the quest begins... November 5th, 2005

It all started innocently enough... make a Low Temperature Differential (LTD) Stirling engine from knowledge that is readily available in the public domain. I quickly realized that Stirling engines were at the nexus of science, the outer fringe element, and a couple of new dimensions I was not familiar with. Along the way, I sought the advice of people whose opinion I respect at a certain "bomber plant" where I work, did my own research, and suffered the prior art of wack-os on the net. I gave them their due, and sifting through it all, arrived at my own opinions which I humbly submit for your approval- the Stirling Cronicles.. note: people looking for a do-it-yourself "how-to" have obviously come to the wrong place!

ball bearings May 9th, 2006

VXB kit665 4 Flanged Ball Bearing 3/32" x 3/16" Ceramic Balls - Stainless Steel - Metal Shields $24.95 McMaster-Carr part numbers: 3826T5 $10.48 each Ultra-Precision Mini Ss Ball Bearing - ABEC-7 Open 4259T4 $11.86 each flanged Ultra-Precision Mini Ss Ball Bearing - ABEC-7 Open myonic RMB bearing part number ULK 3006 UL 3006X The fit and finish of these bearings is noticeably better than the VBX bearings with less wobble. Due to their open ended design (no side shields) they have no fatory applied lubrication.

soldering December 27th, 2005

I went to a local Fry's Electronics store the day after Christmas (appropriately located near MacArthur Park). The first hint of trouble was that available parking spaces were out much further in the lot than usual. The second came as I noticed the returns line was backed up to the entrance door with several hundred people. I knew I needed to hustle before all that busted/defective merchandise made it back onto the shelves. This seems to be the flagship Fry's in the area since it has moderated some of the goofier aspects of other local stores---a restrained theme , no guys on ladders pointing to open registers, no cashiers waving ping pong paddles, only a vestigial grocery store (but they do have a fancy in-store restaurant). In fact, everything looks relatively normal except for what would appear to be an eight foot Steinway concert grand fitted with a cheap sounding electronic player piano device (a genuine Steinway of this type retails in the $100,000 range). I perused their large selection of surface mount resistors and integrated circuits, but for the life of me I can't figure how hobbyists are installing these things without automated placement machines and solder reflow capability. Maybe pre-tin the PCB pads and position the component with tweezers and hope you hit the right component lead with the iron to tack it down. surface mount components shore up my soldering capability by purchasing a Weller WD1000 digitally temperature soldering station soldering station. 35 registers open (I don't know what it takes to get all 70 checkout registers manned===probably have to hire more people).

Toto---we're not in Home Depot anymore December 26th, 2005

Enter the world of W machine tools, the land that modern marketing forgot, with narry an RFID SKEW nor UPC in sight. Much like Disneyland, you need to allocate a whole day for even an exploratory walk through of this place. Upon crossing over into used tool steel land, I came upon a table that I figure must have about a thousand pounds of steel piled on it. There is no good way to sort through this mountain of stuff without a fork lift or something, but luckily in an upper stratum I found a 1/2" X 1" X 6" lathe tool with a custom ground profile on the 1" side. There were several other similar pieces, so I figured this must be a standard lathe tool size for large industrial machines (although I would have thought the 1/2" inch side would have been the normal cutting edge). Most were engraved with the bomber plant ID (they buy a lot of our stuff at surplus auctions). I took my selection to the checkout counter where I was told there was a ten pound minimun purchase from that table. I think there was a sign to that effect, but there are so many signs scattered around (most admonishing the customer to not do something with dire consequences) it is hard to keep track. BTW--there are at least three signs on the front door with their operating hours all of which say different times---through trial and error, I have found that if you get there and a giant padlock is not on the door---they are open. Anyway, when I told him there was no way I was going to make it to ten pounds, he stared off into the distance squinting and a price of a few dollars popped into his head. When I got this bruiser home, it easily hogged out deep cuts in a 6061-T6 aluminum flywheel. Moving on to type 303 stainless steel, I was pleasantly surprised with its performance cutting a very precise square bottomed hole in 1" SS rod for my pyrex cylinder base. I had always had a problem with the tool chattering when attempting cuts such as this, but I think that the large mass of this tool dampens oscillations before they get started. It was all the more impressive as there is about 1/4 inch of exposed cutting edge. It is also high centered since my tool post is set up for 3/8" square tools.

