- FineLine Saturn 4x2 with ClearPath, UC300-ETH / MB2

[DDgitfiddle]

Robb (and others who may be interested),

My system Bill of Materials is attached. The biggest thing that is not on the list is hardware. There wasn't much needed, but the DIN rail, 75V power supply, VFD, EMI filter, and breakout board all have to be screwed to the back plane of the enclosure.

I'd like to put a big broad caveat on the bill of materials. I make no guarantees about the suitability of any of the components I have specified. Safety devices like fuses, relays, shutoff switches, etc. should be reviewed before using these parts in a new design to confirm that they are sized correctly. It is the builder's responsibility to use components that are appropriate. Wire gauges also need to be determined based on the design.

With that out of the way, here are a few more details. I have added some comments to the spreadsheet that may help adjust the design to other people's needs. There are a few parts that I used that were a little less than ideal, so I recommend using something else. I purchased a number of industrial style cable with connectors already attached on at least one end. The location of the control enclosure relative to the rest of the machine has an impact on the length of these cables. My system is a 4x2, so a 4x4 would need longer cables in some cases. I located my control enclosure to minimize the length of the cables that run through the cable chain. Shorter cables are generally better because they pick up less noise.

There is something else to consider on the motors. Teknic sells another version of their motors called the SCSK series. They come in the same sizes as the SDSK series but also offer the option of full software control. This would allow for fully custom closed-loop servo control. These can also be programmed to accept step and direction inputs like the SDSK series. This is a totally different control concept, but might be interesting for someone who really wants to maximize performance.

I'm not entirely convinced that the FineLine design is "better" than the CNCRP system. I am impressed with how stiff the FineLine system is. I think the main frame in particular is probably alot stiffer. The gantry's are pretty equivalent, though. One thing I am finding is that the high frame sides on the FineLine system prevent real access to the work area (without getting covered in grease from the rack). Like you, I do like the ClearPath motors better than a standard stepper solution, and I definitely prefer the safety architecture that I implemented over the CNCRP controller.

I will try to update the schematics over the next week or so. I had to make a few changes over my original plan in order to get the safety architecture to work properly and to get the breakout board to talk to the Hitachi VFD.

-Robert

[rcmcdee]

Many Many thanks for this Robert. I look forward to the shopping and getting this project in motion. It will be interesting to stay with this thread and update my experiences with the comparable equipment.

Robb

[DDgitfiddle]

Many Many thanks for this Robert. I look forward to the shopping and getting this project in motion. It will be interesting to stay with this thread and update my experiences with the comparable equipment.

Robb

Good luck! Definitely make the effort to try to understand the circuit diagrams when I update them. It will help alot when something goes wrong. It has been a couple of months since I first wired everything up and I have already forgotten why I did some things the way I did. I had to re-trace my steps when I recently had to make a change, and I really wished that my schematics had been up to date. At some point in the future, having the schematics will probably help me diagnose some other problem, so getting them up to date will be helpful.

Let me know if you have any questions.

-Robert

[DDgitfiddle]

I've updated my schematics and attached them here. I can't claim these are professional quality or even 100% complete, but I think they are adequate to build from. What is missing is some of the terminal blocks where I ganged up signals. I will post a few photos of my panel with labels to point out where all of these other blocks reside. Its getting late, so I will try to post an updated "theory of operation" for the safety circuit later this week. It has changed a little since some of my earlier posts, but I think it is working pretty well with a good level of safety.

-Robert

[DDgitfiddle]

These are the panel photos. Its a little tough to decipher without labels, but there is at least a little logic to everything. Terminal blocks are colored according to voltages or specific signal type. Gray is for general signals. AC power is isolated to the right side and bottom of the enclosure. Clean DC power and signals are isolated to the top and left. The spindle cable enters right beside the VFD to minimize length inside the box. External signal wires enter as close as possible to the motion controller to minimize wire length. Power and E-stop cables enter in through the bottom. Wires are bundled for convenience, but AC wires are not generally bundled with signal wires.

The main fan blows across the bottom of the box to hit the VFD and motor power supply. Both of these have their own fans, so the main fan is mostly there to pull in enough air to get the heat out of the box.

-Robert

[FineLineAuto]

Very nice layout and documentation. I am sure others going down this route will benefit from it.


Sent from my iPhone using Tapatalk

[rcmcdee]

Beautiful work and documentation. Many thanks for your help. I will keep you posted as our build progresses.

[difalkner]

That's a really nice looking enclosure, Robert! Good job.

David

[DDgitfiddle]

Thanks for the complements, gentlemen!

So with the schematics and panel layouts above as reference, here is the basic theory of operation on the system.

Safety Architecture

The system uses a safety relay as the primary safety circuit and the E-stop circuit on the MB2 as a secondary safety monitor. The safety relay controls all of the motive power in the system. It controls the coil of a contactor which switches AC power to the 75VDC power supply for the motors. It also controls two signals for the Gate Suppress functionality built into the Hitachi VFD. The gate suppress feature is nice because it truly kills power to the spindle without killing power to the VFD. The safety relay has two inputs: E-stop and Reset. The E-stop circuit contains a latching e-stop switch on a pendant in series with the Charge Pump circuit from the MB2. The reset is a physical switch on my "pendant" right beside the E-stop. The reset will only function when the E-stop circuit is closed. The E-stop circuit will only close when the Charge Pump is active and the E-stop switch is closed. What this all means is that the motor power supply and the spindle amplifier will not power up until the motion controller is running AND I hit the Reset switch on the pendant. It is impossible (without a very weird hardware failure) for the motors to move until I am ready to have them move. This was critical for me as I had read that some motion controllers can set outputs to strange states when they first start up. The safety relay helps avoid that problem.

