Abstract
TMC has developed a practical method of rigidly supporting vibration-sensitive and heavy equipment in raised-grid cleanroom floors. The direct application has been in semiconductor fabrication and testing areas. The primary interest is to provide stable support for such equipment where elastomeric, or "spring", suspension systems are not required for actual attenuation of vibration.
A system of modular support stands and rigid platform tops is described. Accelerometers and a spectrum analyzer are used to make adjustments to the stands so as to optimize uniformity of support. Dynamic analysis is therefore an intrinsic part of the "system", and allows immediate evaluation of the overall vibration condition.
Introduction
The process of providing a rigid equipment support stand in a raised grind floor in a cleanroom, or of mounting a large optical table rigidly instead of on isolation mounts, always raises the question “Just how rigid is rigid enough?” We should also ask “What is practical given the site and use conditions?” The problem is that any structure we might think of in reasonable terms as “stand” built above the floor will introduce some new vibration components to the whole assembly. These new vibrations, high or low in frequency, will be due to the structural responses and relative stiffness of the stand carrying our equipment in question. “Rigidity” has its practical limits.
We could pour a solid concrete mass and have a nice rigid support stand. Or, we could design a welded, large cross-section steel stand, grout and secure that to the floor, and perhaps have a reasonably solid foundation for our sensitive piece of equipment. However, semiconductor facilities engineers are increasingly reluctant to make such semi-permanent additions to their valuable floor space. Likewise, the optics researcher typically does not have the option to construct such major solutions. The flexibility to relocate is also an important design and cost parameter, especially for the semiconductor facilities engineer. And, for practical reasons, there is often a compelling need to work around existing obstacles in the subfloor. Thus, the need is defined for a modular, yet predictable, equipment support stand.
The Modular Components
Here, we will consider an alternate, modular stand system consisting of adjustable, rigid supports and laminated, structurally damped platform tops. For heavy semiconductor equipment, the tops are typically 4-inch-thick laminations using steel plates epoxy-bonded in a sandwich construction with a rigid core. The top and sides of this platform are stainless steel for cleanliness and wear properties.
For the optical table user it is a matter of selecting a size and suitable thickness of the honeycomb type of table which will be appropriate to the needs. The construction should be of the highest density, highest modulus honeycomb core practicable, and be structurally well damped. Because these stands are so much more rigid, especially in lateral stiffness, than the conventional “legs” used with most optical tables, we would assume the optical table application would be for the most stringent stability requirement. Again, this is not with vibration isolation as we would achieve with a pneumatic suspension system, but rather made rigid with respect to the floor.
“Rigidity” relative to the motions of the floor depends on the structural stiffness of the stand, certainly, but especially on how well the supporting stand can be connected to the floor. In the modular design concept, each individual support stand is made rigid to the floor with the simplest, stiff, low stress condition: a tripod. Each leg is grouted to give more uniformity of support and contact to the floor. Remember, a few microinches of flexure at the base, or compression into the floor surface, will translate into a much greater amplitude up at the table surface. Tripod legs can then be lagged down, if necessary, for added security.
Consider that the monolithic frame stand must also, somehow, be perfectly, uniformly supported at its base in order to preserve its designed stiffness. Otherwise, its substantial mass and the equipment on top will contribute to some enormous amplifications, especially in horizontal “rocking” modes. It is exactly in the shortcomings of the installation process where we sometimes see the failure of this type of otherwise “rigid” frame design. While the mechanics of the process might seem simple, consider that some of the largest wafer steppers would have a footprint about 8 by 10 feet, might weigh about 6 tons, and might be installed in a raised floor 3 or 4 feet high. A monolithic stand built to a high design standard would, itself, weigh some 3 or 4 tons, and would have to be brought into the cleanroom by men wearing bunny suits and installed with essentially just hand tools. This is where the “practicality and feasibility” argument takes on special meaning.
In the modular stand system we depend on the high stiffness of the platform top to allow some choice in where supporting tripods may be placed. Tripods would be located under major load points, but need not be under all, especially secondary, load points. In fact, a little bending stress in the top actually improves its own flexural damping. In practice, we would therefore have tripod stands, used in some number of 4 or more, supporting a stiff, moderately well damped, laminated platform. The tripod is designed to have high frequency structural modes of its own, with a short exposure of large diameter (e.g. 1 ½ inch) jack screw at the top. Structural frequencies of a typical tripod would be on the order of 100 to 150 Hz, and it may have additional structural damping applied, depending on the height and leg span required for the application. Similarly, the platform top would have higher modal frequencies, and be much less better damped, than a single steel plate of the same size and mass. Most platforms for smaller equipment would have primary bending modes over 150 Hz. While larger platforms will have lower frequencies, the support points are invariably placed close enough together to break up the lowest frequency modes.
The figures below illustrate a typical platform arrangement with four tripods for a 36 inch high floor. Note the laminated structure of the top is covered with a stainless steel shell over the top surface and sides.
Once installed and adjusted, we would see a new, overall system resonance at a lower, moderate frequency which is typically in the 40 to 60 Hz region. This is due primarily to whole-body rocking modes on the floor, and is not unlike the observed performance of unit, welded frame stands. The difference with the tripod and platform system is that the region of resonance can be adjusted up or down a bit in frequency, thereby avoiding a direct coupling to standing frequencies of vibration in the floor. The performance is typically far superior to the scheme of trying to anchor individual support posts to the floor, and then carrying a piece of equipment just at its load points. Here, the lateral stiffness of the tripods and the rigidity of the platform have the effect of creating a far more stable plane of support than the individual posts. Remember, we’re talking about displacements on the order of a few microinches – a few hundreds of nanometers; individual posts may be waving around in the breeze compared to that. At least they will surely be moving relative to each other, and that has the effect of introducing a lower frequency component, and potentially much higher amplitudes, into the equipment vibrations.