3-D modeling December 24th, 2005

I used the Persistence of Vision freeware raytracer to visualize the geometry of the moving parts through animated renderings of a 3-D engine model. Here is the source file if you want to try it on your machine (you might have to sift through some extraneous stuff in there that I haven't cleaned up yet). I also used renderings from the model to print full size templates for making some of the engine components. You would need to calibrate your printer settings to get proper scaling for this. You also need to use "orthographic" camera settings when doing this to prevent dimensional distortion inherent in perspective projections. Note that although modeling the sinusiodal motion of the power piston and displacer linkages is not rocket science, it can be a bit of a brain teaser since all of the motion translations must be referenced from the moving crank pins. It appears that the lost motion slip-linkage could possibly be modelled using megapov mechsim , but I haven't tried it yet.

Fluorosint December 19th, 2005

Fluorosint 207 mica filled PTFE.

"loosey-goosey" principle November 26th, 2005

In researching the web, I was amazed at the number of contraptions that successfully employ loosely connected linkages of thin wire. This must be especially galling to high precision machinists as it would seen to vindicate hastily thrown together claptrap. Lacking a formal name for this, I dubbed it the "loosey-goosey" principle. Upon further consideration, thin music wire with a low coefficient of friction and minimal surface contact (heck, part of the time the linkage elements don't even touch each other!) is indeed a low loss mechanism. I concluded that while inappropiate for machines that produce useful power (most of what we see), it has its place in lightly loaded mechanisms such as laboratory beam balances and LTD Stirling engines.

Vespel^® November 30th, 2005

from the Vespel^® design handbook Genuine Dupont low friction filled Vespel^® formulations: coefficient of friction static dynamic SP-3 15% molybdenum disulphide - .25 SP-221 15% PTFE SP-21 15% graphite .30 .24 SP-22 40% graphite .27 .20 SP-211 15% graphite, 10% PTFE .20 .12 Solvay Torlon 4301, 4275, 4435 This materials amazing high temperature performance is of little value in this application, but I think it is the ultimate in machineability and the best combination of hardness and low friction coefficient. Being a polyimide, I think that it is related to Kapton^® film which I am familiar with. coefficient of friction Now here is the question: Is this the generic equivalent of one of the above Genuine Dupont products? The suppoting evidence is: 1) polyimide is a generic term for Vespel^® (aka Kapton^® in thin film format) 2) The product pictured is black---I believe the unfilled Vespel is tan and all other formulations are grey-black At these prices they really shoud be more precise as to what you are buying even if it is "brand x". Vespel^® and Kapton^® are registered trademarks of E. I. du Pont de Nemours and Company

Torlon^® December 11th, 2005

At lunch, Mr. B tells me that Torlon^® is used in main langing gear uplock rollers on high performance fighter aircraft such as those currently made at the bomber plant and per his evaluation, it meets or exceeds Vespar^® in several key parameters. These parts see extreme forces applied to them during high "G" manuevers. A little research on the web reveals that Torlon^® 4301 graphite + PTFE filled bearing grade material is best suited for this application and unlike Vespar^®, is readily available without a private consultation with a plastics company applications engineer. I move this material to the top of my list. Torlon^® is a registered trademark of Solvay Advanced Polymers Vespel^® is a registered trademark of E. I. du Pont de Nemours and Company

displacer November 26th, 2005

Frustrated from trying to decipher various characterizations of foam used for Stirling displacers, I just went out and bought some. I got a 4' X 8' sheet of blue 1/2" thick Dow Styrofoam* sheathing insulation from Lowe's. I sorted through the lot to find the flattest specimen and cut it in two pieces in the parking lot to fit in my hatchback for the trip home. I transferred hole locations to the foam from a full scale template printed from a 3-D model by piercing holes through it with a pin. I drilled the center hub and regenerator holes with Forstner drill bits to minimize tearout of the foam. I was surprised to find that a regular twist drill wont go through this stuff cleanly. With the foam sandwiched in a custom made Delrin^® arbor chucked up in the drill press, I spun it around at low speed while sanding the outer perimeter to assure it was perfectly round and concentric about the center hub. At the time I was kinda worked up into a paranoidal fit after reading about Dr. Senft's measures to prevent dreaded "latent displacer warpage". I later came to my senses and reasoned that since I have a ten year supply of the stuff, I will take my chances and just make a new one should it occur. Using a 3Mª Rolocª two inch arbor and sanding disc, I removed thin layers of material in sucessive steps flipping over the sides periodically to maintain maximun flatness. I finished it off with the drill press quill lock set at the required .3125" final thickness specified for the Senft N-92 engine. I found that the rigid attachment of these particular abrasive discs to the arbor allows for extremely flat surfaces of uniform thickness to be achieved. ^®ª* Trademark of The Dow Chemical Company Delrin^® is a registered trademark of E. I. du Pont de Nemours and Company

