Automatic Bake-Out Assembly (A.B.O.A.)

Tim RamosSRCCS/CHESS, Cornell University

July 29, 2011

Abstract

The purpose of the summer project betweenJune 5, 2011 and July 29, 2011 was to create afully automatic bake out assembly that wouldnot only automatically bake out an aparatususing ten separate heat tape/thermocouplepairs while comparing temperatures of heatzones, but would also be portable and easilymodified. This was to enhance the currentmethods where variacs are controlled manu-ally, which could lead to possible equipmentdamage and inefficient systems due to usererror.

Introduction

A bake out is a heating process which, with-out the use of chemicals, cleanses an appa-ratus and reduces the out gassing rate (orrate of dissolved, trapped, absorbed, or frozengas in a material being released (”Dictio-nary.com”)) of the volume that is to be placedunder vacuum. The apparatus to be bakedout is already cleansed before it is shippedto the lab and is also shipped in a cleanenvironment. However, a bake out removesthe majority of the particles on the molec-ular level in the apparatus that cannot beremoved by conventional cleaning methods.A bake out accomplishes this by using heatwhile pumping on the volume to acceleratethe out gassing rate that would naturally oc-

cur when the system is under vacuum alone.The bake out commences until the molecularout gassing rate is at a sufficiently low andstabilized rate. (”Alias Aerospace”)

A manually controlled bake out can lastfrom a few days to a couple weeks with con-stant monitoring. Automating the systemwould allow for a timely bake out with theability to adjust the ramping speed, soak-ing period, maximum temperature, and max-imum deviation of temperature between heatzones. Low deviation of temperatures be-tween heat zones is critical because if heatzone temperatures vary too greatly, one heatzone could thermally expand more than an-other causing stress on welds, valves, andother parts. Without human error, low devia-tion is more likely. The Automatic Bake OutAssembly (ABOA) must also be easy to useand should cycle through a bake out fasterthan a manual cycle. The ABOA should havethe ability to continue the cycle over a week-end or another period where there is no staffon hand.

The goal of this summers project was to or-ganize the assembly in which the automatedbake out components would be housed, trans-ported, and utilized. The automated bakeout system would heat ten elements, heat theelements at a given rate, and check the pres-sure inside of the element.

1

The Bake Out Process

The chamber to be processed is evacuatedwith a 2-stage system consisting of a turbomolecular pump, which exhausts into a back-ing pump (Figure 1). If no large vacuumleaks are present, a mass spectrometer calleda residual gas analyzer (RGA) can be turnedon when the pressure is approximately 10−6

torr. The RGA is an extremely sensitive de-vice which detects leaks at the level of the fi-nal pressure approximately 10−9 torr, beforethe chamber is baked.

Figure 1: Diagram of a 2-stage vacuum pumpsystem

Leak Testing

An RGA leak test is performed using a tankof helium gas and a ”wand” that can adjusthelium flow. The ”wand” is moved aroundthe seams, flanges, fittings, and welds of avolume under vacuum to check if the RGApicks up any sign of leaks. A leak would beindicated if there is an increase in the sig-nal given by the RGA. The RGA has a ”leakcheck mode” that looks specifically at a de-sired amu (in this case, helium).

External leaks, caused by mechanical prob-lems such as bad welds or gasket seals, arenot repaired by the bakeout and need to beaddressed beforehand. Once the chamber isdetermined to be leak free, an initial partialpressure vs. molecular mass scan is taken.The total pressure before the bake is usu-ally greater then 1−7 torr using a turbo pumpwhich initially turn on at about 1mT.

Residual Gas Analyzer Scan

The RGA scan is performed to see the molec-ular composition of the volume prior to thebake out. In this instance, the RGA is placedinto a mode that can detect a range of atomicmasses. (Note: RGA scans are always per-formed at room temperature).

