[slimneill]

Hellos all, I’ve finally decided to start my own CNC build. I was able to tear apart an old Thermwood CNC machine at work which had been sitting in disrepair for years, they finally decided scrap it and let me take as much as I could off of it. Linear slides rack & pinion, ball screws motors, wiring, vacuum tables, 3/4" machined plate basically everything I could load in my truck by myself! I basically took everything that could come apart off minus the base and gantry. Now that I have all these parts at home, I have mo garage left to work in, but that’s a welcome issue with the amount of money I’ll save and the caliber of the components I was fortunate to be able to salvage.

To give any idea of the size of this old monster, the X-drive is Rack& Pinion 18ft+ long, Y-Drive was 10ft wide with large heavy duty ball screw, Z-drive was 12". I'll post more info when I'm able to track down the actual dimension of the various parts. This was a HEAVY DUTY machine used to machine aluminum parts for aircraft component assemblies.

I think I have most of the design laid out in my head, some on paper and soon to be assembled in solidworks before beginning the fabrication and machining process. But the part that I am not very familiar with and want to get straightened out before sinking any money or time into this project is the electronic portion.

The old Thermwood machine had 4 LARGE Heavy Duty Baldor DC Servo Motors. All 4 are identical motors, originally 2 on the X-Drive (one on each Rack), 1 for the Y and 1 for the Z. I’ve compared the specs off of these motors to the current line from Baldor DC Servo motors and they appear to be equivalent to the largest motors DC Servo motors that Baldor current sells. I just hope I can manage to power them and use them now!

My plan is to use all 4 motors again, but for X, Y, Z and a rotary for machining. All 3 axis will use ball screws to drive the large gantry around the table. Table size will be somewhere in the 5x10 range. I have a heavy duty 5 x 10 vertical form block originally machined square to produce parts for large aircraft, it’s made of 6x6 square tube that my employer was also throwing away. I’ll be re-orienting it to produce a nice large flat solid table.

I'd like to use these monster motors again because as you will find out when I track down the info on the size of the ball screws and linear rails etc, everything on this machine is over engineered and heavy duty industrial grade and overkill for my purpose.I’m afraid that the DIY servo kits that I have found online will not be powerful enough to handle the sheer size of the machine I plan to fabricate. The plan is to create a machine that will slice through wood like butter and be able to machine metal with super tight tolerances, possibly with plasma capabilities down the road, but one thing at a time.

I’ve studied and based my design on MadVac’s CNC, I hope to have the same tolerances he achieved when done. I have a mill, lathe to help with metal machining fabrication.

Sorry for the rambling, On to the first question in the planning process. The electronics! I’m a mechanical engineer, electric circuits was never my favorite subject.

These are the specs that I was able to read off of the Baldor Motors:

Baldor 5D42-30-A19-A40
Max Current: 15.7A continuous
Max Current: 73A peak
Torque Stall: 50 lb-in continuous
Max Speed: 3000RPM
Max V: 180DC
Continuous V: 42 V/RPM
Tach: 9.6 V/kRPM
The encoder is a 16 pin design, the motors were made in the early 90’s.

1. Where do I find a power supply large enough to power these motors? Granted only two or three of the motors would be in continual motion at any given time.

The servo kits I’ve been seeing online have servo motors with peak torques around 850 OZ-IN (51 pound- inch) (PEAK) which (if I assume these less then adequately documented , no-name chinese motors are to the same standard and build quality as a Baldor equivalent motors equates to a continuous Torque Stall of 17 lb-in, i.e. 1/3 the torque that these Free servo motors are capable of producing.) The peak toque of these motors is somewhere in the range of 4000 OZ-IN!!!


2. The way I understand Servo’s to work is that I will need a power supply, and a controller (typically Gecko seems to be popular) Is this correct? Am I missing something? Will a G320X Digital Servo Drive work with these motors? I know that Baldor has a dc servo control unit, but I’m afraid to ask what the price will be!

Thanks for any advice or wisdom you can share with me here! I know there are a lot of great people on the forum willing to help others along the way. If there is a more suitable place I should be posting this type of info, please don’t hesitate to move it.