Especially in the case of taller raised floors – 3 to 4 feet high – tripod stands will have potentially a much larger footprint at the subfloor level than a conventional stand of simple, rectangular construction. Smaller platforms in taller floors will therefore have the advantage of a more pyramidal shape, and the resultant greater lateral stiffness, than simple stands.
In a 4-foot-high floor, for instance, a 4 ft. by 5 ft. platform would likely have a 6 ft by 7 ft total base footprint. This larger base effectively reduces the overall torsional and rotational moments about the tripod feet compared to the contact points of a smaller simple stand. By extension, it is easy to see why it is so much more difficult to achieve any reasonable degree of lateral stiffness with individual posts anchored to the floor. The figures below compare a tripod and platform system with a conventional stand. Note the increased footprint and control of the center of mass of a piece of equipment with the tripods.
Methodology of Adjustment
Except for the initial process of leveling a platform or optical table, all further adjustments are guided by the “live” monitoring with suitable accelerometers and FFT spectrum analyzer. One accelerometer would be placed on the floor to monitor that spectrum – and prevent us from reacting to spurious events – and the other would be placed on the top. The spectrum of interest is up to 100 Hz.
If we assume the case of a simple, nearly square top supported on four legs, one can easily imagine the dynamic implications of adjusting the fourth leg to give perfect uniformity of support. Too high or too low with that leg and the top is in a torsion and the legs are not uniformly loaded. Fortunately, the vibration spectrum on the top changes significantly as the adjustment is made, so we monitor that “live” as our guide.
As we make the final height adjustment and pass through the point of uniformity of support, there is a reduction, or nulling, of peaks in the spectrum that are due to the torsional modes in the top and eccentric loading on the supports. The stiffness of even the heavy steel plate platforms is high enough, never mind the stiffness of the best optical tables, that the adjustment range is quickly reduced to a fraction of a turn of the jack screw (+- .02 in) in the final setting.
We would monitor the vertical direction first as we approach perfect support. Some simple dynamic testing – a solid thump of the fist on top, for example – is a clear “impact test” indicator of just how “solid” the mounting of the top is becoming as the adjustment is made. Most importantly, there should be no springiness to the stands in the region up to at least 30 Hz. Sustained building and ground vibrations would otherwise drive the structure with the potential for unacceptable displacements. Ultimately, the vertical spectrum should look much like the floor, without significant amplification up to at least 50 Hz.
Our experience and testing of a variety of support stand designs have shown us that most often the horizontal modes are best indicators of the overall “deadness” of support structures. As we approach the “perfect” support adjustment, we would therefore change to monitoring the horizontal modes on the table or platform.
There are, indeed, modes of whole-body motion that are amplified over the floor levels that cannot be entirely overcome. There are practical limits to the stiffness that can be achieved with the various components, and there are limits to the “perfect” connection that can be made to the floor. Ultimately, there are rocking modes of motion introduced simply because of phase stiffness as the waves coming through the floor strike first one leg and then another of any support stand. However, the lateral stiffness of the tripod mounting systems assures that the mode characteristics are reasonably uniform. That is, there is no significantly lower frequency mode of rotation of the whole platform about the vertical axis.
Once we feel we have reached a good adjustment, a further test is to bump the platform horizontally and observe that there is no low frequency response below the region of general amplification noted. We can also adjust any of the stands up or down out of its perfect range and see the same overall changes in the top spectrum.
The process of adjusting center supports under a large platform, or an asymmetric arrangement of supports, is simply an extension of the procedure already described. Keeping in mind that we don’t want to lift the platform off the primary supports already adjusted, we would adjust the new stands to a point of lift while still monitoring with the vibration analyzer. There should be a further slight nulling of the residual system “peak”. In any case, we are always looking for changes in the vibration spectrum that would indicate non-uniformity of support.
The vibration spectrum illustrated below is a composite of the measurements taken at three sites. The curves are representative of the horizontal response on the unloaded platforms and of the floor inputs. It can be seen that there is a generalized amplification in the mid 40 Hz region. Experience with a range of equipment weights on these platforms has indicated an improvement in damping the system “peak” under load, without significantly reducing the frequency. In other words, the support system remains relatively stiff to the load mass.
Additional Performance Options
As with any suspension system, however stiff, the resonance peak is followed in the transfer function by a reduction in throughput. The adjustability of the tripod stand system, combined with the stiffness and damping of the platform, gives us a method of skewing the normal “perfect” resonance peak to avoid coupling too closely with “spikes” in the floor spectrum. This can be done best with the platform under load. Changing a support adjustment, especially center or other inboard supports, will stress the platform a bit more and exercise more of the damping losses inherent in the laminate structure. The peak frequency will shift slightly, and characteristically, the resonance region will broaden a bit.
An alternative method of changing the suspension is to carry out the full adjustment procedure so as to bring all of the stands to the same plane, and then lift the platform up and introduce a selected elastomer between the tops of the stands and the platform. Typically, we would not attempt to create a low frequency suspension out of this, but would find a frequency region at which the floor is relatively inactive, and put the new resonance there. This practice might be done to reduce throughput vibrations at some bothersome higher frequency, or to help damp onboard vibrations generated by the equipment itself.
Adjustable tripod stands and laminated platform tops are rather simple mechanical solutions to the problem of providing sometimes large support stands for sensitive or heavy equipment. The elegance in this solution lies in the method of adjustment and the use of vibration analysis to do so.
Author: Bill Reid, TMC
Back to top^
Download PDF Version
<< Back to White Paper List