5 Bears Home Homebrew CNC bench mill MW54 Gas Turbine Wren PropJet Wright Whirlwind Radial Engine Hodgson 9 Radial Engine Autostart ECU Latest News Useful Links Various downloads Ask and Answer, comments too!

5 Bears: Previous News
The postings on this page are "retired" news tidbits from the 5 Bears Home page. It is growing quickly and takes quite a bit of time to load. Patience please!

15 Aug 2005: Spindle musings, Rhino 3d... Please see the Guest Book page for a pretty sad set of excuses on my web tardiness. With the explosive growth of "blogs" on the internet, people are becoming conditioned to expect fairly frequent updates. Turning 5 Bears into some sort of mechanically oriented blog is the last thing that I want to do. Meaningful content is my goal, and I like to think that I've been successful for the most part. With that said...

The good news - I've made huge strides with Rhino 3d! I've had the software now for a couple of years, and have kept it updated with the correct service packs, but beyond a bit of fiddling with the interface, and some 3d text experiments for engraving, I've been painfully slow in gaining any meaningful skills. I felt a bit like an old dog, incapable of learning any new tricks. The intimidation factor was really ratcheted up by the vast assortment of truly amazing work posted to the web by Rhino Masters. On top of the incredible 3d shapes, these examples are often photorealistically rendered, adding to my anxiety. Think PIXAR, maker of a number of animated movies that I really enjoy, such as Toy Story.

''There's no way I'll ever be able to do that stuff short of a year or two of schooling." Totally false. The reality is quite simple... flowing, 3d, organic forms like Ripley's "Alien" are for movies, not for a garage workshop. 98% of what a machinist will want to do is far removed from such work. The fact that Rhino can be used to model a sea anemone tends to cause potential users to focus on these more cosmic techniques at the expense of "blockier" (but more useful for us) models.

Within two days of serious effort, I was able to create useful 3d models, and I also learned quickly that even for 2D draft work, Rhino was vastly superior to anything I'd used up to that point, which was primarily QuickCAD. The power and simplicity of Rhino can best be shown with this quick and simple demo.

My new skills (still quite basic at this stage) now make me feel a bit better about my selection of Rhino3d and Visual Mill as my software tools of choice for CAD/CAM.

Still MORE spindle thoughts - Those of you who have read the CNC mill saga from the beginning know that I have probably spent more mental coin on the spindle than any other part of the mill. And rightly so, I believe. A powerful, light, accurate spindle truly is the heart of any mill, CNC or manual. I don't believe there is any one spindle for a given mill which will do everything you want. A bench CNC mill spindle and motor combo will fall somewhere into this matrix. Those examples in tan color I have available for my own mill.

General Spindle Type




High Speed HF (High Frequency)
  • KaVo
  • Hofer
  • Precise
  • Extreme precision
  • Near 0 runout
  • Superior surface finish
  • Lightweight
  • Nice variable speed
  • Extreme cost
  • External, expensive power supply
  • Sometimes excessive RPM
  • Torque falls off when slow
  • Slow stock removal rates overall
  • Cutters break easily as diameters drop
  • Very expensive tooling
  • Z usually NOT fixed
High HP
  • CAT30 or largish ER type box or cartridge spindle
  • Adapted BP-style head
  • Heavy duty
  • Max stock removal rates
  • Less expensive
  • Cheaper tooling
  • Fixed Z tooling
  • Heavy weight
  • High power requirements
  • Low top RPM
  • Speed adjustment difficult
Adapted existing spindle
  • Mini-Mill R8 Spindle
  • Sherline or Taig Spindle
  • Low Cost
  • Reasonable accuracy
  • Spare parts available
  • Usually cheap and common tooling
  • Occasionally innacurate
  • Possibly cheap construction (Chinese mini mill)
  • "Jack of all trades; Master of none"

The reason for this matrix - while I really like my KaVo spindle, and find that for certain operations it simply cannot be beaten, the tiny, light cutters simply cannot hack it when the size of the work increases beyond a certain point. I also have a TG75 colleted Gilman box spindle, but I realize now that this spindle is large (and heavy) enough for a 1 ton mill. It is simply too large for a bench mill. And finally, I have a Sherline ER16 spindle which is a real bargain, but it falls squarely into the "a bit too small and light" category. Swapping spindles between the KaVo and the Sherline is not as simple a job as I had anticipated.

New spindle plan - I ordered an R8 mini-mill spindle from The LittleMachineShop.com. Along with the spindle (an incredible bargain at $249) I ordered a huge suite of Tormach fixed-Z R8 spindle tooling, the spindle belt drive conversion kit, and a spindle lock kit. I am going to combine these items, include a new set of precision spindle bearings, and turn the stock mini-mill spindle into a kick-butt, high-speed, precision R8 job with fixed Z tooling. If it flies, it'll do away with the "Jack of all trades, Master of none" syndrome that normally characterizes spindle adaptations of this sort, and I think it will be a "brilliant" (as the Guiness Draft folks would say) match for this 6" X 18" bench mill. About the only issue that I can forsee would be a tendency to be slightly underpowered. The belt-drive kit will help me adapt more HP if needed.

This spindle project will probably merit web space as a mini-project. I expect arrival sometime this week. Stay tuned!

June 2005: The Essential Resources page is up. It will improve with time, as I add additional resources. There are probably 12 or more specific pieces of tooling that I use constantly and which have proven their worth over the years. Along with tooling, I will add books and periodicals, raw materials, and other suppliers. Some of it will be applicable to U.S. residents only, but I hope the majority of the postings will have some value to other countries.

Essential ResourcesAlong with resources, I may throw up a few shop-made fixtures and tools, and perhaps some tips and hints. For example, the CNC mill project and both turbine engines required some extremely precise carbon steel shafts be equipped with precision bearings. This meant that the shafts needed to be turned and polished to +0.0001" -0.0003" or so, normally out of reach unless you have a cylindrical grinder. I've developed out of necessity the ability to do this with a methodical approach that yields excellent results. It includes a custom ground HSS lathe tool, silicon carbide paper, and a good micrometer capable of accurate measurement to 0.0001" or better. I will either describe the technique on the resources page, or link to it there.

Again, it will be built up over a period of probably years, so check it as often as you desire; there may be something new! I should have started this page years ago.


20 May 2005: I hope this modest update finds everyone well. The bulk of "5 Bears" has been around for quite some time, and most of it was written and posted when things weren't quite so hectic on the home front. Everyone knows that the airline industry is in the doldrums, and American is no exception. My own schedule is significantly busier than it has been in years. Most of my flying has been to Central and South America, specifically the cities of Maiquetia in Venezuela, and Guatemala City. For quite some time, the Maiquetia layover included 24 hours on the ground, and this period was one of almost desperate boredom! The city is very anti-American, and it is dangerous to leave the hotel, so we are pretty much stuck there. Reading is about the only way to kill the time, and I always carry a healthy supply of hobby-related books and magazines. That is why Jim Harvey's book has been a lifesaver!

This is NOT the ugly rifle!Over the years, I get many emails wanting to know my background, and how exactly I got started in this rewarding but admittedly obscure hobby. I can trace the roots back to my days in college. I grew up reading Louis L'Amour's Westerns, and have always had a love of older firearms. Flintlocks, especially, fascinated me. In school, I was determined to build one, and given the total lack of tools beyond carving knives, rasps, and sandpaper, I pursued a Pennsylvania rifle design that was moderately prebuilt, in that the barrel channel at least was pre-inletted. The build proceeded, and boy was I proud of the finished rifle! Now, over 20 years later, I cannot physically look at it! The craftsmanship is decent, but the style and the contouring is entirely wrong.

Years passed. I was busy being a father, husband, and pursuing an Air Force career. As my children slowly developed to a state where I was no longer concerned that they would drown, electrocute, or stab themselves if I turned my back for five minutes, my interest in muzzleloaders was rekindled. I began to build again, and with each rifle, my skills improved. Those of you who have Guy Lautard's Third Bedside Reader will find some pictures on page 6, along with my adventures in color casehardening.

The "furniture" for these rifles is available commercialy. At some point, I decided that rather than pay ridiculous prices, I could make my own! This desire launched me into machining! Ignore the logic that says I could have purchased furniture for 20 rifles for the cost of the machinery... that's for your wives to contemplate. I purchased one of the early Smithy 3-1 machines. Not to elaborate too much, I hadn't a clue as to what I was doing, but eventually began to produce nice hardware. I also outgrew the accuracy and capabilities of the Smithy, and upgraded the shop to what it basically is today. More time passed. I found myself enjoying machining metal more than shaping wood. The muzzleloaders were set aside (temporarily, I am sure - this web site has fanned the flame!) and I went on to my other great love, that of aircraft engines.