-D

[Al_The_Man]

You should have taken the control enclosure, you would have had the power supply and the drives, unless they were too old a vintage, the problem you may have is getting the size of drives that are step & direction, unless you go with kflop/kanalog and use AMC non-intelligent drives in the torque mode.
Also if you intend using the two rack system for the long axis as original, you need a system that will gear one motor off the others encoder, that Kflop should be able to do.
Al.

[slimneill]

Thanks Al,

I still can get the enclosure but there is no guarantee it's going to work, they have been scavenging parts off of it over the years. We had Thermwood come in and check it out and they said it was pre-Siemens controller, so I take it to be an old vintage.

Is kflop/kanalog and torque mode as accurate as the current standard used in most new servo models? I'll have to google that to find out what it is.

Would retrofitting the motors by mounting an optical encoder do the trick? Or does that cause even more issues?

I was thinking US Digital | Products » E5 Optical Kit Encoder

[Al_The_Man]

Don't the present motors have encoders? or what was the feedback method?
If you can get the enclosure it will save you a whole bunch of $$$.
The power supply for sure hopefully and if the drives are there they may be usable.
Dynomotion has a forum here as well as a Yahoo group.
If you intend using the two separate racks as they are you need to use a system that has the gearing feature which Mach etc are not capable of.
Al.

[slimneill]

I plan to use the 10' ball screw as my X-drive, the rack & pinon is only to keep the gantry square, not to drive.

[ger21]

I believe those servos are much too large for Gecko drives. Your cheapest option would probably be Viper 200 or 250 drives.
Viper Servo Motor control products

You might need different encoders as well.

[Al_The_Man]

BTW, if going with other modern drives, the motor tach's will not be required.
Al.

[slimneill]

Is there a good thread that breaks down all the different parts of a DC servo system and the improvements overtime and what it all entails?

I assume that if I move to a modern drive, I will need to update the encoder on the motor right? If i go with a modern drive that doesn't require the tach I'm going to have to change something on the motor to make it work right?

[Al_The_Man]

The DC servo system is fairly straight forward, some of it depends on what or who's controller you go with, this includes the type of drive?
To eliminate the tach, just take the brushes out and remove the two tach wires.
You also need to find out what encoder and resolution it has on it now, or if alternatively if they used resolvers?
Al.

[slimneill]

Did some further digging into the cabinet, under the mess of wires appears to be 4 power supplies, connected to 4 boards. The only identification I'm able to read is that the power supply is:

Westamp
model # a7214-17f-303
serial # b1058
another number in brackets (111389)

I tried goggling Westamp with little success, has anyone heard of this brand or where i would go or who I would ask to find out information about these components?

Thanks

[Al_The_Man]

The Westamp's were popular drives for DC servo's in early CNC, but they suffer with problems over time.
If you go with these you will need a controller that uses analogue control, so it will depend on what yopu go with, with Mach on its own, you need step/dir drives.
You should be able to use the P.S. though.
Al.

[slimneill]

I believe those servos are much too large for Gecko drives. Your cheapest option would probably be Viper 200 or 250 drives.
Viper Servo Motor control products

You might need different encoders as well.

I've looked at the Viper drive system and comparing it the the KFlop/Kanalog and i'm not sure if i'm comparing similar things or if they are apples and oranges apart.

Does the viper drive incorporate everything all in one unit? I.E. is this how the system works?

Viper based system for my application:
1 Power supply + 4 viper drives + 4 DC servo motors? (swap out tach encoders to use optical encoders?)

Kflop/Kanalog based system:
1 Power supply + Kflop + Kanalog + 4 amplifiers + 4 motors?

Am i on the right track or missing the boat?

thanks,
-D

[TomKerekes]

Hi D,

Those are two different approaches.

As you describe: with our KFLOP+Kanalog Controller analog drives and servo motors (+encoders) can be used. Our KFLOP Controller would close the position feedback loop.

I believe the Viper drive is an amplifier that can accept Step/Dir and close the position loop itself. In this case you could use a controller that is just a Step/Dir generator (like our KFLOP) or just a parallel port without an externel controller.

I think Henrik did a great job of describing the approaches here:


Regards

[slimneill]

Thanks Tom,

I've been able to piece it together by reading a lot of posts, but i wasn't confident that I had it straight, that link definitely helps sort it all out.

