[5th-axis]

Hello,
We have been looking at replacing our DC servos with the AC servos from Automation Direct. The prices I saw listed include the drive, servo and cabling for around $1500. This is way cheaper than a new servo from Fadal which I think comes in at around $2200. Has anyone here in the forums done this swap? Our machine is a 1993 6030 with a Numatix retrofit. The control is awesome but I think the machine could use a little updating to keep it going.

Thanks guys,

[mactec54]

5th-axis

You need to look at the spec very carefully, check what you have in the way of encoders, the ones you are looking at are only 2500 count, way to low for a good CNC machine

[5th-axis]

Mactec54,
Thanks for the reply. My machine is a glass scale machine so the encoders would not be used.
Thanks again

[JR1050]

I would think a 1993 fadal would still have resolvers,especially if it has dc glentek motors.Fadal used the resolver as an absolute pos device(hence no limit switches).If you still have resolvers,you will have to put them on your new motors,If you have encoders,Im sure there is a parameter in your control to set the encoder count. 2500ppr is fine,in quadature,its gonna be 10,000 ppr!Go for it and let us all know how it works!!

[Neal]

1993 machine did indeed have DC servo motors. The casting machining may pose an issue to install AC servos. Also the Eproms on the 1010 controller cards will NOT accept feedback from AC servos. They will need to be changed. The AC encoder count MUST be 8192 for the software to function properly.
Bottom line is that this is NOT a simple put a different motor on.

Neal

[5th-axis]

I looked at the mounts for the current servos and I believe I can make new mounts for the AC servos with no casting machining required. Please keep in mind that this machine is a Numatix retrofit and the company who did the retrofit initially brought these servos to my attention. He also mentioned that I could do each axis servo seperately as budget and job constraints allowed. When we did the retrofit the resolvers were unhooked. The limits are set in software in the control. I absolutely love the Numatix control and the tunability it has.

[ishaham]

Dear Numatix user

There is no problem to replace the motor with the Automationdirect system, you can email me you email address and I will send you a coplete instruction how to replace it and how to set the control to accept the new system and how to tune the new system

Igal

[ishaham]

1993 machine did indeed have DC servo motors. The casting machining may pose an issue to install AC servos. Also the Eproms on the 1010 controller cards will NOT accept feedback from AC servos. They will need to be changed. The AC encoder count MUST be 8192 for the software to function properly.
Bottom line is that this is NOT a simple put a different motor on.

Neal

Hi Neal

I would be very happy to educate you about the Numatix Control, or you can look at our web site www.numatix.com, we do have a special kit for Fadal that runs ~$20,000 installed, we guaranty increase of 100% of productivity or your money back, we also have a kit that we can install without removing the original Fadal for customers to try for 10 days, after this they can decide, if they dont like it, it takes about 30 min to restore to original system no strings attached

[ishaham]

5th-axis

You need to look at the spec very carefully, check what you have in the way of encoders, the ones you are looking at are only 2500 count, way to low for a good CNC machine

2500 counts on a 6.35mm pitch will give you .000025" resolution which is good enough for fadal or any other mills

[ishaham]

I agree with the resolution, but the resolver is used just for position and reference, when we retrofit with the numatix we remove the resolver and install encoders as part of the installation

[mactec54]

ishaham
Quote
2500 counts on a 6.35mm pitch will give you .000025" resolution which is good enough for fadal or any other mills

If your Numatix control is that good, why would you say that 2500 count encoder was good enough for his machine, have you seen how a large count encoder 17bit / 20bit will make a machine like this run, or can your control not handle a large count encoder

As I have said before in other posts it has nothing to do with resolution, when you have a large count encoder,the drives take care of the encoder, it's how well the machine will run with a large count encoder, supper smooth & fast, You will not get this with a 2500 count encoder

2500 count is from 20 years ago, modern machines have nothing less than 13bit most have 17bit & higher

[ishaham]

Dear Mactec54,

First, the servo is responding to the control’s command. The control defines the motion parameters, not the servo. In most systems, such as Fanuc and Siemens, the motion servo loop in the control has a ~2 msec response time. With this response time at 100 ipm, the control will see the position after ~.003” of motion has already been executed. This lag causes the control to have to “catch up” to the current position of the servo.
The Numatix has a .2 msec servo loop, which is approximately 10 times better than most controls out in the market. With this .2 msec servo loop at 100 ipm, there will be an error of a maximum of .0003” before the control will make the correction.
The smoothness of motion has very little to do with the encoder counts. The smoothness of motion DOES have to do with the ability of the control to process the data fast enough to push the servo to its maximum potential.
In the aspect of processing speed, the Numatix is superior to most controls. The average processing time of most controls ranges from 600 to 2000 blocks per sec, where the Numatix does over 50,000 blocks per second. The processing speed is the reason why when you program at 100 ipm on most controls, you can only achieve an actual average feed of 30-50 ipm. The control has to slow down whenever it gets to a dense amount of data points in the program, just to allow the control to process the data and not lose control of the motion. Of course, the “look ahead” is also a factor. That is why Fanuc, Siemens and other controls invented the NURB feature which reduces the amount of points to “help” the control digest the data, and by doing this increasing speed, but at the cost of degrading accuracy and surface finish.
With that being said, I believe 2500 PPR quadrature that makes it 10000 is good enough for any control.

[Andre' B]

You are both right and both wrong.

For smooth precise high speed motion loop time is where it is at.
For smooth precise slow speed motion (wire and sinker EDMS, and the slow axis when cutting a taper on a mill or lathe) unless you have some other kind of velolcity feed back you need all the encoder counts you can get or you end up with steps in your part.