An RGA is a mass spectrometer that mea-sures the chemical composition of a gas. Todo this, an RGA ionizes a gas to create acharge. In an RGA scan the gas being de-tected moves through an ionizer which ion-izes the gas using an electron beam. A massanalyzer then sorts the ions with respect toatomic mass by an RF quadropole that onlyallows certain ions to pass through basedupon its mass-to-charge ratio for a given fre-quency. The ions are then detected by ei-ther a Faraday Cup or an electron multiplier.(”Thomas Net”)

RGA scans are critical to the bake out pro-cesses at Cornell High Energy SynchrotronSource (CHESS). Since CHESS uses high in-tensity x-rays for their experiments, it is crit-ical to have clean parts with low outgassingrates that would not interfere with the x-raysand cause intensity loss. An example of anRGA scan (Figure 2) represents a graph ofan RGA scan on a Wiggler Beam Stop beforeand after a bakeout. Comparing the two im-ages in Figure 2 shows a sufficient decrease inpartial pressures of the ions and compoundsin the system, especially with large masses.

2

(a) RGA scan before bake out

(b) RGA scan after bake out

Figure 2: Graph of an RGA scan of a WigglerBeam Stop before and after a bake out. Thetop graphs in each image represents PartialPressure (torr) vs. Time (hr.) and the bot-tom graphs in each image represents PartialPressure (torr) vs. Atomic Mass Units (amu)

Wrapping

Thermocouples are attached to the apparatusto read the temperature of each heat zone.

The apparatus is wrapped in thick aluminumfoil, heat tapes are wrapped around the ap-paratus (with each heat tape coinciding witha thermocouple), and wrapped again in alu-minum foil (Note: Aluminum foil is used toinsulate the apparatus and thermocouples andalso transfers heat evenly from the tapes tothe object. The stainless steel of the appara-tus is a poor conductor while aluminum is anexceptionally good one).

/subsubsection*The Bake OutThe bake out begins when the apparatus

is ”ramped up” to a desired set point tem-perature via manually controlled variacs thatcontrol voltage to the heat tapes (Note: vari-acs are only used when a manual bake out istaking place). If previous data is available,it can be used to configure the rate at whichthe apparatus is heated by using small set-points. If there is no available data then aneducated guess must be used. Heated nitro-gen gas is passed through the apparatus toheat the inside components of the volume.Once the apparatus is at the desired set pointtemperature, it is left at that temperature foran extended period of time in what as knownas a ”soaking” period.

Finally, the apparatus is ramped down toroom temperature using educated guesses de-rived from the ramp up settings. Note: Itis OK for the heat zones to cool at a differ-ent rate than when ramped up as long as eachzone cools at a similar rate without too muchdifference between each zone. The ramp-uprate is also used as a guide to ensure the ap-paratus does not cool too rapidly and causedamage to the components. Once at roomtemperature, another RGA scan is performedand compared to the first. The second RGAscan should show a lower out gassing ratethan what was observed initially based off ofthe pressure in the volume. Also, out gassingof specific elements/compounds should de-crease. The apparatus is then prepared forstorage or is put into use.

3

A Need for Automation

As one can see, the manually controlledmethod is both time consuming and difficultto control. Figure 3 represents a manual bakeout and the time span in which it took place.Observing that the bake out process took al-most a month to complete it is obvious that abake out takes a large time commitment fromthe user.

Figure 4 (b) shows how the pressure in asystem varies during a bake out. The steepspikes in the graph represent a rise in pressurewhen the temperature is ramped up. Thisoccurs because of thermal expansion. TheABOA will eventually compensate for thisand hold heat zones until the pressure is at adesired point.

Figure 4 represents a period in the bakeout over a weekend where no changes wereperformed. An automated bake out couldcontinue control within this time span. Com-paring this time span to Figure 3, it is easyto see how much time would be saved withan automated system where no user input isneeded during the bake out.

Figure 5 shows an example of a prototypetest that was performed to test the iToolslogic program with a two channel automatedsystem. Although the system does not re-act to pressure, the goal is for a ten chan-nel system to show similar results. (Note:adding pressure to the system might cause a”stepping” to the temperature when observedclosely, but the overall ramp rate over an ex-tended period of time would be constant).

