new build opinions required

[Boyan Silyavski]

Silyavski If you are thinking of Linuxcnc and need to slave motor on one axis there is a problem with homing that has not been ironed out yet. See here LinuxCNC Support Forum :: Topic: Homing and limits with X axis 2 joints (1/2) and here LinuxCNC Support Forum :: Topic: Configurate Slave Axis (1/4) This might be being fixed in the next release. ..Clive

Thanks for the info. Though here is what i think of it in my case, may be concerning this build also:
-if the machine is rigid like mine and like this will be , may be it shouldn't be a problem. With Hiwin 20 HA bearing spacement at the gantry sides of 40cm/ or even 30/ i don't see how it would go out of true, its not a flimsy machine we are talking about here.
-i would use servos, they will simply stop due to error

again, a few very important bits of info between you lot there.


2 rails per x axis side, one above, one below, 2 barings per rail, seems to make the most sense on an equal loading/force approach. Maybe its not a bad idea to have the barings as the week point, Because they are cheeper to replace in time and money, compared to unbending a machine gantry. If, from the design stage, you consider the barings as consumables ( not to short a life time though ). it free's up some design options.

I still don't like that design of yours, but that shouldn't discourage you :-)

R&P suits a heavy gantry, and i can see why most heavy machines use it, but, when your building your 80k to 100k + to sell to a customer, the customers not going to acsept thaty every n100 hours of running, they will have to replace the barings, and every n1000 hours , the ballscrews, so you design a machine thats heavy enough to absorb the forces, to give your barings and ballscrews some help. but the extra weight requires bigger motor or servoes, and causes other problems, or pushes you design in certan directions.

I don't know if i am understanding you wrong. But:
RP is far more wear prone than ball screw. Ball screw is far more better and smooth. for 6m long table, yes RP, but for anything under 3m-ballscrew.

wereas, if you specificaly design your machine for the barings to be the week point, you can make you Y gantry a box, that contains a Z Axis, thats a box iteslf with rails on both insides of the Y gantry. The hole Y ganty, clamped by rails and barings on the top and bottom of the X axis, with the ballscrew in the gap between them (between the edge of the Table frame/X axis and the Y ganrty box ).

IMHO the weakest points in order of a DIY machine FYI:
1. Router bit - deflection
2. Z overhang-plate twist
3. Gantry twist and bend ,
4. Gantry sides
5. bearings/assuming you use square
6.Spindle power
7. Balscrew support bearings, spindle bearings
8...etc

My main reason for designing like this, apart from equal forces in all directions, is machine down time if a mistake is made, This is through experiance of owning/running a Gocarting track 97-2001. Go carts have lots of expencive barings in them, and after 2 years of large parts bills, we redesigned and refitted the frames, to eas of on the barings, What a mistake, we spent a lot less on barings that season but down time per cart frame due to bending was horrendus, each frame had to be stripped and bolted in the jig for 2 days, heated and coolded, then brazzed, then rebuilt, That job went from once after the end of each season, to frames queing up to be placed in the jig.
My point being, make a mistake and bend your machine, your down time is a lot more (maybe days ), and you dont know what the cost of repair might be, whereas, if you pop your barings, its costly, but you know how long it will take and how much it will cost before your machine is running.

You can not bend with steppers a properly constructed machine. They will simply stall. Remember this. Nor you can wear Hiwin bearings if properly mounted. When you have them in your hands you will see why. They will be probably that last element in a machine to fail.

Back to the BOBs

I have another questions in mind concerning the BOBs. Please some one more knowledgeable clear it for me:

-With steppers: If Mach 3 outputs as max 100kHz pulse/ even less normally/, then a normal reliable board that supports 100khz should be ok, yes?

-With servos: If Mach 3 outputs as max 100kHz pulse/ even less normally/, then why should i care to buy a board like the CSMIO or other? I know, apart from the reliability and connections, i mean. Every modern servo drive can upscale by gearing the frequency and generate the proper signal , so why waste money on expensive cards that will do exactly the same thing in fact? I start to believe that the driver generating the pulses will be quite better and reliable..
It seems that mach3 is the limiting factor, not the board?

[JAZZCNC]

-With servos: If Mach 3 outputs as max 100kHz pulse/ even less normally/, then why should i care to buy a board like the CSMIO or other? I know, apart from the reliability and connections, i mean. Every modern servo drive can upscale by gearing the frequency and generate the proper signal , so why waste money on expensive cards that will do exactly the same thing in fact? I start to believe that the driver generating the pulses will be quite better and reliable..
It seems that mach3 is the limiting factor, not the board?

Mach3 is only the limiting factor when using the parallel port and even then it's Not really Mach3 it's the Parallel port and it's driver.
When you use an external motion control card all these restrictions are taken away and infact Mach3 does very little other than watch I/O's and this is why low spec PC and Laptops can be used.