[mactec54]

Andre' B

ishaham You need to do some catch up, Softservo systems is way better than your control, & costs a lot less, they can take up to 20 bit encoders, & it makes for a nice machine

This is were it's at, These guys are using 16 million count encoders for there high speed control, with nano smoothing

New designs are often produced on 3D solid-model CAD systems using Non-Uniform Rational B-Splines (NURBS) to describe complex surfaces. Some high-end CNCs have advanced options to machine NURBS surfaces directly with specially formatted part programs. Creating NURBS part programs requires special software, and it can often be a time consuming and expensive process, which can be frustrating during part program validation. Many current CAM systems still output short line segments, even though they support NURBS internally.

There are also thousands of pre-existing part programs that use a short line segment approximation of a NURBS surface. These part programs have been validated and certified, and reverse engineering them back to the CAD system geometry is not practical.

The high-speed machining of short line segment part programs is difficult. The rapid and abrupt direction changes described by the short line segments create overshoots and vibration in the machine mechanics that limits processing speed in order to maintain accuracy. However, nano smoothing provides a powerful alternative when the CNC or the CAM system does not fully support native NURBS.

Nano Smoothing
Nano smoothing essentially reverse-engineers the NURBS contour on the fly by developing a “best-fit” curve using the part program line segments. Nano smoothing uses advanced techniques to create a curve as close as possible to the original NURBS curve.

When CAM systems generate a line segment part program from a CAD NURBS curve, the part programmer typically specifies a tolerance value. When a tight tolerance is used, the post will typically create more line segments. More line segments means a better path profile, but it may also overload the CNC’s processing power. Because of this approach, most of the points generated by the CAM system lie on the tolerance curve rather than on the original NURBS curve. This bias error is compounded by rounding errors used to match the minimum CNC part program increment. Regenerating a NURBS curve by only considering the points in the part program would create a distorted profile.

Nano smoothing overcomes this problem by inserting additional points between the points specified by the part program. Many of the inserted points are much closer to the CAD system’s original NURBS curve than the points specified in the short line segment part program. The nano smoothing function uses the additional insertion points to approximate potential NURBS line segments. Finally, the inserted points can be corrected so that all the points fit a continuous NURBS line. Nano smoothing is performed using nano meter precision calculations (0.000001mm/ 0.00000003937”) so that an extremely smooth contour can be modeled, producing noticeable results in surface quality.

Programming Nano Smoothing
For many companies, programming and editing NURBS part programs is complex, time consuming and expensive. Programming nano smoothing in contrast is very simple. A single part program block activates nano smoothing mode, and a second block deactivates it. All the part program blocks between the activation and deactivation blocks are standard line segments.

The G5.1 Q3 activates nano smoothing for each axis specified in block. Axes are specified with their address letter (X/Y/Z/A/B) and a zero value. No motion is commanded by the nano smoothing activation block. Activating nano smoothing also activates high-speed machining contour control, providing automatic velocity control to further smooth impacts on the machine’s mechanical systems. The G5.1 Q0 deactivates nano smoothing and high-speed machining contour control.

With the smooth NURBS curve interpolation created from the straight line segments and with the automatic velocity control of high speed machining contour control, the machine mechanics receive smooth and precise motion commands and overall processing speed can be increased significantly.

Summary
Complete NURBS support is still the technique of choice for those companies with the resources to invest in state-of-the-art CAD, CAM and machine tools, and do not have a significant investment in existing part programs. Nano smoothing provides NURBS-type performance for everyone else.

Advanced algorithms reverse engineer the NURBS curves described by the line segment part program, using nano meter precision. The nano smoothing commands are simple and can be quickly added to existing part programs, either by modifying the CAM post processor or by keying in the activation and deactivation commands directly on the CNC. With nano smoothing, even smaller contract manufacturers with limited resources can achieve the benefits of high-speed machining with smooth NURBS curves.

[ishaham]

Mactec

First, Numatix can handle 4 milion line encoders, (not 1 milion) so this is not the issue,
Second, do you have an idea why all these controls are using NURB? why not be as close as you can to the real surface and not creating short cut with the NURB, and the answer is, they can NOT digest the points so they reduce them with the NURB system, but if you have a control that can process the data fast enough and keep the machine running at the requiered feed, why do you need NURB?

[mactec54]

ishaham

You were the one that brought up the Nurb thing, I just showed you what fanuc is doing about it & it's fanuc thats using the 1millon count encoders,on there HSM controls, it may be more, I will have to look, I did not say the 1 millon count encoder, was anything to do with your control

Are you having any machines at the IMTS show that have your control on

[mactec54]

Well I was a little off in the other post with the fanuc encoder count it should of been 16 million counts per rev

http://www.fanuc.co.jp/en/product/ca...A(E)_v06_s.pdf

[5th-axis]

We do alot of surfacing on our Numatix equipped Fadal and the parts come out of the machine faster with a high degree of accuracy. The machine is also much smoother. Tool life has increased as a result as well. As a matter of fact, the spindle is now the limiting factor on our machine. Ishaham, you were just in our shop in North Carolina a few weeks ago. This should give you an idea of who I am. I will admit that when we first looked at the Numatix I was a doubter. But after running it for 2+ years with no issues, faster cycle times, and increased tool life I am now a believer. The Numatix control really woke up a tired machine. I would post up pics of the type of work we do, buts it is proprietary and I would get in big trouble. Just take my word for it that most parts are pretty complex.

[mactec54]

5th-axis

It's not hard to wake up an old control like on your Fadal, even the $170 Mach software would of made your machine run a lot better,the Numatix I have had & used as well, but for the money??I would go with other controls that work even better than what you have for a lot less money