[peteeng]

Hi all you lot! - I've been working on a 5 axis router as a precursor to Milli. There are two designs at present A) a trunnion/rotary design and B) a BC 5 axis design. I'm having trouble getting a suitable slewing bearing for the trunnion rotary. Most are for heavy equipment so are really big for the job. A small crane would be OK on this sort of 300mm thing. A THK one sourced from OZ is $4250AUD WOW!! a Chinese version similar about $300USD.. But I like the BC model. I can make this from current parts I have and just make the BC head... but I'll keep at it and see which one wins. Peter

Note some of the parts in the trunnion model are leftover parts from other machines. I think I'll start from scratch once I get a bit further down the road and tailor solutions to composites and grout...

[joeavaerage]

HI,
the cost effective alternative is two opposed taper bearings, say 100 to 120mm OD and about 75 to 100mm apart. Not as low profile as crossed rollers, but not bad either.

Crossed roller bearings are all very nice, but it's just to damn expensive to get them.

Craig

[joeavaerage]

Hi,
you have made no provision for driving the fifth axis, The common way is a worm drive but that will require a motor and worm all of which add to the profile height.

Maybe you could start by saying 'I want no more then 100mm in profile'......and then see what sort of components would allow that. Personal experience would suggest that 100mm
profile is very optimistic.

My fifth axis is 175mm total profile height. Note that includes the trunnion and the 45mm platter. In the attached pic the 32mm steel plate that forms the base of the trunnion
fits between the underside of the platter and the upper surface of the worm reducer. The worm reducer has the two taper bearings and seal and obviously the worm and worm wheel.

Craig

[peteeng]

Hi Craig - The rotary size requested was 400mm dia. This is very big to support from a central bearing, That's why I want a slew bearing,. I've knocked the rotary down to 350mm dia. Its a router not a mill so the plunging loads are smaller... There are no drives in the model yet as I'm figuring out clearances, interferences and the general arrangement. Waiting on some feedback BC head vs trunnion. Currently I prefer the BC head arrangement . I think the trunnion will be integrated into the machine walls vs having the columns if it goes the trunnion style... Peter

I've just added an ISO20 1500W ATC spindle to the model and it makes the BC head really long!! Every corner taken is new challenge...

[joeavaerage]

Hi,
taper bearings are used as wheel bearings. Two tapered bearings say 120mm in diameter will easily bear a disc of 400mm, if not every truck and trailer axle out there is going to
fall to bits.....and that's not happening.

What I like about the trunnion fifth is that it is rigid....commensurate with a steel capable mill.

I suppose with thick enough sections, outsize bearing etc you could make a BC head rigid also, but I suspect the idea of design simplicity would evaporate.

Craig

[peteeng]

Hi Craig - I'm sure two mini hubs will support the rotary, its not about strength its about stiffness. And adjusting two tapered bearings seems simple but its not, plus all the other parts that have to be made to make that work. Shafts, spacers, nuts then it has to be driven. many things to solve. A unitised bearing with a 150mm flange seems to be the go.... Unless I just use the Chinese thein flanged slew bearing.... Peter

slew bearing attached $200USD plus freight

[joeavaerage]

Hi,

I'm sure two mini hubs will support the rotary,

Mini hubs....what are you talking about??? Tapered bearings of 100mm to 120mm are seen on truck and trailer hubs, vastly stronger and stiffer than mini hubs. You think about the huge stress that
go on to a trailer wheel, not only the weight but the cornering loads...and yet you are of the opinion there not stiff enough? BS.


Craig

[peteeng]

Hi Craig - I do deal with a lot of large Timken bearings, here's one that goes on one of my trailers that weighs 116kg and its radial rating is 623 tonnes. The dims are inches... and here's the shaft it goes on coming out of heat treat. I've also attached a unitised bearing used on a mini and it will fly through the application. I could use a jeep one also... The minis static rating is about 4.2T so that's good enough for a small router.. Peter the jeep is about 4.9 tonne. I'll keep whittling away at the issues... Peter

[ardenum2]

cross roller slewing rings aren't the most compact, true. but there's the option of cross roller bearings, these guys are very compact, just like your regular bearing. In cross section. The question will be, small OD YRT bearing stiffness/price vs large OD cross roller bearing. As you figure out the drive, one will become more appealing.