slip-link November 25th, 2005

Early on in my experiments, I noticed that by moving the displacer by hand, with proper timing, I could get much more power out of the engine. Of course, in doing so you are adding energy to the system, thus relieving the engine of the inherent losses in the displacer mechanism. Still, my gut feeling is that I am getting more out than I am putting in. In particular, if you rapidly shuttle the displacer to the oppostite end of the chamber as the crank pin is rounding top or bottom dead center, the timing will be such that the connecting rod will supply maximun force on the crank pin as it enters its mid-stroke position to achieve maximum torque. Of the four engine designs I have made so far, all versions have exibited this behavior, so it is not a quirk related to any particular design. Applying my pea-brained logic to this phenomenon, I conclude that this unutilized thermodynamic energy is partially recovered by the so called "lost motion" linkage to increase the dwell times of the displacer at the top and bottom plates. This notion is reinforced by the following factoids: 1) Certain low speed steam engine designs can operate in a so called "expansion mode" in which the time the intake an exhaust ports are closed is prolonged (increased outer valve lap) to extract the maximum amount of thermodynamic energy out of the expanding gas in the cylinders. A regenerator equipped Stirling might be able to carry some unused heat energy forward into the next cycle, but the lesson to be learned here is to convert all of the thermal energy into mechanical motion while you've got it (IOW---use it or loose it). 2) The idealized Stirling cycle has four discrete segments with squared off edges on the pressure-volume graph. Since the area enclosed by the P-V curve represents the thermodynamic energy available to the engine per cycle, any rounding off of the sharp edges represents unrealized heat energy conversion. If I ever get the instrumentation done, it will be interesting to see if the lost motion linkage generates a P-V curve closer to the ideal Stirling cycle. 3) The sinusoidal displacer motion of conventional Stirling designs is a consequence of the spinning crank motion. This is purely a mechanical conveinience with no thermodynamic advantage. 4) Last but not not least, Kolin and Senft already figured this out as it applies to Stirling engines and published it years ago! Using the Senft P-19 as a prototype, I scaled the print to get a rough starting point on some critical dimensions. I get about a 1/4" horizontal "lost motion" slip. Unsure of the exact purpose of the vertical motion component as it seems to get cancelled out in the final analysis, I ignore it for now. The bellcrank pivot bearing appears to be offset relative to the axle shaft about 1" vertically and 2" horizontally. I am not too worried about the horizontal offset on my machine right now since it is variable over a wide range by loosening a set screw. escapement drops to bottom loosey-goosey principle paper clip starts working (it always the trick paper clips always work better than the real thing!) slipping at both ends Studying the phasing of this mechanism, I notice that the power crank pin and displacer crank pins are brought back into alignment (or 180 degrees out depending on which direction you want to go) with linkage. My first theory attitude Unstable analogous to a "flip-flop" with two stable states (just so you know, I view everything in electrical terms) friction to achieve a precise balance

Delrin^® November 23rd, 2005

I ordered some black acetal rod from smallparts.com. Judging from markings on the product I received, this appears to be unfilled acetal sourced from Quadrant Engineering. From the Quadrant material spec sheet, this product would have a dynamic coefficient of friction of about .25 when in relative motion with steel. For moving parts in relative contact, DelrinAF^® Teflon^® filled acetal with a dynamic coefficient of friction of .19 would be an improvement, particularly for the displacer gland and the cone shaped spacers for the bearing races. The displacer gland with its oscillating motion would seen to benefit from the improved "slip-stick" behavior of the filled material (static and dynamic coefficient of friction values which are closer together). The original graphite sleeved displacer gland I used seems to have a stutter to it which is audible with the contact stethescope. I dedicated a new carbide lathe tool to use on only acetal with excellent results. There is virtually no pre-loading of the cutting tool with this material, resulting in very accurate cuts that can hold well defined edges. I made a lot of experimental acetal prototype parts that I later copied in stainless steel purely for aesthetic reasons, but there is no technical reason for not utilizing this material for many of the major engine components. There is a slight "spring-back" to it as a hole drilled the same diameter as a shaft will grab the shaft with a pretty strong grip (although not as pronounced in this regard as Nylon). It works especially well for making the cone shaped shaft spacers that engage the inner bearing races (referred to in the Senft book   ). I found that tapering steel to a fine point at the cone apex can leave a burr that is almost imperceptible but can still damage the bearing. Delrin^® & Teflon^® are registered trademarks of E. I. du Pont de Nemours and Company   James R. Senft, An Introduction To Low Temperature Differential Stirling Engines, Moriya Press, River Falls, WI, 1993