They should make that Henrik link a Sticky on the top of the forum, it would eliminate a lot of Newbie Posts!!!

Thanks!!!

[slimneill]

Been a while, been very busy......

[slimneill]

Any help would be greatly appreciated.

I am trying to determine what Encoder size to use to maximize accuracy without slowing down my motor. And after reading a lot of posts I still have a few questions which I can't seem to get straight answers for:


1. The Viper 200F Driver has a step rate of 300Khz. And a Max encoder frequency of 625Khz.
Isn't your maximum speed limited by the lower of either the input and output count rates? In this case the 300Khz output?
2. When using quadrature, is the 4x multiplier a result of counting 4 times as many points on the encoder wheel, or is it done in the driver after receiving the single count form the wheel? I.E. will quadrature multiply a 1000 line encoder and make it appear to be a 4000 line encoder (not in quadrature) and reduce the max encoder/motor speed by a factor of 4?
3. I also read that Mach3 has a max 100Khz output, and that parallel port is somewhere in the neighbourhood of 40Khz? Will something else be the bottle neck in the chain and reduce my motor speeds if I go for a higher line count encoder?
Am I reading too much into these different factors in determining the corrector line count for an encoder?

My aim is to have machine accuracy to 0.5/1000". I read that you want to have your encoder accuracy to be 5x more precise then your machine accuracy. I.E. 0.1/1000"
The encoder will be mounted directly to the servo motor shaft, there is a 5:1 pulley gear reduction to the ball screw which was a pitch of 1.847 inch/rev. The motor has a max RPM of 3000RPM.
The fastest speed i would want this machine to go would be 600ipm, doing the math that means the motor will only ever reach a max speed of 1624 RPM.

1000 Cycle/Rev Encoder in Quadrature
1000 x 4 =4000 pulse/rev (assuming 4x as many pulses, which could be the wrong assumption)
4000 pulse/rev x 1624 rpm / 60 sec/min = 108,266 pulse/sec
=108.26 KHz input to the driver

2000 Cycle/Rev Encoder in Quadrature
2000 x 4 =8000 pulse/rev (assuming 4x as many pulses, which could be the wrong assumption)
8000 pulse/rev x 1624 rpm / 60 sec/min = 216,533 pulse/sec
=216 KHz input to the driver

2500 Cycle/Rev Encoder in Quadrature
2500 x 4 =10,000 pulse/rev (assuming 4x as many pulses, which could be the wrong assumption)
10,000 pulse/rev x 1624 rpm / 60 sec/min = 270,666 pulse/sec
=270 KHz input to the driver

I was looking at the E6 encoder from US-Digital and they have different wheel sizes for different applications, US Digital | Products » E6 Optical Kit Encoder has anyone used these encoders? Are they a good fit? Do they work well?

Encoder Resolution
Ball Screw Pitch = 1.847 inch/rev with 5:1 reduction between motor and ball screw: 1.847 / 5 = 0.3694"
1 Rev or motor = 0.3694" of ball screw travel

1000 Cycle/Rev Encoder in Quadrature
1000 x 4 =4000 pulse/rev (assumes 4x pulses are physical and not created electronically in the processor)
0.3694"/4000 pulse = 0.0000923 inch/pulse

2000 Cycle/Rev Encoder in Quadrature
2000 x 4 =8000 pulse/rev (assumes 4x pulses are physical and not created electronically in the processor)
0.3694"/8000 pulse = 0.000046175 inch/pulse

2500 Cycle/Rev Encoder in Quadrature
2500 x 4 =10000 pulse/rev (assumes 4x pulses are physical and not created electronically in the processor)
0.3694"/10000 pulse = 0.00003694 inch/pulse

Does my Math make sense? Am i off by a factor of 4?
Am I splitting hairs, does it not make that much of a difference?
Thanks for all of your expert help, it is GREATLY appreciated.

[slimneill]

Status Update:

We went with:
-Viper 200 DC Servo Motor Control
-KFlop Motion Control Board from Dynomotion
-2000 CPR E6 Optical Encoder from US Digital
-Baldor Motor (taken out of original CNC machine)
Baldor 5D42-30-A19-A40
Max Current: 15.7A continuous
Max Current: 73A peak
Torque Stall: 50 lb-in continuous
Max Speed: 3000RPM
Max V: 180DC
-Power Supply: 165V DC

The good news: It all works! I can drive the motor with mach3 or KmotionCNC.