[peteeng]

Hi Ard - Yes appeal for different solutions waxes and wanes... Ideally its a bearing with flanges cross drilled so I don't have to make the flanges. I think for this machine the automotive Hub 3's as they are called are the ones... I have used them on a scara robot some time ago... I'll have to buy one to be able to model it... Peter. I found some agri bearings but they have a bit of a wobble in them... then I have to drive themmmm....

[joeavaerage]

Hi,
all of the unitised wheel bearings I've ever fitted, and I fitted quite a few over twenty years as a mechanic, all have a little clearance. They are after all made for highspeed operation
and so a little clearance is mandatory....not so for CNC bearings where speeds are slow but rigidity is upmost.

Craig

[peteeng]

Hi - The catalogue states that "non driven" wheel units are preloaded, there is a preload discussion section.... I could use Hub2 bearings and adjust the preload myself.... But then I have to make flanges and shafts. More details to sort through, pick your poison. Think I'll order one and have a "play" Peter

[joeavaerage]

Hi,
yes they widely and readily available, and modestly priced overall. If there is a suitable combination it would be a real step in the right direction, ie trying to build a cost competitve fifth axis.

The worm reducer I pictured earlier is new old stock and cost $180USD. The list price is over $1500USD. It has <1 arc min lash, and its as solid and rigid as can be, proper German made engineering components.
If you are going to buy a bearing just to experiment with then you should consider one of these Atlanta Drives servo reducers, even if only as an experiment, its likely as cheap as a wheel bearing after all.
The reducer is the whole deal, opposed taper bearings adjusted to zero lash, adjustable precision worm adjusted to < 1 arc min, housing everything. Add the platter and a servo/stepper and your in business.

This is just one example:

https://www.ebay.com/itm/20230179089...Bk9SR7DB0oGlYg

This is the same size as my fourth axis and you'd be happy to swing a 20kg platter with another 20kg as the part with it. They are seriously nice bits of gear and hard to beat the value at $149US.
The list price is over $1500USD.

Craig

[peteeng]

Hi Craig - All of my CNC designs are intended for production & they are Hobby/Maker level machines not commercial level machines (so $$$ are tight as the buyer is tight). So I can't spec surplus stock as it may not be available when the time comes to build 10 units. Hopefully more. So design solutions have to be production scalable, with readily available parts at good prices. Not always easy to do, 100's of hrs sorting solutions and suppliers is the usual task on a new design. A bespoke design is different can just dive in and make parts and get it done... Peter

[joeavaerage]

Hi,
that's why after three years of this thread running you have not actually built anything yet.

Good luck.

Craig

[peteeng]

Hi Craig - I have designed and built two other machines and tested a few control systems in this time frame. Designing a good production machine takes time and Milli is not a priority project. It's a development project, especially in regard to materials. It takes a long time to acquire materials, make coupons and have them tested then get over the disappointment. I'm still waiting on Sika to deliver something they said would be here a week ago... everything blows out... Plus last year we had a flood which has screwed lots of projects over the last year, I'm still doing stuff thats related to that event that takes up my time...Once you have the right answers things can accelerate. 3 years hmmmm, time flies when your having fun .... Peter

[ardenum2]

that's why after three years of this thread running you have not actually built anything yet

pretty sure Pete build a few other things in the meantime.

What about you? You take commissions? How much do you want to develop a drive for iron core linear motors? Board must be producible and assembled through an online service like pcbway or jlcpcb or similar so it arrives ready-to-go. What language do you use to program chips by the way?

[peteeng]

Hi All - So I have quickly reviewed the thread as it has come to my attention that's its 3 years old. The thread has wandered a bit but the main idea has been material development. In the last 3 years I have built a workshop, built 4 routers, tested several controllers, endured the pandemic holdups on everything, had a flooded workshop and half my block, of which I'm still cleaning things up and it happened a year ago... but we move on. My epoxy+stuff has been disappointing in terms of stiffness. Fibreglass and carbon fibre are still the top materials to beat outside of metals. Steel and aluminium are the easiest metals to make machine parts from. Cast UHPC or grout is the easiest material to cast parts from if you can design with the low modulus (compared to steel, CI or AL). Fabricated glass or carbon is complex but results in good parts. Laminated metal is attractive as I can make larger parts from laminated aluminium or contract laser cut steel to laminate from. So the sika grout will be the last material left to test then a result matrix can be established. I leave out cast iron as it's a known material and known process if you want to go that way, same as if you choose steel, the design and build process is well understood. So onward to wherever this leads. Peter

[joeavaerage]

Hi,

I have designed and built two other machines

And have either of those designs resulted in production orders? All very well researching and developing a design to a production ready state, but if it takes years then the market has moved on.