I'm drifting a bit. More than anything else, with the sole exception of practical experience, books and magazines have helped me gain skills. Even tool catalogues, like MSC's mighty Big Book, helped educate me. Tooling is the great mystery for most machining newcomers, and if you browse enough tooling catalogues, the knowledge ultimately will seep, via osmosis, into your brain.

A terrific place to begin your library has to be any decent used book store. Unlike so many other hobbies and activities, with the exception of CNC, not too much has changed over the years. A text on machine shop practice from the 1930's will be full of relevant material. Even better, it will be loaded with practical techniques for manual machines, no longer in vogue today due to the preponderance of CNC equipment. Personally, I will snap up nearly any machine-related book published this last century.

With this rambling complete, here is what will happen...

I have decided to create a page for 5 Bears which I will simply call essential resources. On it, I will post reviews of books, magazines, tooling, accessories, etc, that I have found invaluable over the years. And Jim's book will be there.

It will be more than a generic links page. Give me a day or two to create a page which will be permanent and useful.

And if you are a casual web surfer, a regular reader, or simply am fascinated with home machining, start now. Don't wait. Get your hands dirty. Find out what it's all about! You won't regret it.

30 Apr 2005: Soo much to catch up on! First, the hacking has stopped. My host service set my email to a "black hole" state, meaning if an email is sent to the 5bears domain, and there is not an appropriate recipient, the email will be jettisoned. This seems to be working, at least I haven't heard of any more attacks.

Model Engine BuilderMike Rehmus, the editor of Model Engine Builder magazine, has posted to the guestbook several times. I was (and still am) a longtime fan of Strictly IC magazine. This new magazine has taken the reins of this niche hobby. I received my first copy a few days ago and still haven't gone through it in detail, but what I have seen is very impressive. The quality is evident; great layout, color, and above all content. If you like engines in miniature, may I strongly suggest you support Mike and his crew with a subscription. There really is no other alternative other than to hope for the occasional IC engine article to pop up within one of the current crop of home engineering publications.

I am trying to come up with some article(s) of interest for Mike. Support of this magazine, via its readership, will ensure success. Unlike a typical R/C enthusiast magazine, where hundreds of thousands, if not millions of people participate, the current crop of interested readers is not large, and such a niche magazine will need support both via subscriptions, and also via contributions.

James on a sweet latheMr. James Harvey, author of Machine Shop Trade Secrets, contacted me via the Guestbook and has kindly offered to send me a copy of his new work. I have heard good things about this book. One of the most common questions that I am asked is this: "What books do you recommend for a beginner?" In the past, I have always replied "Any of Walker's Modern Metalworking series; Technology of machine Tools by Krar, et al, and Machine Shop Work by Shuman." Some of these are hard to find. I'm hoping that I can heartily recommend this book, and I suspect I'll be able to do just that.

In addition to books on technique, one other aspect helped me greatly to learn about machine tools and metalworking... that is, simply browsing through any decent catalog, especially ones that have a wide selection of tooling. You'll be exposed to odd tools, tapers, and nomenclature. Shell end mills? M2 High Speed Steel? C6 carbide? Morse Taper? ER Collets? Kurt Vises? Dividing head and plates, that thing looks insane! And what the heck does 'Center-cutting' mean as it relates to an end mill? You get the idea. There is an osmosis effect that occurs. You'll want to know just what that ISO30 shank cutter does, and how it differs from a collet. "Wow, that 5C spin jig looks like a killer tool, and cheap, too!" Then you find out that it is far more suited to grinding operations than milling.

Beginners have a pretty intuitive grasp of what a milling machine does, for example. It looks like a super drill press, except it can cut on the sides of the cutter! A metal lathe's function is also obvious. What is not obvious is the vast array of tooling and techniqies which will ensure success with these primary tools.

I look forward to reviewing this book!

Finally, I have exchanged a series of emails with Mr. Lee Hodgson of Ageless Engines. Lee offers a series of plans for a variety of radial and rotary engines in miniature, one of which is my flagship project. Lee has described several improvements made to the design since I worked on mine. I'm going to list some of them on my Hodgson-9 project page in the next few days.

For those who have awaited updates with great patience, please accept my apologies, I'll try to do this a bit more frequently.

Now go get your hands greasy!

1 Feb 2005: Distractions; Delays: I know quite a few people are waiting for progress on the Whirlwind... I keep getting distracted by a number of issues. The primary issue is the move of our household, people and goods, to a new home. I currently like in Keller, Texas, which is a suburb of Ft. Worth. The old Chisolm Trail (cattle trail) is located literally a hundred meters from my front door. The dust from the cattle drives is oppressive in the summer, and the rattlesnakes and coyotes are a real nuisance.

Please, I'm not serious! In my visits to Europe, I've had more than one "Old West" afficianado, when hearing "Ft. Worth", assume all is as it was 120 years ago. The Chisolm Trail really does run through our neighborhood, but Keller is a modern suburb, quite green, thank you, and is unlike the mental image of the Old West, which usually pictures the Utah Badlands and West Texas. There are several "Gentleman" ranchers that keep small herds of Texas Longhorn cattle nearby... other than that, and the number of saddle horses maintained, there's not much different here than any other U.S. suburban town. The real reason I am moving is so that we can live in an airpark! I have been flying since I was 14. Now 42, I have not flown for pleasure in over 15 years, and I feel a strong need to regain the joy of flying for myself, when and where I want to. A very happy bonus is the presence of a full hangar on my new property! When I view what appears to be hectares of floor space, my mind soars to the possibilities of the ultimate shop. Everything done on this web site was completed in a floor space roughly 10 feet X 12 feet. Packed in this space are 3 machine tools and almost all of the stash one accumulates after years in this hobby. My wife is unhappy when she has to move my THK rail collection so she can get at her stuff in the garage

Another distraction - I keep playing with the CNC mill and not really getting anything done. After perfecting the new tool length sensing procedure in Flashcut V2.1.2 software, I took a number of test cuts to validate that yes indeed, the new tool is set to program zero. At 25,000 RPM, a 1/4" carbide end mill really rips through aluminum, leaving a beautiful finish behind. It also showers everything in the vicinity with an unbelievable spray of very fine aluminum swarf. Similar to party glitter, but finer. Almost a powder. This stuff goes everywhere! Without proper shielding, my linear motion hardware would soon be rendered polluted and useless.

I tried eBay for a while, but the pickings were slim. Keywords like "bellows, swarf, way, guard" etc didn't help. As usual, MSC came through. Pictured here is a 24" / 600mm square Gortite bellows of special synthetic rubber, immune to oils. The interior dimension is 6" / 152mm square; the outside is 8" / 203mm. Be warned, these bellows are expensive. MSC also sells bellows which are made for covering ways like those on my mill, at 80% of the cost of this square bellows, but the square bellows, when cut, will actually give me two bellows of 24" each, a better deal than buying the standard bellows. It will also allow me to customize the shape a bit more.

Hard to believe it, but this compact bellows, which naturally compresses here to about 20mm high, less than an inch, will open to 600mm long! I really like this product so far. It is a tough, high-quality material, with fabric woven inside, quite unlike the cheap, all rubber way covers one sees on "mini-mills" and similar. My plan is to install this bellows, after cutting, on both the front and the rear of the Y-axis, and on the lower portion of the Z axis. I should have plenty of material left over for the upper Z if I so desire.With proper swarf guarding finally in place, I will have no fear of really going to town with this mill. If you are working on your own high-speed bench miller, you must plan for swarf guarding from the beginning. It is no fun, and difficult, to add this feature after the machine is finished.

One final CNC tidbit - just for fun, I replaced the more than adequate NEMA 23 motors on the X and Y with a relatively large brushed NEMA 34... the same motor which powers the Z. Those who've read through the CNC construction remember this page - Adapting a NEMA 34 for the Z - where I destroyed the encoder which came with this set of motors, and laborously modified and installed a RENCO 2000 line encoder. I bought three of these motors in a package deal from eBay. They came with a mystery 500 line encoder, and each also had a very high quality Thomson Nema-True planetary 10:1 gearbox. I remember paying next to nothing for this set of three, which would probably run >$2,000 new. Prices for motion control hardware on eBay have been up quite a bit recently due to market pressures. If you are stocking up for your own project, don't wait, get right in and buy. I digress, back to the motors.

Two motors remained with the mystery encoder still installed. Not wanting to bore out any more encoder disks, I unleashed a DVM meter, applied +5V to the encoder, and began to examine the outputs. Fairly quickly, I identified the Index, and Index~. Identifying the A, B, and complements took a while longer, but ultimately I wired it up and was rewarded with a correctly functioning encoder! The destruction of the encoder from the Z-axis motor was a total waste! Also, 500 lines is better than 2000 lines.

Between work, the "move", and general shop fiddling, I've made no progress yet on the Whirlwind. Please be patient, I'll make some progress soon. In the meantime, I'll try to update the site with stuff that is hopefully of interest.