thermometers November 22nd, 2005

Searching for the finest instrument American engineering/China manufacturing has to offer, I naturally headed straight for the local SuperTarget store. I was immediately drawn to the Target Tru-Temp model in the housewares/kitchen department, based on its suitability for mounting directly to my flywheel support post. Further research revealed this most likely to be a Taylor model #9840. When I got this beauty home and dissected it, I found it to have other desirable properties, to wit: 1) These appear to use a Resistive Temperature Detector (RTD) sensor rather than a thermocouple. RTDs (thermistors) offer the advantage of producing a relatively linear electrical output and luckily my installation did not require a lead extension which might throw off the calibration. 2) The sensor bead is not encapsulated in epoxy as in others I have seen (in fact it doesn't even have any adhesive at all) and is easily extracted from the probe tube without damage. 3) The low thermal mass of the tiny sensor bead gives a response time that is very fast. I estimate that they give a stable reading of the plate temperature within 10 seconds of initial surface contact at light pressure. 4) I doubt that any digital thermometer in this price range will have much better than 2 degree absolute accuracy, although the particular pair I have seem to track each other within a fraction of a degree which would bode well for relative differential readings such as this. Rather than being a useless decorative add-on, they proved to be very informative as to what is really going on thermodynamically. The thermometers show that the engine in it's current configuration consistently starts at 6 degrees F. The thermometers also show that under certain conditions my hands cannot generate this six degree differential (iron poor blood?). Before I installed the thermometers, I spent countless hours trying to account for what I thought to be irregular engine starting when the irregularity is elsewhere. I dissected a digital aquarium thermometer which I believe to have a type K Alumel^® and Chromel^® thermocouple sensor element. I suspect this to be typical for indoor/outdoor thermometers with a remote outside probe on a long cord. I had to dig the thermocouple bead out of some kind of silicone encapsulation, almost destroying it in the process. This probe has a much slower response than the RTD device. One tantalizing advantage of thermocouples I have yet to explore would be the ability to connect the two sensors in series differential mode and get the temperature delta directly with one readout...maybe a future enhancement opportunity. I reviewed this treatise on the non-linearity of heat energy over wide temperature variations. While a noble effort to straighten out some cloudy thinking of our forefathers, I concluded that we are stuck with the flawed system that history has handed us, and likely this is of little consequence at the temperature ranges an LTD Stirling operates. I believe others have referred indirectly to this phenomenon when comparing the increased energy released by cooling the cold plate below ambient using ice verses heating the hot plate the same delta above ambient ...anywho, it has been duly noted and we can refer back to it if proven to be consequential. I will defer judgment on it until I can get some Stirling cycle energy readings by integrating pressure vs. volume curves. Alumel^® and Chromel^® are registered trademarks of Hoskins Manufacturing Company