It can be seen that an automated systemnot only allows for a much quicker bake outprocess, but also allows for an unfamiliar userto perform a bake out and choose a spe-cific max temperature, ramp speed, and soaktime. Also, in a manual bake out, an oper-ator uses mostly educated guesses to deter-

(a) Temperature (oC) vs. Time (hr.) graph ofbake out

(b) Pressure (torr) vs. Time (hr.) graph of bakeout

Figure 3:

4

Figure 4: Zoomed in graph of manual bakeout between 5/29/11 and 6/1/11

Figure 5: Graph of two channel prototypeover a 12 hour period

mine how much power to send to the heattapes. This allows for a high probablility oferror. It is difficult to establish a constantramping rate and to ensure that each heatchannel ramps evenly. The automated sys-tem would make these constraints easier tosatisfy and would decrease the amount of er-ror and damage a user could cause.

The ABOA Project

Materials

• 10- Solid State Relays

• 10- Heat Sinks

• 12- Thermocouple Wires/Ports

• 1- 47”x19”x26” Equipment Rack

• 5- 3 Prong 15 Amp Comp. Plug PowerIn

• 10- 3 Prong 15 Amp Power Out

• 1- Eurotherm Mini8 Multiloop Temper-ature Controller

• 1- Vacuum Gauge Controller

• 1- Moxa Ethernet Converter

• 1- Ethernet Box

• 10- GFCI Circuit Breakers

• 1- 120V Rack Mount Power Strip

• 1- Cooling Fan

• 1- 4 Amp Fuse

• 4- Custom Made Rack Panels (1/8” Alu-minum)

• 1- 24V Din-Rail Mounted Power Supply

Automated Bake Out Logic Sys-tem

A logic program was designed with iTools tocontrol the temperature. The logic was de-signed to not allow for a change in tempera-ture with respect to distance (dT/ds) greaterthan a set amount. Similarly, the logic wasdesigned to not allow for a change in temper-ature with respect to time (dT/dt) greaterthan a set amount. The logic would heat up

5

all loops evenly and at the same rate. Also,if the pressure exceeds a given set point, theheating loops would be held at the controllerscurrent setpoint until the pressure is allowedto drop back to the desired set point.

The Eurotherm Mini8 was selected for thetemperature controller due to its ability tobe programmed to suit the project’s require-ments and conditions. Figure 6 representsa visual diagram in which the logic is inter-preted. In this ”circuit” the thermocouplesattached to the specimen conducts a volt-age with respect to the temperature of thespecimen. The Mini8 reads the temperaturefrom the thermocouples and then activatesthe SSRs which switches 120V AC power tothe heat tapes.

Figure 6: Layout of how the Logic HeatingSystem operates

Project Progression

Space Diagrams

To receive a visual of how everything wouldbe placed in the portable cabinet a space di-agram was created in Autodesk Inventor. In-ventor allowed for the creation of a 3D visualrepresentation of the project with the abilityto create, destroy, and manipulate parts to

the users liking in order to fit all of the mate-rials into the cabinet in a safe and convenientmanner. Figure 7, respectively, shows a front,rear, and isometric Inventor representation ofthe finished product.

(a) Front View (b) Rear View (c) IsometricView

Figure 7: Front (a), Rear (b) and Isometric(c) views of Inventor Space Diagram

Design/Order Needed Parts

Once a space diagram was created and a ba-sic layout was agreed upon, a design pro-cess was implemented to create certain pan-els that were not readily available. The SolidState Relays were too heavy to be held byDin-Rail alone. Therefore, panels were madeto support them. The Gauge Controller wasa panel mounted component, but the Mini8Logic Controller was mounted to Din-Rail.For convenience, and to save space, a platewas made to mount both of these compo-nents next to each other. Finally, the powersupplies, thermocouple ports, and the 4 AmpFuse had to be mounted to a panel for safetyand convenience purposes. Therefore, a sin-gle panel was created for all of these compo-nents.

Each plate was designed using Inventor inthe original space diagram to fit into the cab-inet. The Inventor file that represented eachpanel was converted into an Inventor DrawingFile (.idw) and then plotted for a machinistto create.