Finally, sadly, the Guest Book has been discovered by advertising bots. All I can do is shake my head. The internet is becoming useless due to ads, spyware, and evil bots. Skynet is born; where are Sarah and John Connor when you need them?

7 Jan 2005: Happy New Year! Please say a prayer and contribute to the Tsunami relief fund in your country. What a horrible way to end 2004, not a particularly happy year for so many people. Keep the big picture in mind; so many of us are blessed beyond description, let's not lose sight of that.

On a happier note - the holiday season is behind us all, and I hope to begin to make some visible progress on the Whirlwind. No cutting yet. I have been thoroughly enjoying a recent upgrade that Flashcut has made to their 2.1 series of CNC control software... they have incorporated tool length sensing to their system! Before the automated tool length sensing, tool changes without fixed toling have been a real pain, requiring one to reset the program 0 using either a fixture or touching off to the work. The latter doesn't work if you have milled down the original Program zero! With tool length sensing, when a tool is changed, and you select the tool from the tooling entered into Flashcut, you can activate the process with a button click. The spindle will drive to a predetermined location, and slowly descend until the switch is tripped. It measures the difference between the current machine Z, and the spindle nose (determined earlier), and then applies a G43 Z offset value to the tool. All this means is that when swapping tools of different length, the system takes that length into account and no correction to program zero is necessary.

Sensing requires an input line equipped with some type of switch. Since all 8 inputs are currently in use, I "borrowed" the Z-axis servo fault line for some preliminary tests. Here, the Z-fault is wired to a simple microswitch. Preliminary results show a repeatability of no worse the 0.0004".

One other extremely slick device is this pendant based upon an X-keys product, the X-keys "professional". Flashcut has extensive keyboard shortcuts, but I have never been one to use them, relying on the mouse. For everyday processes like jogging, resetting 0, etc, this can be tedious. Enter X-Keys. X-keys makes a series of USB keypads which are entirely programmable for any number of keystrokes by the user. Programming a key's functionality is nothing more than specifying the key, via their excellent macro-works software, and entering the keystrokes which you want the key to emulate. They also include a MS Word template for the stickers which you see here on a portion of the keypad. The stickers reside below plastic covers, so you need not fear getting them dirty in the shop.

The X-keys pad has greatly increased the ease with which Flashcut can be used. This is a great product, and I recommend it highly. Oh yes, X-keys can be used with any windows software whatsoever, so don't hesitate to try it with other Window's-based controls.

28 NOV 2004: The Cirrus rebuild is complete. Unfortunately, I lost some of the photos, so it will end rather abruptly... my apologies. I decided to publish what I did have so as to freshen the site a bit.

Since the completion of the CNC mill, I have been looking for a new, major project, one which will hopefully incorporate some CNC production. Some of you may be inhabitants, like me, of the CNC ZONE. This is a message board devoted to CNC, both major commercial machines and practices, and homemade equipment as well. In one of their forums, the I.C. Engine forum, I learned of a set of engine castings available which I had thought were "long gone." The engine is the Wright Whirlwind, first published in Strictly IC magazine years ago, beginning with issue #24, Dec 1991. This miniature was designed and executed by The Wright WhirlwindKarl-Erik Olsryd of Sweden, to power a scale replica of Charles Lindberg's Spirit of St. Louis. I remember at the time thinking "What a neat engine to build sometime in the future." As the years went by, the castings were no longer being offered in the classified advertisements, and I assumed that they were no longer being made. I was pleasantly mistaken.

The Whirlwind castings are still available, complete with what appears to be an excellent set of CAD prints! All appears to be of top quality. Further pictures of the castings can be seen on Einar Sjaavik's web site.

The Whirlwind is a 4-stroke, spark ignition, 9 cylinder radial aircraft engine, 25mm bore, 28mm stroke, and a displacement of 124cc. I am not even going to attempt to estimate a completion date. Suffice it to say, it will probably be a year or more. I plan on serializing it in a hopefully high-quality format. The Hodgson radial engine, which has been completed for some time, was perhaps halfway completed when I began to post to my original website. The serialization was limited and crude, and many of the pages were simply photos of already-completed parts and subassemblies. The feedback on the Hodgson engine was terrific, and as my subsequent projects progressed, I made an effort to post more pages reflecting the actual machining processes, culminating with the CNC mill, my last major effort. With the Whirlwind, I will do my best to show as much as possible, in a manner similar to the CNC mill and the turboprop. One exciting aspect - with the CNC mill now operational, replication of many of the components should be a snap, and should make for some interesting postings.

I don't anticipate that I will be modifying this engine to any degree. I will probably make use of Imperial fasteners, but dimensions, gears, ball races, etc, will remain Metric. Thankfully, I am becoming more comfortable with that system. I really appreciate the effort that Mr. Olsryd made to create an engine as close to scale in appearance as possible. In my opinion, the Whirlwind is one of the most attractive radials ever created. One other exciting benefit of this design is that the head and rocker castings are not too expensive, and any number of aircraft engines can be engineered around the head and cylinder. See the Whirlwind web site for examples of air-cooled 1, 2, and 4 cylinder variants.

31 OCT 2004: After months of waffling, I have decided upon a new project. It will incorporate CNC to the maximum extent possible. I have come to the conclusion that the only way I can learn CAD is to be motivated enough to actually produce meaningful parts, and that requires one more major project. Simply creating small desktop dew-dads and saying "yup, it works", no longer cuts it, pardon the pun. On to more smoke and noise!

You'll have to wait a while before I begin serializing, or even leaking the project. I need to gather the requisite castings, and don't want to create a run on the individual who supplies them. I will say this much... it will be round.

In the meantime, I have begun a very short series of pages which detail the teardown and cleaning, and general restoration of my first IC engine, the 1/6th scale deHavilland Cirrus inline four aero engine in glow. I anticipate three pages with eight to nine photos per page. My Cirrus had been shelf-bound for quite some time and had gathered quite a bit of dust and grime. While not construction, it'll hopefully tide readers over until "the big one" commences. The first installment has been posted. Within the pages, I will attempt somewhat to relate some of the construction techniques used. The engine runs superbly on hobby glow fuel. I am tempted to convert it to spark, but I will do this at some future date.

This engine was serialized in Strictly IC magazine some years ago. The engine required a number of castings which are sadly no longer available. I will take this moment to advocate the purchase of engine kits or castings which interest you when they become available. Over the years, I have sadly missed out on several fine designs, and short of creating your own patterns and casting your own components, once you miss the castings, the engine is pretty much gone!

1 Mar 04: Things are not as quiet as they seem at 5 Bears! I seem to go through cyclical periods where I post to this site quite frequently, and others where I simply let the site "age" like a red wine (or a bottle of Scotch, pick your poison). Almost all shop effort has been towards an overhaul of the spindle backplate and spindle mounting systems. I should have paid more attention to Z-axis mass and overhang. The CNC mill, as it is currently engineered, forces me to create a spacer system which moves the spindle centerline quite a distance away from the spindle backplate. The backplate is a fairly hefty, 1.5" thick cast and machined aluminum plate which is mounted onto the NSK Z-axis rails and trucks. It was designed to allow any number of spindles and mounts to be attached with ease. I machined a clunky and ugly spacer block for both the Kavo and the Sherline spindle systems, but along with it's unaesthetic lines, it was flawed in that swapping spindles was a horrible exercise in alignment and DTI use. I decided, then, to create a spindle foreplate (see photo), which will be added as an installment to the CNC mill saga sometime soon. This plate is dimensionally similar to the backplate, and is sandwiched with the backplate using six 5/16" X 18 cap screws through two beefy standoffs.

This hollow-sandwich method will allow me to sneak in a hex-wrench between the layers and remove the spindle for swapping or maintenance. All of this effort is part of the general spindle upgrade plan I have in mind. Earlier, I mentioned the machining of a pair of sheaves for the Sherline ER-16 spindle, and the eventual supercharging of that device with a 700+ watt 20,000 RPM DC motor... along with the sheaves, I have engineered and machined very nice motor mounting plate (not shown) which adjusts laterally (parallel to the mill's X axis) on a pair of THK RSR12 linear trucks for belt change and maintenance. The motor mount will be precise and tight, important for proper belt tensioning at 20,000 RPM.

The Sherline factory assured me that with the correct preloading, the Sherline spindle bearings can handle 10,000 RPM. I am planning on creating a tachometer system using components from the turbine ECU I engineered a while ago. A pair of neodymium magnets embedded in the bottom of the spindle sheave will trigger a hall-effect tach probe. A PIC16F876 uProcessor will sample the pulses generated and output the spindle RPM to a small LCD display. I'd like to create a cool spindle electronics interface somewhere above the spindle mounting area, with I/O for CNC control of the spindle, manual control, RPM output, and possibly an RPM feedback system to maintain exact RPM's. The current plan is to limit the spindle to 8,000 RPM for a bit of safety overhead.