Now with servo's the freqency rate becomes important because the encoders and depending on the encoder count will depend on if you need higher than 100Khz. Which for most servos you will with out using the electronic gearing feature built into the drives.
Most who are using servos avoid electronic gearing because it lowers resolution and often the whole point of servo's and encoders is to give high resolution.

Now lets say you are using typical servo's with quadrature encoders having 2500cpr this means you need 10,000 pulses per rev so divideing the controler Pulse rate (IE: KHZ) by this amount gives Pulses per sec then times x 60 will determine the maximum speed you'll get from the Servo's. . .IE: 100Khz / 10,000ppr=10pps x 60s=600RPM

So has you can see 100Khz won't be much good if have 3000rpm servos has the fastest speed you'll get is 600rpm.!!

Now if the drives have electronic gearing multiplier then you can use this at the expense of resolution. So when 4x multiplier is applied then for every 1 pulse Mach3 controller sends out the drive will see it has 4 so 100Khz becomes 400Khz but the resolution is divided by 4.!!
400khz/10,000x60=2400Rpm better but still not enough so you'll need to use the nearest or above to give the 500Khz you'll need for 3000rpm with 2500CPR encoders.!!
Electronic gearing can be ok for some machines because the resolution will be very high to start with.!!

The next problem comes from the BOBs if used and it's Max pulse rate causing a bottle neck and yep most typical BOB's don't go that high so you'll need a very fast BOB and high quality BOB if your using servos with high pulse counts.
This is also why lots of High end motion control cards don't use BOB's so to speak (IE Csmio) and connect direct to any devices attached or thru dedicated daughter boards which don't cause a bottle neck.!

So be warned to all those dreaming of using servo's just be aware they take you to another all new level of head scratching and expense.!! . . . . . So I honestly suggest you ask your self do you Really Really Really need Servo's.??? . . . . . . . Oh and just be aware they are still not closed Loop with Mach3.!! . . . . . The best they can do is Fault if they see a following error.

For 99.9% of DIY machines and even relatively slower Light industrial machines steppers if spec'd and setup correctly are far far far less hassle and plenty accurate enough.

[Boyan Silyavski]

Dean,

thanks for the great input, cleared some clouds :-) , but still my question stands:

Mach3 outputs normally 25khz pulse. On a perfect PC it could output stable 100kHz /or even up to 300kHz before it locks, on a top computer, according to its creators/ .

Now, how these 100kz become 1Mhz or 4Mhz on CSMIO, Galil or whatsoever when mach3 outputs only pulses, not trajectory information. They generate them, the same way a geared servo drive generates them, so in fact - no real resolution benefit at all with expensive boards for the $$$, coupled with servos, steppers and Mach3. Apart from the reliability , ethernet or usb connection and I/O. It seems to me that these Mhz they are offering are a hype. No body can prove to me that a generated signal from a BOB is better from the same signal generated from a Panasonic servo drive electronically geared, for example.

Dont ask me who put this in my head :-) but now i have it clear after that digging, it took me a week to wake up from the hype- until Mach3 or 4 or.. closes the loop at the PC and transmits trajectory signal to the motion control, i wouldn't care for the Mhz of the boards. It wouldnt happen on windows, except on some kind of dedicated hardware pc probably.

That investigation answered the other questions of mine, why the good DSP motion control cards have + encoder inputs, / not only for Analog servo drives i mean/ . And why a separate dedicated motion control solution for a machine is so darn expensive.

Now seems Linux a kind of do that what we are speaking of. Wow, i am starting to defend Linux

[JAZZCNC]

No you miss understand. Mach is only limited and prone to crashing when using the Parallel port at high Khz ratings.
The Parallel port and it's driver are the problem when using high Khz and struggle above 60Khz.

Motion control cards free Mach of these problems and it's Now the job of the Hardware on the Motion control card to generate the pulses NOT Mach3.
Mach3 just passes the trajectory Data to the Motion control which Buffers it in memory then generates the pulses needed, which it can do at very high pulse rates.

The main differance between Mach3 and Linux Cnc is that Mach3 and it's Motion control cards have to buffer information and Linux works in real time.
Can't speak for Linux has I don't use it but I can for MAch3 and it's motion control cards and trust me it's not hype. They do exactly what they say and generate very clean high quality pulses very fast.!!. . . . . If they didn't there would be lots of very very unhappy people who are now daily running industrial machines for there living which have been converted to Mach and using high count Servo's with encoders.!!

Also not sure what your on about regards Bob and Servo drives.? . . . . Bob does nothing more than distribute signals and servo drive does nothing but receive signals. Again I think you confused here.??