Wiring for the apparatus had to be doneas well. Each wire was cut, stripped, and sol-

6

dered to fit the given circuit. Twelve gaugewire was used for each cable that was switch-ing 120V. This included the cables going fromthe Power In to the GFCI and the cables fromthe GFCI to the Solid State Relays and tothe Power Outs. The 24V signal cables go-ing from the Mini8 to the Solid State Re-lays were 24 gauge twisted-pair wire(Figure8). (This circuit was repeated five times forthe final product.) Also, twelve thermocouplewires had to be stripped and assembled intoplug ports that were mounted to a panel andwired to the Mini8. Ten GFCI Breakers, eachrated for 10 Amps, needed to be ordered.

Figure 8: Circuit diagram for a two-channelrelay system

Preparation of Cabinet for Assembly

The cabinet needed to have a cooling fanmounted to it so that the components, espe-cially the Solid State Relays (SSRs), wouldnot overheat inside the cabinet. The coolingfan was mounted near the top of the assem-bly on the vents and close to the Solid StateRelays to expel the hot air that rose to thetop of the cabinet. Four 17/32” holes weredrilled so that the fan could be mounted us-ing four 6-32 screws, nuts and washers. Itwas found that the factory vents in the cabi-net were not sufficient enough to expel the airthat the cooling fan was trying to displace.Therefore, six 1/4” holes were drilled around

the circumference of the fan path to allow formore air flow.

Figure 9: 6 1/4” drilled out holes to allow forbetter air flow from cooling fan

Also, the cabinet needed two 17/32” holeson its chassis for the grounds on the circuitry.These were each drilled on opposite sides ofthe cabinet and were mounted with lock boltsand washers. The paint on the cabinet’s sur-face near the ground bolts was scraped off toensure a reliable electrical connection.

Assembly

Once all of the parts were set up, receivedfrom the machine shop, or received from com-panies, the ABOA was ready to be assembledin a pattern resembling Figure 7 and wiredwith respect to Figure 8. Note: Parts inthe machine shop were broken and on order.Therefore, the assembly of the A.B.O.A. wasput on hold for roughly one work week.

A panel mounted power strip was mountedto the racks of the cabinet on the rearside. The SSRs were mounted to the Heatsinks with a thin layer of silicon thermal

7

paste in between for sufficient heat trans-fer. The SSR/Heat Sink assemblies were thenmounted to the Din-Rails, which mountedto the custom support panels. The supportpanels mounted to the racks of the cabinet.The GFCI Breakers were mounted to Din-Rail and the rack. The SSRs and GFCIswere wired together, and all of the plugs,fuses, and thermocouples were mounted intothe custom panel for the power. The con-nectors for the plugs and fuses were solderedand shrink tubed. The wires were routed, sol-dered, and assembled and the ground wireswere mounted to the chassis of the cabinetusing the pre-drilled holes. Cable mounts, zipties, and wire loom were then used to orga-nize and route the circuitry (Figure 10)

(a) Wireloom aroundwires

(b) Zip ties to hold wirestogether

Figure 10: Examples of cable management

The panel supporting the Gauge Controllerand the Mini8 Logic Controller was mountedto the rack of the cabinet. The Ethernet con-verter and Ethernet box were mounted to thepanels of the cabinet and connected up to oneanother and the Mini8 via Cat5e cable. Allcomponents needing external power were con-nected up to the rack-mounted power strip

Safety Features

To ensure safe use of this assembly, the ap-paratus was designed with certain safety fea-tures. The Ground Fault Circuit Interupterbreakers (GFCIs), not only cut power if the

circuit is shorted or is sending more powerthan desired, but also allows for a manualshut off for each relay. The GFCIs, din-railpower supply for the Mini8 Logic Controller,and Heat Sinks were all protected by a FlexanPolycarbonate shield so that users have a lowrisk of accidently touching terminals that areexposed. (Figure 11)

(a) Heat Sink cover (b) GFCI cover

(c) Power cover

Figure 11: Flexan Safety Covers

Every wire connection has shrink tubingapplied to it so, once again, the user can nottouch any wires that may be exposed by un-foreseen circumstances.

As an extra precaution, all high voltage lo-cations have been labeled according to theamount of voltage each area is switching sothat the user is aware of where the high volt-age is located.