One of my favorite tools, an automatic boring and facing head for an R8-shanked mill.

Long ago (it seems that way at least, say 1996) I was deep into the creation of my deHavilland Cirrus 1/6 scale engine. The engine has 4 cylinders, but makes use of a single hunk of aluminum for the heads, with some artful fin cutting creating the illusion of four separate heads. Each cylinder required a corresponding flat-bottomed bore of significant depth in the bottom of the head, and the only way to do this with manual equipment is either on a lathe faceplate (a horribly taxing setup to do correctly), or with this tool. This is a small ENCO automatic boring head which costed roughy $200 in 1996. It is Asian made, Chinese I believe, but the quality is very nice. Chinese tooling tends to be either very good, with a super price, or garbage. In this case, Enco had a local outlet and I was able to examine the tool before I bought it. I knew quickly after a few moments of handling that it was of fine quality.

An automatic boring/facing head can function either as a manual boring tool, such as a Criterion or APT type, or as an automatic head. When the head is set up for automatic use, the tool bit feeds outwards (to a larger diameter) a fixed amount with each revolution of the head. Normal feed rate is 0.002" per rev. To begin an automatic pass, the cutting bit is manually run to a small diameter position, the quill depth is set either static or under power, and when ready to cut, the large knurled ring is simply grabbed by hand and held firmly! The head has an internal pawl which creates the feed, each outfeed accompanied by a very audible click! As the cutting bit describes an ever-increasing diameter, an adjustable stop device causes the pawl to disconnect, and the tool will stop feeding at the same diameter each time.

In this photo, I an creating a shallow, flat-bottomed bore into the motor-mount plate. Into this bore will go a small, indexed aluminum handwheel for adjustment of the belt tension.

The reason I describe this process, along with the photo, is this - while fairly specialized, and not cheap, this is a really cool tool! A lot of fun to use, all sorts of fiddly adjustments, noise, all stuff designed to endear it to a machinist's heart. If you can find one on eBay, snap it up.

16 Feb 04 - I recieved an recent eMail from Aaron Astle, who is launching the sale of what appears to be a very functional small spot-welder. I haven't personally tested the unit, being happy with the one I do have, but it looks like it would be ideal for construction of a lot of the fiddly items needed for small gas turbines. Hopefully, those of you who have arrived at 5 Bears looking for spot-welding information will find Aaron's site helpful in acquiring the parts, plans, or even a complete welder at a very attractive price.

AeroBug - Spot welders at hobby (not industrial) prices.

CNC mill stuff - I can't seem to give up the hardware side of this project and focus more on actually "making" parts with the mill and the software. I think this is a common bug among hobbyists, and I used to think it was a flaw, but have grown to understand that toolmaking itself is legitimate, and the items produced have every bit as much relevance as, say, an IC engine head machined in 3-axes from a Rhino model.

With the Kavo spindle running well, I turned my attention to the little Sherline spindle. First, sneak a peek at this incredible device - an automatic toolchanger for the Sherline spindle! <drool>. Be sure you view the movie. OK, I need one. And in fact I am going to order the plans and perhaps some of the parts. But before I embark on that project, which will be serialized and will make use of the CNC mill, I wanted to improve upon the stock Sherline spindle and motor. My first concern was with the horsepower and rotational speed of the original DC motor. I happen to have on hand a compact and very high-quality Dayton universal (AC/DC) motor, which peaks at nearly 20,000 RPM. It has an internal fan and quality ball bearings front and rear. It will replace the Sherline motor, and be speed adjustable with this "Dart" motor control from eBay.

Left- The motor on the left is the Sherline; in the middle is the spindle, and the bigger "Dayton" motor is to the right.

The Dayton motor's output shaft is larger than the Sherline's, and this, combined with the need for a different pulley ratio, forced me to create two new pullies, one for the motor and one for the spindle. The stock Sherline drive belt is a bit light as well. Having a second identical belt on hand, I decided that doubling the belts would be suitable, i.e. running two belts simultaneously. Since the Dart controller allows me to "Dial-An-RPM", I no longer needed nor wanted the "high/low" arrangement of the stock Sherline pulley setup. With the motor peaking at about 20,000 RPM (measured with my turbine ECU and a magnet mounted on the shaft), and desiring a spindle speed of ~ 10,000, I have turned two pullies to retrofit the spindle and motor, with the spindle sheave being ~ 2.4" / 60mm, and the motor sheave being 1/2 of that.

Right- The spindle sheave is being faced in one of the final operations after the bulk of the turning was completed. A precision form tool was used to create the 30 degree belt channels, which were copied directly from the Sherline pullies. I have taken enough photos to make another installment to the CNC mill series, and will document the process from start to finish. Besides the sheaves, the preloading of the spindle will have to be reduced a bit, and this is easily done by loosening the retaining nut on the top of the spindle. I feel confident that this entire spindle "turbocharging" will bear fruit, but I have some nagging doubts about the longevity of the spindle at 10,000 RPM and higher power levels. The good news is that the spindles are relatively inexpensive, and if I get a reasonably long life, I will be happy.

I was tempted to go further and replace the Sherling bearings with high-grade angular contact bearings, but will save that for another project, and only if the stock bearings fail or are not rigid enough.

Finally, I am sorry for the lack of updates recently, work has kept me very busy. Oh yes, remember the laser? I was toying with the idea of setting up a 25 watt CO2 laser for cutting/engraving duties, but after a bit of research, I must reject it for now as being too expensive, unless I stumble across something really cheap on eBay.

Make chips!


News Flash! There is a gentleman on eBay selling Kavo 4041 spindles. A fellow CNC enthusiast from Norway, Einar Sjaavik, dropped me a note mentioning that he had purchased one too. Sure enough, a search for "Kavo" on eBay located the spindles.

He apparently has several of these spindles if he is offering a bulk discount. At $525 U.S. the price cannot be beaten. The only drawback that I can see is the lack of a "cheap" HF (High Frequency) controller. From the data sheet, the motor uses 3-phase AC power, and by changing the frequency, the spindle's speed is controlled. There is nothing exotic about the matching Kavo HF controller... I suspect any decent HF supply will power the spindle just fine. What you need is 3-phase, variable frequency (80 to 800 Hz) at 48V. The motor draws 13 amps.

The thermistor leads could probably be safely ignored. The obvious answer is to monitor the temperature the old fashioned way (touch it!), or you could always mount a thermocouple to the spindle. I am not sure of the spindle's internal thermistor arrangement.

This is the same guy from whom I purchased my own many months ago. I talked to him via email, and he says that these spindles are in excellent, useable condition but not rebuilt with new bearings, hence the much lower price. The current price is too attractive to pass up, so I snapped up a spare. If you are in the market for an exceptional spindle, look for his auctions in the future. Kavo Spindle on eBay If you have browsed this site at all in the past few months, you know what I think about this piece of hardware. Unmeasurable spindle runout, and at 50,000 RPM, you can barely feel any vibration at all.

27 Jan 04 - I get quite a few emails about the projects posted here. The vast majority are very complimentary, and I appreciate them. What few guys don't seem to be aware of is the large amount of scrap I create. Not everything works out, and the stuff that does, gets posted! Here is an example of a significant effort which fell short.

Available collets for the KAVO spindle for the CNC mill are mostly metric, and quite expensive at over $100 each. The KAVO spindle can handle tooling up to 1/4" shank. Right now, I have only 2 collets, and unfortunately both are 1/8". In one sense, this is fine, as 1/8" tooling is very plentiful, and cheap too. But for some roughing operations, I'd really like a couple more collets in a larger diameter. Not wanting to pony up $100 for a collet, I decided to try making my own from carbon steel.

I dutifully set up the camera, planning on adding a web page to the CNC saga. This was a challenging job, with some very precise boring, an OD measured in tenths, taper turning, plenty of good material for a web installment. The collet on the left is my effort, needing only the slits... the collet to the right is a factory KAVO in 1/8". After several hours of work (I'm slow, what can I say?), the collet had been completed to the level shown.

The collet boring was accomplished with a boring tool, followed by a cobalt spiral reamer. The fit of the collet to a test tool (carbide-shanked) was quite nice, what I call a "hydraulic" fit. The rear of the collet was tapped 6mm for the internal, pneumatic drawbar of the KAVO spindle. After a bit of fiddling, the collet, sans slots, would mount into the spindle bore. I wanted to check the runout of the collet bore before going to the trouble of setting it up for the slots.

First, I checked the Kavo spindle internal taper without a collet. I expected it to be good, but my very best Brown and Sharpe 0.0001" indicator was simply immobile. Absolutely zero visible runout. Very impressive accuracy in this spindle. Germany makes some very fine tooling.