pressure transducer November 23rd, 2005

egad!---the French got the jump on us ...but seriously, analyzing what is really going on here is quite interesting and I will have to replicate this setup in some form or another to get thermodynamic performance metrics out of my engine. A better explanation of a similar setup is here. backing up a bit to get my theory straight: The area enclosed by the pressure-volume curve represents the total thermodynamic energy available to the engine per cycle. In practice, this is usually an ellipse rather than the squared-off appearance of the ideal Stirling cycle (this is an important feature we will get back to when discussing the so called "lost motion" linkage) To get the data points for this P-V graph, we need to measure the instantaneous pressure as a function of volume within the displacer chamber. measuring volume: They are measuring the piston position with an optical gradient filter (overly complicated, but effective) recognizing that there is a one-to-one correspondence between the piston position and the internal volume of the contained air. This gets us the volume parameter that will make up the abscissa of the PV curve and is sent to the horizontal input of the oscilloscope. measuring pressure: They state they are using a Motorola (aka Freescale) MPX2010 piezoelectric low pressure transducer. From comparisons of the photos with the Moto spec sheets and other nosing around on the web, I conclude that this part is available in the Digi-Key catalog as part number MPXM2010D-ND for $7.90. This configuration has a low level 25mV output range (the extra components on their circuit board appear to be an additional amplifier stage to boost the signal prior to sending it to the vertical input of the oscilloscope) and a pressure measurement range of 0-1.45psi. Due to the "plumbing" aspect of the pressure sensing interface, this part comes in a multitude of case configurations. I think I can machine a custom adapter to allow mounting of the bare transducer directly on the top plate of my engine thus eliminating any "dead space" from any connecting tubing. Because I also want to mount the supporting electronics remotely, I opted for the amplified version of this device for a better signal to noise ratio. I ended up ordering Digi-Key part number MPX5010GP-ND which is an amplified version of the same part they used with a cost of $18.19

machining November 22nd, 2005

the trusty prazisionsdrehmaschinen in action

Flitz polish November 23rd, 2005

I found that Flitz polish worked well for taking a small controlled amount of material off of the centerless ground rod I used for the flywheel shaft. This was necessary in that the ball bearings were such a tight fit on the shaft that they could be damaged (found this out the hard way) by forcing the shaft through in it's original diameter. Used this same technique on the power piston and displacer crank pins. I also used Flitz to burnish the displacer rod. Concerned about the purported "protective film" of this product gumming up the graphite sleeve in the displacer gland, I cleaned the polished surface off with a solvent (Methyl Ethyl Ketone). To finish it off, I burnished with Dow Corning Z-Moly powder.

carbon aerogel November 26th, 2005

Determined to use the most exotic and atrociously expensive materials I could get my hands on, I am studying the properties of carbon aerogel and nanofoams for possible use in regenerator cores embedded within the displacer. I have reservations about its suitability for this purpose if the cell structure does not allow air to circulate through it (photos of people in lab coats protecting their hands from blowtorch flames does not instill confidence in this respect). Short of committing big bucks for this, I did order an inexpensive carbon aerogel "super capacitor" from Digi-Key to dissect and report on what it looks like. Whatever the case, this material is too tempting to resist experimenting with. Dropping back a notch into reality, I am eyeing a product called PocoFoam (a steal at only $50.00 for a tiny sample). This appears to have the proper open microcellular pore structure with a high thermal conductivity required of a Stirling regenerator.

turnbuckles November 29th, 2005

Talking with Mr. B at lunch, I mentioned that I needed to incorporate tiny turnbuckles to allow linkage adjustments without any disassembly of the components. Bemoaning the prospect of buying both left and right handed taps and keeping track of which is which (#0-80 threads are so small, I can't tell which way they are going without magnification), he pointed out that I should use right handed threads at both ends with different pitch. This advances at the difference between the two thread pitches with each revolution, allowing much finer adjustment over left/right equally pitched threads which advance at twice the thread pitch per revolution. He says that this is common knowledge and is used to good advantage in many aircraft applications such as servo feedback potentiometer linkages requiring fine grained adjustments. Doh!

Teflon^® December 12th, 2005

I ordered a solid Teflon^® rod from smallparts.com. It may have the lowest coefficient of friction (as low as .04) of all solids, but it has the dimensional stability of a pencil eraser. Although I think it potentially could be used as a sleeve backed by a more rigid material, for now, I am moving on to more exotic hybrid materials with a hard base resin filled with a dispersed lubricant.

Nylon December 16th, 2005

I bought some molybdenum disulfide filled Nylon from smallparts.com. It suffered from the same lack of dimensional stability as solid PTFE so I never used it.

electronics December 18th, 2005

working on schematic diagram for pressure transducer electronics package.

oil May 2nd, 2006

moebius 9010 watch oil

mark-V specification February 12th, 2006

displacer gland: Torlon^® 4301 filled extruded bearing grade Polyamide-imide displacer rod: .0625" type 303 CRES steel centerless ground rod polished with Flitz and burnished with Dow Corning Z-Moly powder regenerator: carbon nanofoam top plate: 1/4" copper with 1/8" pin cooling fins bottom plate: 1/8" copper