter between the stringers and 700 mm wide off them, on a platform area of 0.359 m2, each platform carrying at most 100 kg including its own 15 kg with a point load limit of 50 kg at the center and mass distributed evenly [1]. The Long Drop Capsule raises that height to 1780 mm but is available in drop mode only, and the capsule interior itself is a sealed vessel at about 1013 hPa, with a specified pressure loss below 1 percent in 3 hours and an operational band of 980 to 1300 hPa outside which the run is aborted. The operators state that “all dimension limitations are hard limits”, and capsule internal pressure is monitored continuously as housekeeping.

Integration is on ZARM’s own aluminum, plywood and aluminum sandwich platforms. Drilling holes and threads is permitted; opening the outer contour is not, and machining is done only after consulting the drop tower engineers.

Integration hall

Parameter	Value
Test article limits	ZARM platforms, aluminum and plywood sandwich

The hall at the foot of the tower is where capsules are built up, qualified and, since the GraviTower’s commissioning, flown: the GraviTower stands inside it, unpressurized and in room air. Users must arrive at least five working days before the first flight for integration, ground testing and qualification, and a risk assessment by the experimenter is mandatory before work starts. No dimensions, crane capacity or cleanliness class is published for the hall.

Compared to other ground facilities

The Bremen tower is not the only ground option, and it is not the deepest one. NASA’s Lewis Research Center (now Glenn) operated a 5.18 s, 132 m evacuated shaft at about 10^-6 g, the same order as Bremen’s tube but with a shorter fall and, at 65 g over 0.15 s in a decelerator cart, a harder landing than Bremen’s roughly 50 g case [6]. The same center’s 2.2 Second Drop Tower and the Marshall Space Flight Center’s 100 m tower sit at the shallow end of the same family, trading duration for a simpler drag-shielded rig and a same-day turnaround [5]. None of the drop towers approach aircraft duration: the NASA Lewis DC-9 gave 18 to 22 s per parabola, forty-odd parabolas a flight, at a gravity quality two orders of magnitude coarser than any tower and set by airframe jitter rather than a physical floor [7]. Bremen’s own answer to that duration gap is not an aircraft but the catapult, which roughly doubles the tube’s free fall without leaving the ground.

What a facility is asked to reproduce also differs by where the disturbance actually comes from. A catalog of on-orbit microgravity disturbances measured on Spacelab, Mir and the shuttle middeck finds crew exercise, pump activations, latch motions and thruster firings all injecting transients from a few micro-g to tens of milli-g superimposed on a quasi-steady drag floor, none of which a sealed falling capsule or a parabola-flying aircraft reproduces [8]. A drop tower, a centrifuge and a parabolic aircraft each remove or add gravity by a different mechanism, so a result qualified on one does not stand in for a result on another: a penetrometry campaign validated for the MMX sampler was flown on parabolic aircraft rather than in a drop tower specifically because the several-second contact events it measured needed sustained low gravity with the operator able to intervene between shots, not a single few-second fall [10].

Instrumentation

The Standard Capsule Equipment is mandatory and is supplied and programmed by the facility, not by the user, a National Instruments PXI real-time system whose sequence is written in LabVIEW by the assigned ZARM engineer to the experimenter’s specification. It carries 40 differential analog inputs at 16 bit and 100 Hz to 10 kHz, 10 analog outputs, 8 RS-485 or RS-422 lines and one RS-232, 48 isolated digital inputs and 64 digital outputs, and every channel is logged at a common 1 kHz into per-minute TDMS files exportable to ASCII so a campaign’s data arrives on a single time base [1]. Power inside the capsule is 24 V from two 25 Ah lead-acid packs, capped at 15 A per line and 1500 W total across eight switchable lines, and a high-sensitivity inertial measurement unit, an ASC 7.050LN.075 with a 3-axis gyro to plus or minus 75 degrees per second and a 3-axis accelerometer to plus or minus 50 g, is mandatory in every capsule and logged on the capsule control system. The published GraviTower microgravity profile was measured with an iIMU-FCR-03 and the partial-gravity profiles with an ASC 4311LN [1]. External power, coolant and battery charging are all disconnected before the run. The standard capsule sensor set also reads ambient light, 0 to 1000 lux to plus or minus 50 lux, and illumination hardware is supplied as an integration service [3]. A separate Drop Tower Sensor Pack logs the internal capsule environment for the operators and can be released to the experimenter on request.