But with the collet mounted, the bore of the collet had a full .002" runout. I haven't a clue as to why this happened. All of my lathe setups looked correct. I would have been happy with .0002" or even a bit more, but 0.002" was just nuts, and rendered this collet worthless.

It sure looks nice, though. Maybe I'll turn it into a lapel pin or something.

12 Jan 04 - 5 Bears has cracked 100,000 hits. It has been quite a ride, lots of fun, since I started this site a few years ago with no other intention but to network and brainstorm with like-minded individuals. With the exception of one long dry spell last year, I feel like I have added some decent material at regular intervals, and that has generally coincided with my real-life projects.

Statistics for the site are always interesting. The #1 method of getting new readers to the site is google and other similar search engines, and very surprising to me, the most popular search string which has routed readers to 5 Bears has been "spot welder". Many, many guys have made the welder, and ultimately all of them have worked, though not without some issues, the #1 issue being too small of a transformer. So if you are pursuing your own spot welder, be sure you have a jumbo transformer. Try Surplus Center, they might be able to help.

The next most popular search engine string has been "radial engine" and with the cnc mill adventures posted, the latter will probably overtake the radial for sheer volume of traffic. Somewhat surprising is the fact that the turbines are not as popular. I think there is still a perception that model turbines are exotic, expensive, and unattainable, so let me say without hesitation that the Wren engines were the easiest projects posted, by far. Even the "scratch built" wren 54 (I used the factory compressor, NGV, and turbine) was a snap compared to the radial. The radial took over 2 years vs. 3 months for the Wren, so if you itch for turbine power, don't hesitate. I would put the difficulty factor of the MW54 at roughly the same level as a simple twin-cylinder IC piston engine.

CNC mill - I have been spending quite a bit of PC time at The CNC Zone, a great forum for CNC enthusiasts. Come on by and look for me online, as Swede, and contribute to the nice library of knowledge there.

I have been a bit busy with work and other non-hobby activities, and haven't done much since the holidays. I did engineer some solid-state limit switches for the little mill. I have a vague notion of creating an "I/O" box which will extend the functionality of the Flashcut CNC hardware, by using an intelligent PIC microprocessor, specifically the 18F252, which is a pin-for-pin replacement to the proven 16F876, driving my turbine ECU. The I/O box will accept cabling from the Flashcut system from both the input and output ports, and by hooking the TTL, I hope to implement a few neat external devices, KAVO spindle speed control, and the previously mentioned solid-state limit switches. It's still pretty conceptual for now. Programming skills quickly go stale, and I want to retain what I have learned of programming the PIC from the ECU project.

7 Nov 03: Goodbye Netfronts. I was being nickle-and-dimed to death there, with Netfronts begrudging every kilobit of transfer. My new hosting service, Lunarpages, seems to be a much better deal. Plus, their customer service, so far, is excellent.

While the site was in limbo, I checked out in the B737-800 for American Airlines. Longtime visitors remember when I was flying the B777... since then, due to layoffs, I have been steadily downgraded, first to the 767, and now the 737. The B737 is a fine jet, but it ain't a 777. Oh well. Beats pounding the pavement.

I have made some progress on the CNC mill, though. It is all on camera, and I've got quite a bit of work to do to bring the web site up to speed. The Z-axis is complete, except for the spindle mount. The column has been attached to the base. Both X and Y axes are completely done, and functioning. The first 2-axis G-code has been run, with an ink pen on paper! Everything looking tight.

Thanks for the patience everyone. More good stuff to come soon.

Two new CNC milll installments posted. The rails for the Z-axis have been mounted, and the ballnut carrier for the Z-axis is complete.

I'm trying hard to catch up. I have literally hundreds of photos to wade through, pick out, edit, and turn into logical, sequential installments.

But I will say that the mill has cut aluminum, with complete success, but a lot of work yet to do to refine the mechanics of the system. I will need to install a gas spring for the Z-axis, as the weight of the spindle assembly is higher than I'd like, and not only does it freewheel downwards with the power off, it is giving the Z-axis motor a bit of grief. I have a NEMA 34 motor which I will adapt, and it will easily have the needed torque to handle just about anything I could mount.

The entire structure is gratifyingly rigid. The Bosch extrusions are more than adequate to make an extremely solid mill, and the addition of the necessary sideplates on the column has added both mass and stiffness.

28 Aug 2003 - A peek at the column structure: Any concerns I had about the rigidity of the Z-axis column vanished when I added the two fully 1" thick side plates to the Bosch 9090H Extrusion. The extrusion alone probably would have performed fine, but to add two THK rails to the column required more space than the 90mm wide extrusion gave me. A bit of CAD work showed that the 1" thick side members would act as a base for the rails. Note how the side members extend forward of flush with the extrusion - The rails will be mounted on these forward edges, giving roughly 1.5" of space between the extrusion itself and the back of the spindle plate. This space is needed to fit the ballscrew and bearing housing.

The side plates will be firmly attached to the extrusion with 8 ea. 1/4" x 20 SHCS per side, with heavy T-nuts. The plates measure ~700 mm or 27 inches. I have no doubt about the integrity of this column, especially when the 25mm THK rails are added.

31 Jul 03: I feel that I am catching up a bit with the project, with the posts coming faster. It was simply a matter of dragging out the HTML editor and getting to work. If you've missed the last few home-page "News" postings, you can find them here or click the Old News button on any page. Since I have basically torn down the CNC mill to serialize it, what I decided to do was machine some of the more boring, assorted attachments which will be needed in the final machine, such as limit switches, swarf guards, etc. Then, as the mill goes together for pictures, it will hopefully be its final assembly.
Milling what I thought would become the Y-axis limit-switch bracket... instead, it became the Y-axis master rail guide.

X axis (inverted), saddle plate, ballscrew and ballnut.



27 Jul 03: Bear with me while I prepare for more details on the little CNC mill. It was roughly 65% complete when I decided to "serialize" it like the other projects presented at this site. To properly photograph the project and to create a logical order to the pages, I am tearing it down to loose parts, and will then begin to build it up again, with photos along the way. In the meantime, I have a couple of pics of one of the more vexing issues facing anyone building a mill... what the heck do you do for a spindle? For a mill this size, the choices are limited; either spend big coin on a high-speed, precision HF spindle, or refit an existing bench mill spindle. In my case, I decided to create a modular Z-axis which will allow the fitting of any modestly-sized spindle, and enable different spindles to be tested or substituted.

The more "advanced" route would be an HF (High Frequency) spindle like the KAVO 4041 shown middle. These normally go new for $3K to $5K, with the required 3-phase speed controller being like-priced. What you get for your money is an extremely precise, high-speed, quality spindle with limited cutter capacity, expensive collets, and phenomenally smooth and accurate running up to 50,000 RPM. The KAVO spindle seen here was another ebay buy at a fraction of the cost of new. It came with this bad boy...

A BIG rack mounted industrial-grade HF controller! The first time I ran the spindle, I was amazed at the smooooth running. Right now, the KAVO is my #1 choice for 3d profiling, engraving, etc. Spindle capacity (pneumatic collet action, very cool) is 0.250" shank or less.

The Harbor Freight mini-mill spindle is a robust, respectable unit with a big R8 capacity, but it must be considered low-speed, and the runout of the R8 taper socket itself is a rather poor .001", which will translate into iffy accuracy. I have considered the possibility of rebuilding this spindle with precision bearings, but have rejected this notion.

A trio of spindles for the bench mill. Left to right, a Harbor Freight mini-mill R8 spindle without the motor attached, a KAVO 4041 high-speed HF spindle, and the old standby, a motorized Sherline spindle with an ER-16 collet nose.

The KAVO nose compared with the Sherline ER-16 nose. The Sherline spindle will probably work fine for light work at lower speeds. The advantage it has over the KAVO 4041 is the larger ER-16 collet, which can take tooling up to 3/8" shank. A big problem with both of these spindles is this - how the heck do I do a drilling cycle? I'm thinking I will need to find either a small albrecht chuck, or rely on the collapse range of the ER-16 collets to grip drill shanks directly.

13 Jul 03: Let me fill everyone in on where I am at with this CNC project. Several months ago, I bought a Harbor Freight "Mini-Mill" with every intention of converting it to stepper control, and gaining some experience with CNC. This topic has always fascinated me, and I really got the itch when I saw guys milling turbine wheels, IC engine heads, etc, out of both aluminum and machinable wax. Wax can be used as a pattern for the lost-wax casting process, another interest of mine.

Anyway, the little mill was marginal in every way. I know it has been done (CNC conversion), but the thought of having a substandard piece of hardware with which I would always be struggling was not appealing. I began web browsing at a furious rate. Off the shelf bench mills of high quality would set me back 15 to 25 grand - yikes! What makes them so freakishly expensive? Hmmmm, ground ball screws? HF spindles? What is all this stuff? Yup, it was that 'stuff' that made the quality mills so expensive... the precision motion-control hardware.