What about you? You take commissions? How much do you want to develop a drive for iron core linear motors?

I assume Ardenum this question is directed at me? The fact is I don't really want to try to develop a linear servo. I'm sure you are aware just to buy a single seat license for motor/magnetic software costs a small fortune.
Second issue is that there are many companies around the world with vastly better resources than I trying to do the same thing. The probability of me coming up with some new solution or technique or cost
competitive design is slim, not non-existent but slim. Having designed and built my own servo drive I can tell you even a basic Field Oriented Control servo drive is not easy. The servo drives offered by commercial vendors
represent years of design and development, ten or even hundreds of thousands of man hours of design time. How can I compete with that?....I cant.

I own a business that repairs automotive instruments. Yesterday I had a commission to build a tacho circuit for a Range Rover to which a Nissan engine has been fitted. I designed a new PCB, admittedly re-using circuit ideas
that I use elsewhere. I used my mill to make the board, I then populated it by hand, programmed it (it is a uP design) and fitted it. The company has already paid the $350 plus tax that I quoted them and the job is gone.
Started yesterday morning at 8am, finished by 11:30am. That afternoon I made another PCB for a truck speedo, a $400 plus tax design. The job was finished shortly after 5:00pm last night.

These are the sorts of jobs that I'm good at, and there is, to my knowledge, no-one else doing that sort of thing, and it pays. I don't make a fortune but I do make a living.

By comparison developing a linear servo sounds like a sure way to invest big dollars, inordinate amounts of time with a slim prospect of commercially successful result.
Meanwhile I have customers wanting me to finish a new design for Nissan Terrano (and similar) speedo PCBs. They would cost around $600 to which customers have responded
'I've got to have it', so the price does not deter them.

I have taken on overly ambitious electronic design projects before, a extra high voltage electrostatic precipitator is a good example. I spent many months and some thousands of dollars getting it to work,
before realising that I was never likely to get this device approved for domestic use in New Zealand, it runs at around 75kV. Electrostatic precipitators are industrial machine....for a reason.

You might say 'that I pick my battles', and with age my success rate is still increasing, not because I'm getting ant smarter but am making smarter choices about where I invest time and money.

I would, as a pure hobby, be interested in designing a linear servo drive, they are not hugely different to my existing rotary design, but to try to design and build the electromagnetic parts
does not appeal.

Just as a matter of interest during my first year at University studying Electrical Engineering, myself and about ten other students made a model electric train, ie a linear motor. We had to hand wind the
coils, and we had about 12m of 'track'. Took us weeks on and off, but we did it. Even the Prof was impressed, the 'train' disappeared off the end at something like 15m/s! We were grade on it, and
it was quite a bit of fun, but more importantly taught us the fundamentals of moving magnetic fields.

What language do you use to program chips by the way?

The majority of software development tools written for micro controllers use C/C++. For highly intense processing like a FIR filter you write in assembler, tediuos and demanding of concentration
but blindly fast results. Most of my Automotive stuff I use Microchip 8 bit AVR devices. For servo drives I use Texas Instruments 32bit C2000 with single cycle floating point multiply and accumulate.

Craig

[joeavaerage]

Hi Ardenum,
I use this development board:

https://www.digikey.co.nz/en/product...BoC_DoQAvD_BwE

Lots to recommend it. The micro is optimised for rotating machinery and has all the periphials that you can dream of and more, including really tightly timed complementary PWM with programmable deadzone, sequential PWM
signals for polyphase systems, cycle by cycle current and voltage limits, dual encoder inputs, plus all the usual timers,analogue inputs and all the rest. The board has a galvanaically isolated D100 JTag. If you use the
onboard JTag then you get free unlimited use of CodeComposerStudio, Texas Instruments own IDE including the Insta-Spin software libraries. You can use this development board right up to production. In my case
I just drop it into my drive complete, saves having to write a bootloader. All-in-all Texas Instruments have tried to squeeze everything in that you might ever want, all for free or at least cheap to encourage your use of
Texas Instruments silicon. This particular IC (TMS320F28069M) is about $20USD each (1000 unit rate), so hardly expensive.

Craig