The biggest hurdle in this entire process was the electronics, now that I know I have motion. Actually fabricating the rest of the machine will be a little less stressful now that the big question mark is gone.

I’ve learnt a lot so far and there is a lot more I need to learn to complete the electronic piece of the CNC puzzle.

The way I have it set up currently is that the Viper accepts the Encoder signal and closes the position loop itself. The KFlop currently only sends the step and direction from the PC to the vipers which power the Baldor Motors. I know the Kflop can take the encoder signal and close the loop that way too but I’m not sure if the Viper can function strictly as an amplifier. I’ll focus more on this option at a later date.

Everything is currently set up on the work bench. I have realized that these motors are big powerful units, which is making the tuning sequence with the Viper very trial and error and difficult to optimize. Have you ever try to upset the shaft of a 50lb-in servo motor working against you? It barely budges while trying with all my might.

According do the Viper manual you upset the motor and then adjust the parameters so that it returns to steady state as quickly and smoothly as possible with out continuous oscillation. The problem is that, we can only budge it 3 to 5 degrees and then is returns to it’s park position right away with no visible oscillation, only a slight hum and clicking sound. This is telling me that the servo is constantly trying to fight it’s way back to the true home position even after being a rest for a min or so. We found that applying a hand to the shaft to dampen the shafts motion helps it settle quicker but it doesn’t stay completely still for long. Taking a stethoscope to the side of the motor, you can hear it clicking back and forth as it tries to find rest and once it does it will start clicking and self adjusting randomly many times over the course of a minute, for as long as it is turned on.

I’ve never dealt with servo motors before this point, is this normal? Will there always be some live adjustment going on or should it come to a complete rest and not have to self adjust if set up currently? Or is 2000 line encoder to accurate, and the play in the motor/encoder wheel bearing causing the motor to struggle to stay stationary? The motor temperature gradually increased from 80F to 130F just sitting stationary on the bench as it clicks back and forth for about half an hour. I’m not sure how hot motors should run sitting stationary but it’s starting to concern me. The current going to the motor was less then 0.3 amps the entire time. Is this something that requires additional tuning or is it a problem I caused with a high resolution encoder? Will going to a 1000 line encoder help solve this issue?

The current Viper Servo settings are:
Kp=60
Kd=120
I=1
Loop=5

Started off with higher values but found the issue got better as the values decreased.

Thanks for the help!!!

[Al_The_Man]

The desired action of servo's is they when move to a position is to achieve it without any oscillation or dither, and position within one to a very few encoder counts (ideally zero), any force applied to the shaft should be met with resistance and deter any movement of the shaft up to the continuous torque value of the motor.
Al.

[slimneill]

The desired action of servo's is they when move to a position is to achieve it without any oscillation or dither, and position within one to a very few encoder counts (ideally zero), any force applied to the shaft should be met with resistance and deter any movement of the shaft up to the continuous torque value of the motor.
Al.

I think that the drive is perfomring as it should, it resists any force applied to the shaft and keeps it in position. When I say that the drive is contstantly trying to fight itself to bring it back to position, i should clarify further, there is no noticable motion. I can only tell this is occuring becuase A) i can put my hand on the shaft and feel it vibrating and B) i can hear the motor silently humming/clicking when it is doing this if i put my ear next to it in a completely silent shop. I can't see it being noticable once assembled to the actual machine. My concern comes out of being new to this and not wanting to burn out an expensive motor by improperly tuning it, and assuming this is standard practice when it may not be.

I should rephrase the question. Is constant motor correction (very very small amounts) to be expected from servo motors in a CNC machine while stationary? Or am i causing additional wear & tear to the motor because of the high resolution endcoder constantly correcting itself?

[Al_The_Man]

Depending on the drive, on my torque mode drives, IF there is no perceptible pressure or opposition of the motor it is completely silent and vibration free when it servos to around zero to 1 pulse error.
If I have the gain too high it will tend to cause oscillation or vibration.
Al.