After a long period of study, I decided to pursue the "roll yer own" method of CNC bench milling. The Harbor freight mill was put aside. eBay became my 2nd home as I browsed for, and bid upon, surplus motion control hardware.

While I mentally prepare for the photo onslaught, and how it will be organized, bear with me. I am going to have a projects page for this job similar to the turboprop and radial engine pages.

The dark, greasy, but oh-so-precision underworld of the X axis table, showing the THK SHS-15 linear ways, and an eBay "kill"... a C1 grade ground 12mm ballscrew. Everything in this photo is new eBay fodder at 10% cost.

29 June 03: Never before have I let my pages stagnate as badly as they have. After some mild chastisement from some of the site regulars, I have decided to continue adding material which is hopefully meaningful and enjoyable to all.

First, the turbines - the turboprop is complete, has been run many times with almost flawless results, and like any shop project, there finally came a time when I simply couldn't do anything else to really improve on it. The fall and winter of 02 and 03 were months of shop waffling, doing minor projects and jobs which didn't merit a lot of web space. Around March, I finally decided to press on with something new and major... I wanted to either buy or retrofit a benchtop mill for CNC.

Three months later, in my typical fashion, I have overresearched, overbuilt, and generally really gotten into this project! I have a LOT of catching up to do on the web site, so hopefully over the next few months, I can bring this site up to speed with some cool photos and other enjoyable goodies. Please check back more often.

20 Sep 02: Gearbox Oils I am trying a new oil mix in search of a DANCO replacement. In previous tests, I had found that traditional viscosity additives such as STP or Lucas had reacted with the Exxon 2380 turbine oil to form a gelatinous "gooze" which had really fouled the plumbing during final assembly. Research on the internet (with high speed spiral gear trains the topic) revealed that a flood oil system, as designed by Mike Murphy, is in fact optimum.

The next question involves viscosity, and the ability of the oil to withstand extreme pressures (EP), shear, and its ability to cling to the gears. Excess viscosity will cause the oil to heat rapidly, rob power from the system, and perhaps inadequately lube the bearings. Too little viscosity (such as pure turbine oil) will not properly protect the gears. Castrol Syntec Gear Oil is a fully synthetic gearbox oil which should, when mixed with turbine oil, perform nicely. Unlike thick, viscous, traditional slow-speed gearbox oil, it is relatively thin, and mixes perfectly, without reaction, with Exxon 2380 turbine oil. A 50/50 mix is being tested.

"Gooze!" - After the first few runs, the Turboprop was torn down partially to install one of the new Flight Works pumps. From the beginning, I was surprised at how clean the oil appeared in the sight glass of the oil tank. Here's why - the magnetic "trap" I had installed had done an admirable job collecting ferrous swarf so fine that the result was a black gelatinous mass clinging fiercely to the rare earth buttons. This method has merit! It has kept the oil relatively clear with no other filtration of the oil system.

15 Sep 02: The small Flight Works pump (model 100) has been tested with excellent results. See the details here.

9 Sep 2002: New fuel pumps! The beautiful little pump on the right is a new pump by Flight Works, Inc, based right here in the U.S.A. in California. Initial inspection shows top quality in a very small package, perfect for MW54 and other small turbines. The left hand pump is a Hausl... note the size difference.

I plan on mounting one of them into my turboprop within a couple of days. In the mean time, here are some more photos.

2 Aug 02: As of 1600 this date, the TurboProp is complete! I am going to run it tonight. Tomorrow I am off on a business trip so you'll have to check back later and see how it went. I am roughly 4 "topics" behind in the serialized construction - I have added pages detailing some plumbing work, which was pretty intense given my goals for the completed engine. When I return, I will continue posting pages as well as report on the run.

21 Jul 02: Work proceeds, mainly plumbing and mounting. To do the job I envision requires quite a bit of imaginative mounting of the engine itself and all of the accessories. Most of the last two weeks has been spent on the gearbox lube system. This includes the oil tank and the oil pump pack. The latter is a modular unit which consists of a pump, pressure switch, and batteries mounted in a removable bloc for maintenance.

22 Apr 02 - The Gas Gennie has been test run.

Taking a break from the turboprop, I decided to finally test the gas gen turbine which will be the powerplant for the turboprop. It has been completed for a couple of weeks, but I have been too busy machining to do any testing! Not being able to ignore a new turbine with 0 running time, I decided to fire it up today.

I loaded my ECU turbine data set 3 with data appropriate for the Gas Gen. Really the only difference from my thrust MW54 was a suspected lower EGT; hence, I dropped the EGT MIN RUN to 400. At this time, I haven't added the electric starter, so I decided to use an air start, which is always smart for a new turbine anyhow. With the ECU announcing READY, I applied the air. At 3,000 RPM, the gas and glow came on. Ignition was instantaneous, much faster than my thrust MW54. With the air reapplied, the ECU took care of everything else and the turbine was at idle with no fuss about 10 seconds later.

The big question? VIBRATION! There was NONE! I was delighted with the results of my meticulous balancing. I ran the turbine up to 140,000 RPM, checking vibration levels every 10K or so. Absolutely smooth! EGT's were higher than I expected, with levels only slightly lower than my thrust variant. The cone obviously helps boost EGT and will not be used for the turboprop.

In a nutshell, it was a non-event. Instant start, right "out of the box" after maybe 5 hours of construction and 3-5 hours of balancing. The Wren Mark2 comes highly recommended.

25 Mar 2002: The ring magnets were a bust. Every dimension except the length of the ring was perfect, but every attempt to shorten the ring shattered the brittle material. I even potted the entire ring inside an aluminum tube, but even with the support of the epoxy, attempts to part or saw ring sections failed. Back to square one.

More upbeat, my kit gas gennie MW54 arrived! A new "serialized project" will be started, the Wren prop jet. Rather than reinvent the wheel by machining another MW54, I went the MII kit route, and will spend most of my time on the prop section. I will devote one page to construction of the MII kit, and will give my impressions (very good so far) of the quality of the new kit. Some clever improvements were already visible compared to my set of plans of almost 18 months ago.


14 March 2002: - With either Hall Effect or magnetoresistive sensing being the technique of choice for tachometry, I wanted to simplify the process of embedding strong magnets into a stock Wren compressor nut, or one machined to replicate it so that the Wren bendix would correctly couple. After much searching, I located a source of very powerful Neodymium ring magnets which will be perfect. To help me produce these magnetic compressor nuts in any quantity, I created a deluxe, precision arbor. It was good to do some machining!

I have on order with Wren Turbines all of the necessary parts to construct the turboprop. Rather than machine another MW54 (gas gen) from scratch, I went with the kit to save a bit of time. I'm sure that there will be plenty to challenge me in the gearbox section. I will serialize this entire process as another major project here on this site.

For you piston-heads, I have added another simple yet effective spark ignition, the TIM-7, to the Spark Ignitions page.

Note the name change of the site from "5 Bears Engineering" to "5 Bears Research". Apparently, the Texas State Board of Professional Engineers took exception to the use of the term "engineering". I am not a licensed, scrutinized, regulated, certified, bonded, taxed, and otherwise "real" engineer, hence the forced name change. Makes my head want to explode. For those baffled by the name, my last name, Bjorn, is Danish for "bear". There being 5 of us in my family... 5 Bears!


5 March 2002: - I had a blast yesterday running my turbine for almost 1/2 hour total. Lotsa kero converted into heat and noise. I added a bunch of new MPEGS to the downloads page, mostly of the LCD display from the 5BECU in its current coding state. Mpegs of a running turbine just aren't that cool unless torching or other excitement happens, and that is my goal, to make such excitement rare. Look for the little symbol on the mpegs which I have added. I apologize for the size of these. Anyone know of an MPEG editor which will allow me to clip, or otherwise edit MPEGS to make them a bit smaller?


27 Feb 2002: I have decided to pursue construction of the Wren Propjet. Rather than convert my current thrust MW54, I plan on building another MW54, this time an appropriate gas gen. version suitable for the turboprop. Ultimately, I'd like to make my turboprop as a modular unit, meaning items such as the fuel pump, filters, sensors, any external oil reservoirs, and the like, be more or less permanently mounted as "accessories", as the real turbines do. Plumbing on the jet would be metal tubing. All that would be required to run the engine would be a fuel line and an 8 to 12 pin milspec style DIN plug into the ECU. I am already hard at work on a dual-processor ECU which will be able to accomodate a second tach input (for the prop shaft), as well as pressure sensors, etc., pretty much any analog or digital signal for future use.

I also rewrote a bit of the start fueling code for the current 5BECU and had some astonishingly fast auto-starts. I'll take some MPEG movies and get these posted.