High-speed imaging is available at facility level. Six Phantom Miro C321 cameras give 1920 x 1080 at 1480 frames per second for 4.48 s, or 94,510 frames per second at 640 x 8 for 28.4 s, with 240 GB of internal memory and a 1 us minimum exposure [1]. Up to three Photron FASTCAM MC2 units give 512 x 512 at 2000 frames per second, and their live video can be relayed from the falling capsule by an infrared optical link at the top of the tube. The operators recommend switching the Phantom’s cooling fan off during a run to avoid disturbing the microgravity, and single the Phantom out for GraviTower work because its large local memory holds several recordings across successive runs without a download between them [3].

Telemetry differs between the two machines. The tower uses WLAN at up to 450 Mbit/s down the tube, with live remote control throughout the flight; inside a metal slider that will not work, so the GraviTower uses an industrial Li-Fi optical wireless link at up to 250 Mbit/s, with the capsule radio unit relocated to the top of the capsule structure [3].

Access is open to external users worldwide through ZARM FAB mbH, with two assigned drop tower engineers, one mechanical and one electrical, through preparation, integration and campaign. The operators state that lead time “varies from experiment to experiment”, with examples “with only days or a few weeks on the one hand and several months or even years on the other” [3]. No prices are published.

What it does not reproduce

Its own quoted microgravity level, as measured by its own standard instrument. The mandatory inertial unit (see the ground truth and fidelity limits in the table above) is not sensitive enough to confirm it, and the operators say plainly that proving the figure requires high-sensitivity accelerometers that are not standard equipment, pointing users to the MICROSCOPE literature for the measurement.

A constant residual acceleration. The 10 Pa tube pressure exists to hold residual drag down, and an independent user quantifies what it holds it to: 10 Pa “minimize[s] the drag effect at 200 um s^-2” [2], about 2 x 10^-5 g. The time behavior differs by mode: “in a standard drop, the drag due to the residual air in the tower is continuously increasing from the time of the capsule release to the end of the drop. During a catapult shot the drag decreases to zero at the apex of the trajectory i.e. at the very precise moment when the velocity is also zero” [2]. A drop degrades monotonically; a catapult shot is best in the middle and worst at both ends.

A gentle landing. Each payload must withstand deceleration peaks of about 50 g, and about 30 g of initial acceleration in catapult mode, treated as quasi-steady, with a factor of two recommended on top: “all capsule payloads shall be able to withstand accelerations of about 100 g (100 % safety margin included) in drop tower operation, at least” [1]. The GraviTower is the exception, at about 5 g in total, and the operators warn that a payload built to that level is a GraviTower payload only. This is the constraint that governs whether a robotic mechanism can fly in the tube at all. Independent measurement puts the deceleration at about 0.2 s and not exceeding 400 m s^-2 [2].

Compliance. The operators state that shock absorbers within the experimental setup “shall be avoided”, that industrial shock absorbers “might lead to an amplification of accelerations as they delay damping of the payload”, and that “the damping of a payload will be the best, the more rigid it is connected to the capsule platform”. For catapult operation, “cantilever beams or any kind of systems that might oscillate during the capsule acceleration phase must be avoided”. A deployable boom, a compliant arm or a suspension cannot be flown in its operational configuration.

Motion of the article during the run. This is the direct limit on robotics testing: “change of motion of masses during free fall shall be avoided. If this cannot be achieved, accelerations must be compensated by accelerating counterweights on or around the identical axis. Please consider the fact that you do not know about the exact location of the COG of the complete capsule in no direction” [1]. A manipulator or a mobility actuation exercised during the 4.7 s either carries a counterweight or degrades the microgravity it is being tested in. The requirement is written jointly for both machines, and thruster effects “must be avoided in the GraviTower as well”.

Thrust or venting. The vent line valve at the capsule lid must be closed before release or launch and all gases stored in onboard containers; “a pressure release is not valid during free fall”. In the GraviTower, gases are stored onboard and vented between runs.

Partial gravity uncoupled from the drive. This is the GraviTower’s own stated principal limit. Partial gravity is “only possible without the advantage of the RCM, since a mechanical connection of experiment and slider is required in the current development stage. This reduces the quality of the targeted acceleration level to the order of 10^-2 g, depending on the input kinematics and the experiment mass”, and vibration of the drive “might couple into the experiment” [1]. An article’s own mass is therefore a term in the accuracy of the gravity level it is tested at. No vibration spectrum is published. The microgravity figure is hedged the same way, resting on “preliminary measurements of the residual acceleration… during several dedicated test runs” at “the current stage of development of the GraviTower”. A campaign whose result depends on the residual acceleration floor belongs in the tube.

Gravity levels other than lunar and martian. Those two are what exists; further levels await additional test campaigns, and specific levels can be prepared on request for a dedicated campaign.

Both GraviTower modes in one campaign without a gap. The operators state: “Switching between microgravity and partial-gravity operation mode takes approximately one day. Thus, a break in the campaign must be considered if both operation modes are requested”.