Results

Figure 12 shows the completed bake out as-sembly before the system was powered up for

8

the first time.

(a) Front View (b) Rear View

Figure 12: Front and Rear views of finishedproduct

On 6/29/2011, the ABOA was powered upfor the first time with nightlights in the PowerOut Plugs to ensure the heat tapes wouldfunction properly. Three specimens, a hollowaluminum pipe, a solid stainless steel pipe,and an aluminum block, were all mountedwith thermocouples, heat tapes, and wrappedwith aluminum foil. The logic was set toramp to a max temperature of 60oC with aramp rate of 30oC/hr and a maximum tem-perature deviation of 5oC. The Mini8 andheat tape channels 1 through 3 were con-nected up to mains power. Once the Mini8showed that is was receiving power, the iToolslogic program was started. The logic pro-gram successfully activated the relays whichswitched power to the nightlights and heattapes.

Channels 2 and 3 switched at an expectedrate while channel 1 switched power at anincreasing rate until full power was reached.This occured because the thermocouple fromthe power panel to the object in channel 1was wired incorrectly. Once this error wascorrected, the ABOA was powered up and allthree channels switched power at an expectedrate.

The bake out was monitored periodically

and it was found that all three channelsramped up, soaked for an hour, and rampeddown successfully with error. A temperaturedeviation of 10oC as opposed to 5oC was theonly found problem with the cycle.

Conclusion

The purpose of this summers project was todesign, fabricate, wire, and assemble an au-tomatic bake out apparatus that was safe,would be able to protect equipment, be mo-bile, and easy to use. The automated systemshould heat the elements evenly at a set rate,watch pressure, and complete a full bake outcycle faster than a manual method.

Once the ABOA was fully assembledand passed all safety inspections channels 1through 3 were connected to power and testedwith nightlights. Once it was clear that thecircuitry was working and the heat tapes wereconnected up to the system, a bake out upto 60oC with a ramp rate of 30oC per twohours and a maximum temperature deviationof 5oC commenced.

The bake out ramped up, soaked for anhour, and ramped down with very few errors.In the upcoming weeks more channels will betested and a pressure monitoring system willbe integrated.

Project Problems

During the design of the ABOA a few prob-lems arose. Wires for the circuitry were foundto be cut at insufficient lengths and were cum-bersome, some power outlets melted duringthe soldering process, and the power platewas too thick to hold the power outlets prop-erly.

If this project were repeated it would be ad-visable to wait until the components were as-

9

sembled in the ABOA before cutting the wirefor the circuitry to ensure sufficient lengths.14 guage wire would also be suggested to usefor the 120V wiring as opposed to 12 guage.The 12 guage wire proved to be cumbersomeand needed more heat to solder and thus al-lowed for the power outlets to melt during thesoldering process. The power outlets had tabsto hold them in the power plate. The plugswere originally designed to fit in a much thin-ner sheet metal than the 1/8” that was used.It would be suggested to use screw mountedoutlets next time instead of panel-clip poweroutlets.

References

1 ”Outgassing.” Dictionary.com. Dic-tionary.com, n.d. Web. 27 Jul 2011.http://dictionary.reference.com/browse/outgassing

2 ”Thermal Bake Out Vacuum Proce-dures.” Alias Aerospace. Panamet-ric Inc., n.d. Web. 27 Jul 2011.http://www.aliasaerospace.com/GOES-RTP190(Bakeout).pdf.

3 ”Residual Gas Analyzer.” ThomasNet. N.p., n.d. Web. 27 Jul 2011.http://www.thomasnet.com/articles/instruments-controls/analyzers-residual-gas.

4 ”CHESS Bake Out.” CHESSSignals. Web. 27 Jul 2011.http://signals.chess.cornell.edu/.

Acknowledgements

Special acknowledgement goes out to allwho made this project possible includ-ing, Bob Seeley, Chris Whiting, Eric Ed-wards, Jim Savino, Ernie Fontes, LauraHine, Cornell University, Cornell HighEnergy Synchrotron Source, Wilson Lab,

The Summer Research for CommunityCollege Students Program, and MatthiasLiepe.

10