19 Feb 2002: Testing of the Radiometrix modules has confirmed the ability to transmit turbine data over relatively long distances! It was a simple matter to patch some code into the ECU which packetizes critical numbers, performs a checksum addition for data integrity, then transmits the whole packet in one burst. I found that 2400 baud transmission seemed to deliver the most consistent results. While this is slow compared to what we work with daily on computers, there are only 8 to 10 bytes which are delivered in each packet, and one packet is enough to generate an LCD "screen" of data.


11 Feb 2002: I have updated the 5BECU manual. This version is current and will be the correct manual for the first release of the ECU. The format is Adobe Acrobat Reader (*.PDF) V5.0.

The shipment of the RadioMetrix modules was delayed. They should be here on 14 Feb. Tests forthcoming. Keep your fingers crossed.

Wren Turbines has begun initial sales of critical turboshaft engine components! Now I am faced with this delicious decision... should I pursue and serialize the awesome Wren turboshaft engine, or should I serialize this incredible Pratt & Whitney Wasp Jr. 9 cylinder radial in 1/6 scale... Sooo many engines, sooo little time.

PCB's the easy way!

These nice traces were created with a toner transfer system. This small board is an experimental board for the RadioMetrix TX3 RF module for transmission of turbine data to a ground display.

Find out how to make simple PCB's like this one by clicking here.


8 Feb 02: Radiometrix Modules are on the way! Hurrah for the Brits, who make these modules for both European (868 MHz) and U.S. (914 MHz) markets. 1 each TX3 and RX3 RF data modules for 914 MHz should arrive by 7 Feb. The data sheets look promising for transmission of critical turbine data to a ground display. Big questions are the potential for interference between the modules and the R/C system, and filtering the rather noisy uProcessor / output sections of the ECU, keeping this hash from overwhelming the low powered transmitter. The EGT section is also prone to noise, as the thermocouple signal is quite miniscule and the long leads of the EGT probe can be easily overwhelmed with RF, causing false EGT indications. Testing to commence.

New PCB (Printed circuit board) art - See the Express PCB art for the 5BECU and two conceptual designs. Please don't ask for Gerber or similar PC art files; I use software from Express PCB and am happy with both the software and the excellent finished product. If you want a *.PCB file, I'd be happy to supply one. Be warned that small quantities of boards from Express PCB can be costly. The best way is to bulk order, thus bringing the per-unit cost down.

5BECU Status - The AECU project has been complex and taken far longer than I anticipated. Please remember that this is still a hobby site. I am a family man with a "real" job, with normal responsibilities, and I simply cannot proceed at a pace which a dedicated individual or team can manage, one devoted to a specific goal. Please be patient with me. I have also posted updates to the ECU page. Several people have emailed asking for schematics, parts lists, etc. At this time I am not placing everything into the public domain. Perhaps at some point in the future... thanks.


4 Feb 02: I have produced a PDF document of the current manual for the AECU. (275K). Feel free to download and browse. The current code revision has added elapsed time, simplified programming, and added some excellent fault handling logic for both flameout and overtemp detection, using trend information. This simply means that the ECU can detect to a certain degree if the turbine is recovering from an EGT excursion, or if in fact the situation warrants a fault shutdown.

Testing and data are now being gathered from several turbines in the U.K., Canada, and my own here in the U.S. Mr. John Wells in the UK is running a gas generator MW54, and has provided critical input, as his turbine has exceptionally low EGT; the flameout logic was revised and a new SETUP page added to allow user adjustment of flameout threshold EGT.


6 Nov: We are finally acheiving good success with Bill's KJ-66. Apparently, an overpumped fuel system was a portion of the problem, as Bill was spanning only 15 units of pump output from Idle RPM to Maximum RPM. We also made some additional code modifications refining fuel delivery. Much of my early progress in the software was dedicated to the MW54/Orbit pump combination that I am using, without adequate thought to other systems. I tended to optimize the code for it, rather than for model turbines in general. To solve this problem, I took a long, hard look at how my ECU maps fuel to requested throttle, and how that relates to RPM. The new coding now relies heavily on a time-based ramping scheme.

The new code improves upon the old by altering the run loop fuel calculations. It incroporates compact and intelligent algorithms with automatic EGT limiting, pre and post ramp tapering, and trim at 0% and 100% throttle. Most importantly, it interprets the pump output range, from 0 to 100%, and incoporates that into the ramp profile.

First, realize that DC motors (such as a fuel pump), are normally controlled using Pulse-Width Modulation, or PWM. 8-bit PWM allows one to control a DC motor with 255 discrete steps. The 5 Bears ECU uses 10-bit PWM, but 8 bit is easier to understand. Since fuel plumbing, pumps, and turbines vary widely, one cannot assume that a range of, say, 30 to 100 pump units will work. Since we want a given turbine to accelerate from idle to max in a short period of time, I arbitrarily chose 3.5 seconds as a desireable starting point, with the user having the ability to adjust this via SETUP on the ECU. For example:


Idle RPM Pump DC

Max RPM Pump DC

PumpDC Range

Turbine A




Turbine B




Note how turbine A has a wide pump DC range from Idle to Max RPM, while turbine B has a narrow range. The PumpDC Range is correctly integrated into the ramping profile. This means that if both turbines have a Pump Ramp Up of 16, a throttle burst from 0 to 100% will result in an acceleration profile of identical time periods, roughly 3.5 seconds from idle to max. Pump Ramp Up settings below 12 are extremely aggressive and must be tested with caution, although the EGT limiting should keep things safe... the most likely result would be a minor RPM overshoot. Ramp settings above 30 are very mild.

So what does this all mean? The latest firmware correctly programs fuel delivery for a wide range of systems. This code has been tested with great success on my MW54 and on Bill's KJ. It has reached a point, after 15 gallons of kerosene and countless test runs, that I am comfortable with releasing the ECU. I am rewriting the manual, a 30 page Word document. If you want a .ZIP copy of the manual file, please request it via email and I will be happy to send it to you. Bill is also hard at work on kit documentation and setup.

On a happy personal note, I flew my first revenue trip in the Boeing 777 for American Airlines, from Dallas/Ft. Worth to Chicago. All I can say is, it is a helluva transport, beautifully made, easy to land. Boeing has a real winner here!


20 Oct: The ECU is being tested on my friend Bill's KJ-66 as I update this page. Starting fuel was too aggressive, resulting in a large overshoot of idle and some torching. I am refining the code to fix this. I was a bit surprised that a start profile so suitable for an MW54 would be entirely too much for a KJ-66. I have also made the user entered Pump Ramp Start value more flexible and able to deal with different turbines' appetites for kerosene during this critical phase.

Here is Bill's exceptionally clean and functional test stand, complete with dual pressure guages, thrust transducer, backup EGT meter, and an uncased ECU in the foreground.

Bill didn't mention the function of the orange button at the very bottom of the picture, but I would guess it is a big "panic" switch, always a useful accessory when testing new equipment!

I am also experimenting with a new method of ramping fuel during the run phase. This newer technique feeds the RPM back to the ECU in a different way, resulting in a proportional throttle setting and more predictable behavior. This also means that a throttle setting of 75% will deliver exactly 75% of available RPM. The older method compacted far more thrust variation in the upper 1/2 of throttle stick travel than the bottom 1/2.

The electronics and safety features have exceeded my expectations, with the ECU responding correctly to every fault I have intentionally or otherwise caused in the system. Refinement contines, but I feel that I am very, very close. I am currently going through Boeing 777 training for Amercan Airlines, and this has really put a cramp in my development time.

The custom type-k probes from Omega have arrived! These are sized perfectly for the MW54 or any smaller turbine. The probes are 4" long with a .062" Inconel sheathe, and a 1/4" diameter stainless junction for mounting. They are available at the online store.


16 Sep: Grrr. My electric start clutch has shredded itself. I had been having trouble with electric starting these past few weeks, with the starts being consistently hot. I believe I may have found the problem - the inlet cone showed signs of rubbing on the inner portions near the periphery of the compressor. I cleaned out the inlet cone a bit more and worked on air starting while a new Bendix was on the way.

Previously, I had been using normal compressed shop air via a blow gun for starting. This worked but was not optimum. Cleaning the yard Saturday, I realized my 2-stroke leaf blower provides serious airflow and would be a much more satisfactory air source than my heavy compressor. A bit of work with some PVC and flexible ducting resulted in the rig on the left. You can see the blue hose ending in a 90 degree PVC fitting.

The middle photo shows my Wren with the starter motor removed for Bendix replacement. Since one leg of the starter motor pod housed my tacho, I removed just this leg from the pod and reinstalled it. Note the large air gap between sensor and spinner nut, no fuss at all with the Hall IC sensor. This single leg support works very well.