Run-to-run identity in catapult mode. The free-fall duration is not fixed [1]: “Single periods of microgravity may vary some milliseconds from flight to flight in the catapult mode (up to 9.3 s). It depends on the actual pneumatic pressure that is applied for the specific capsule launch”.

The rotation a separating body leaves with. The MASCOT separation mechanism was qualified here and then flown to asteroid Ryugu, and the two halves of that pairing agree on one axis and not the other. Time to clear the support frame, a frame about 20 cm long internally, came out at 3.9 ± 1.2 s in the tower against 3.1 s from flight dynamics and 3.2 ± 1 s from the lander’s magnetometer: agreement, within the test scatter [4]. Rotation did not agree. The flight rotation period was 138.9 to 139.2 s, measured to ±1 s by MasMag, while the drop tower article spun with an 84.4 s period and almost purely about x [4]. A separation qualification in this tower bounds the clearing time and says nothing about the attitude the released body arrives in.

Vacuum, dust, terrain or thermal flux at the article. The article sits inside a sealed capsule at one atmosphere; the tube vacuum is not its environment unless the experiment brings its own vacuum vessel and evacuates it through the capsule vent line, and ZARM operates separate thermal vacuum and vibration laboratories and a 12.5 m, 30 g centrifuge for that work [3].

Campaigns run there

MICROSCOPE T-SAGE electrostatic accelerometer qualification, to June 2014. Twenty-five campaigns and 71 falls, 49 drops in free-flyer configuration and 22 by catapult, to assess the dynamic behavior of the twelve digital servo loops of the MICROSCOPE differential accelerometer; the flight models themselves were dropped only 8 times to limit the stress [2]. The capsule was 2 m high, 80 cm in diameter and about 300 kg with electronics and data acquisition, with a 50 cm cone at the bottom to stabilize it during deceleration, and the instrument axes were arranged so that the residual drag was measured along the vertical [2].

What the campaign established is also a statement of the facility’s limits. The drops are described as “the only way to test the servo control dynamics”, but the electronics had to be run at 100 V instead of 48 V, and the test mass bias raised from 41.4 V to 89.2 V, to generate enough electrostatic force to acquire the test masses inside the short free-fall window [2]. That widened the X full-scale range to 2.7 x 10^-5 m s^-2 against 1.6 x 10^-6 m s^-2 in the flight high-resolution mode, and the authors conclude that “since the instrument is configured in a coarse acquisition mode because of the short free fall time, its ultimate performances are out of reach” [2]. The catapult apex was used to estimate instrument bias, but the authors record that “this determination is not precise because the test-masses are not yet controlled along and about all directions”. Control loop dynamics can be qualified here; end-to-end sensitivity cannot.

MASCOT separation mechanism qualification, for Hayabusa2. The mechanism that pushed the MASCOT lander off Hayabusa2 above Ryugu was developed and qualified by dropping it here, across at least three drop test campaigns. The tower is where the mechanism’s release velocity was measured at all: optical tracking of the lander leaving the support frame gave 4.79 ± 1.69 cm/s (3 sigma) on the test article, scaled to 4.94 ± 1.7 cm/s predicted for the flight model after correcting for the flight spring rate, 0.21 against 0.19 N/mm, and the flight mass, 9.66 against 9.29 kg [4]. Flight gave 6.1 cm/s from flight dynamics, 5.9 cm/s from ONC image analysis and 6.0 ± 1.5 cm/s from the magnetometer, so the prediction was low, though the flight figures sit inside the test article’s own 3 sigma band.

Two facility-specific results came out of the same campaigns. Flight-to-flight scatter in eject velocity fell from 5.5 to 1.9 cm/s once the push-off plate tips were padded to force three-point contact, the original scatter being caused by manufacturing tolerances on hand-made CFRP parts changing the contact area [4]. And the residual scatter that remained is attributed to the tower setup rather than to the mechanism: the push-off spring preload could only be adjusted to 250 ± 50 N per flight, which is what moved release velocity over 4.26 to 5.25 cm/s between shots. A drop tower campaign measures the article and the rig together.

Other space robotics. The operators state generally that users “perform technology tests preparing and qualifying instruments for future space missions”, and offer the facility for that purpose [3]. Beyond the MASCOT mechanism above, no campaign involving a manipulator, rover, sampler or free-flying robot at this facility is published.

GraviTower. No GraviTower campaign is published. The operators describe the intended use generically, “to test preliminary setups or experiment components, to qualify new technologies for space and / or exploration missions, or to facilitate dedicated microgravity and partial-gravity research with a very high repetition rate”, but name no user or result.