The right photo displays the PVC air supply held by hand to the inlet. I had to make a cut in the nozzle for the tacho support leg, but otherwise it is very simple. The first start using the modified leaf-blower was a revelation! The ECU was able to feed the fuel aggressively, and the turbine was at idle within 15 seconds or so, with the EGT in the 400's all the way! Consider these leaf blowers... they are very portable, easy to use, and FAR more powerful than a modified hair dryer. They can also be had cheaply, especially when used.

Guys, my web site feels like one long apology these last 3 months. I simply cannot release the ECU until I am confident in every way that it will provide satisfaction, and that includes the auto-start capability. So until I can verify the current start fuel coding, I need to hold off on its sale in the online store. Thank you all.

Oh yes, one interesting occurence today during a run... the fuel line from the fuel pump to the fuel filter let go at 145,000 RPM! A jet of fuel began to spew from the pump and the turbine rolled back immediately. My first instinct was to hit the panic switch and kill the power, but I decided to see how the ECU would react. It correctly detected the flameout and secured the system, and the fuel was cutoff nicely. It was good to see it perform exactly as intended.


1 Sep: Here's what is happening... I have been working with a nice and knowledgeable fellow in Canada named Bill Blackburn who has been with me almost from the beginning of the ECU project, offering excellent advise and practical knowledge. Bill will hopefully be working on a kit version of the ECU, which in many ways is more complex than offering completed units. The kit will probably include the printed circuit board, a programmed PIC chip, and the difficult to source integrated circuits, along with a machined case and sundry fittings. The LCD display will probably have to be offered completed, as it is a very dense fit inside the case, and the soldering is point to point in some places and can be tricky. Instructions will have to be drawn up as well. I will post more news on the "kit" if and when it becomes available.

I have installed the new sticks from Wren and had several excellent runs. The ECU coding is in the final tweak stages. I am optimizing the fuel delivery to allow a good, crisp acceleration. Air starts have been cooler than motor starts, and I have been talking with the fine folks at New Creations RC who have an extensive knowledge of electric motors and batteries. Kirk Massey has been helpful in selection of the proper batteries for the ECU, one which will deliver the higher energy needed to crank the turbine, and also provide the correct capacity for extended flight. Once the best battery has been determined, I will offer these online. I am also going to test a higher energy motor for starting than the rather modest Speed 300.

Omega Engineering has an order in place for some excellent custom type K probes, inconel sheathed, dimensioned for the Wren 54. Once these are on hand, I am going to engineer a very simple bracket which will mount onto the MW54 case using the pre-existing case front external cap screws. The probe length will be 4", and will run parallel to the turbine, with the final 2" curving gracefully into the outer cone.

Gathering the necessary stock for the online store has been difficult and slow. Thank you for your patience!

24 Aug: A Tale of 2 EGT Probes - OK more than 2 probes. From my earliest running of the MW54, I have been baffled by EGT probe placement. I did some experiments recently which showed that even minor changes in probe placement can cause EGT fluctuations (constant power setting) of tens or even hundreds of degrees. Not knowing which is the "right" temperature, I asked the guys at Wren. What they said makes sense. We are concerned with what you can call the average work done by the turbine. Gasses exiting the turbine at different spots will obviously have different temperatures... those with lower temps have transferred this energy to the turbine and thus to the compressor, happily making the turbine sustain. The turbine, being metallic and in fast motion, transfers heat along the blades, hopefully allowing the temperature the turbine wheel "sees" to be somewhat of an average of these escaping gasses. Any guys disagree with this? Am I off base here? Let me know.

With that question settled, I decided to check the accuracy of several type K probes I have on hand, comparing them to each other and to my ECU. I have an excellent Omega TC meter which has 2 inputs, allowing me to compare 2 probes simultaneously.

The first test was to clamp two probes and heat the ends with a hot air gun, which can deliver temps up to 600 C or higher. The two probes showed huge variations, often 50 C or more between them. By shifting the hot air gun slightly, the temps flip-flopped, indicating that the hot air stream was not even. I next buried the probes in some core sand, and heated it up with a propane torch to 600 C or so. Again, the two probes did not match.. the sand, like the hot air gun, did not distribute heat evenly.

The only thing which would test the probes with any consistency was a molten bath, either salts or lead. I decided to fire up a lead pot and find out once and for all if our typical probes can be trusted to be reasonably consistent.

The lead pot has sufficient mass to provide a good, stable source of heat. I was more concerned about probe and meter variations than I was the actual temperatures.

The first test was to simultaneously measure two probes with the Omega TC meter. This showed an excellent accuracy, with no deviations beyond 1 degree between any two type K probes.

The next test was to measure the ECU versus the Omega meter. One probe was connected to the ECU, the other to the Omega. I was quite pleased to see the two agreed within a degree or two, even when probes were swapped.

Overall I have gained confidence that the typical type-k probe from a reputable supplier will do a fine job measuring EGT. Place the probe axially to an average location on your turbine. This also sugests some experimentation would be appropriate for initial installations in a new turbine to avoid misleading high or low temps.

23 Aug: The inconel sticks from Wren have arrived, and they are exceptionally nice. Thank you Roger and the gang for helping me out on such short notice. I plan on securing the CC rear (the sticks came with the rear as an assembly) with inconel tie wire. SS wire would probably do as well as this portion of the combustion chamber doesn't appear to be subject to excessive heat.

The new inconel sticks as recieved from Wren. Weld quality was excellent. The sticks are very uniform, and slid right on to my machined NGV with just the right amount of force... no modifications required at all. It does pay to machine your parts to print!

I plan on reassembling the engine today, verifying the final ECU coding with an intact and properly functioning engine, and finally getting the store up and running. It's been a long run, but I think it will be worth it!

I had promised earlier to show some pictures from my torn-down MW54. There really isn't too much to show. With the exception of the original sticks, everything looked pretty much as it did when the engine was first assembled. My swirl jets showed some very minor corrosion and scaling, but were otherwise OK. These I had made from 316SS stock, which is a fairly corrosion resistant alloy.

Here, the case rear, NGV, and combustion chamber are being slid off the back of the outer case. The fuel injection needles have backed off slightly and can be seen between the NGV and the combustion chamber. One needle was partially blocked with carbon, but this was cleared with a strand of copper wire, and the needles passed another flame test after disassembly.

The turbine wheel, nut, rear bearing, and shaft. Note the scoring on the right portion of the shaft... this was caused by a sloppy preload spring on the very first run attempts. The scoring was more cosmetic than anything and the shaft was serviceable after replacing the preload spring. The bearings looked new, due to the well-designed air filter on the MW54 and apparently excellent heat transfer characteristics of the tunnel, keeping excessive heat from the precious ceramic bearings.

More Bad News: 28 July - Bad things seem to come in groups. The last few days, I had been working feverishly on refining the code for the ECU. My engine was running hot, and I thought it was the software. Finally, I had reached a point where I couldn't even get the turbine started without frightening temperatures. All along I thought it was my coding, but a teardown of my MW54 revealed the problem... severe erosion of the sticks at the forward ends was the culprit.

First, the good news. The interior of the turbine other than the sticks was perfect. The bearings had a nice pool of lube keeping them in order, and the races, both inner and outer, were as bright as they day I had installed them, indicating excellent thermal properties and nice cool running. One compressor blade had a small nick (probably from a bug) which I dressed with 400 paper. Other than carbon, the ngv and the turbine wheel were pristine. The fuel injector needles were likewise in excellent health. All of the surfaces which I had nickel plated were smudged with carbon, but underneath were still perfect in every way.

The outer and inner CC wrappers were surprisingly clean, and the swirl jets were still in their correct positions, having shifted not at all.

But the ends of the sticks showed some very odd patterns. 3 or 4 were split longitudinally a good 1/3rd of the way from the front to rear, allowing fuel to escape well aft of the correct location. Others were "pinched" more or less shut. The surface of the sticks was fluffy with corrosion, and pieces of the sticks could be flaked off with a scribe.


Wow! I expected some wear or erosion after much hard running, but the damage was stunning. No wonder the engine's characteristics were becoming a bit ragged.

Note the stick upper right which is split a good 1.5 cm. The sticks with the worst erosion were those which the swirl jets most closely blew across. Per the instruction manual, the swirl jets were correctly placed to blow not on the sticks but in front of them, but I theorize that enough fresh, oxygen-rich air struck the red hot sticks to erode them in this fashion.

If you are making an MW54, get the inconel sticks. Also, don't do what I did, which was secure the CC rear to the rest of the chamber via welding... do what the plans suggest and use stainless wire to secure the CC rear/sticks, allowing you to replace just that part rather than the entire CC, which I must now do.

While not a death knell for the ECU project, this is going to further delay it. I'm very sorry, I know there are a lot of guys who are interested, but I am not comfortable releasing it until I am 100% sure of its performance. More good news for the ECU at least - I have beat the prototype to death with massive running and the circuitry is working exactly the way I intended, and until the erosion became evident, performance was excellent.