The secondary safety circuit on the MB2 is used to monitor errors from the motors and spindle. The motors each output a HLFB signal that closes when the motor is enabled. The VFD has a NO relay output that opens when there is an error or if the power drops to the VFD. All of these switches are arranged in series so if one motor or spindle has an error, the MB2 sees an E-stop. This puts the UCCNC software into Reset (soft reset) which in turn drops the enable signals to the motors and the spindle. I had a situation where my spindle had an over-current error (it was spinning the wrong direction), and the motors kept moving. Fortunately it was not a deep cut, but its hard to drive a stationary bit through wood. Similarly, if a motor ever has an error, I would much rather have the spindle shut down than have it keep running. The MB2 also monitors an auxiliary contact from the safety relay so it knows when the safety relay has tripped. This ensures that UCCNC will go into a "soft reset" to stop a program whenever a hard E-stop is pressed.

The schematic also shows that the monitoring circuit for the safety relay is connected to an auxiliary contact on the contactor. I had also tried to connect this to a signal on the VFD thinking that it would trip the safety relay if there was an error, but that didn't work. I am now unclear what this monitoring circuit really does, other than check that contacts actually closed before allowing the safety relay to latch in the operational position.


Operational Architecture

To operate, the ClearPath Motors need 4 things: 1) DC Power, 2) a 24VDC Enable Signal, 3) Pulse train of steps, and 4) A Direction signal. DC Power is pretty straight-forward and shown in the schematics.

The enable signals are routed through 4 separate outputs on the motion controller, but the 24V power also runs through the contactor to avoid being able to have the motors have an enable signal before they are powered up. Probably not strictly necessary since a lack of HLFB signal from the motors would also prevent the enable signal. The 24V power for the enable signal comes from the MB2, and is distributed via terminal blocks to the 4 motors.

The Step and Direction signals for the motor are handled via standard 5V connection. The ClearPath motors will not accept a differential signal, so I had to live with the 5V pulse signal. So far it seems to be working pretty well. The 5V power comes directly from 2 terminals on the MB2 and is distributed to the 4 Step and 4 Direction signals via terminal blocks.

Originally, I had hoped to use a leftover Step and Direction motor signal for the spindle speed control, but the signal was too noisy and resulted in poor speed control. 1Jumper10 had a similar issue and had contacted the folks at CNCRoom for advice. The result is that a standard output on the MB2 is used to create the pulse train needed to control the spindle speed, and a resistor is added between the EA and P24 pins on the VFD. (See the Safety Schematic) This provides a 24V signal which is robust to noise. Spindle speed has been rock-solid in this configuration. This connection scheme does require connecting the VFD "L" terminal to MB2 0VDC and the PLC terminal to MB2 24VDC. Technically these are two separate 24V power supplies, but the voltages should be pretty close and shouldn't cause any trouble.

Limit switches (prox sensors) are connected as recommended in the MB2 manual. This is a straight connection of wires to 0V on the MB2 and the specific input selected. The sensors also need 24VDC power, which is provided from the clean 24V power supply via terminal blocks.

The other items requiring power are the main enclosure fan and the external fan for the 75V motor power supply. The main fain is powered by 220VAC and is connected directly to 220VAC distribution so it powers on when the main switch is turned on. The external fan for the 75V power supply is powered by 24VDC and is connected to the clean 24VDC power distribution, so it also turns on with the main switch.


UCCNC Configuration
There is not much unusual here. The A-axis is slaved to the Y-axis. I had to reverse a few signals to make things work right. I had to set a couple of input triggers, but those are escaping my memory at the moment...I will update this when I have a minute to check it. The Charge Pump signal is set to always on. A little math is needed for calculating the number of steps per unit of motion, and for calculating the right spindle speeds. I have enabled soft limits because the system goes into reset every time a limit switch is triggered (appropriate). I find that to be a pain, so I like the soft limits.

I think these are all the major points. I'm pretty pleased with how everything functions. I'm still working out maximum speed and acceleration. In theory 800 in/min should be feasible, but even 200 in/min is looking pretty fast right now. I still need to work with the various settings for acceleration and smoothing to see what I can do with that. This is definitely not the cheapest way to build a system, but I think there are alot of advantages that will pay dividends in reliability and performance.

As always, if there are any questions or comments, just let me know.

-Robert

[DDgitfiddle]

Finally got around to getting the spoil board finished. I made the front section removable for future vertical work and added grooves for T-tracks. I figured I would start with that because it sort of suits the way I think. We'll see how it goes. I can always change the spoil board out in the future. I'm also thinking about adding a few sets of 1/4" holes to help locate parts using pins. At least I can get the Y-axis straight that way



I need to work on the dust collection. My little 3/4 HP portable might not have enough power. It also might just need to have the bag emptied. The machine is definitely dusty now, though!

I have my first real project planned...I need to earn back some brownie points, so time to make Christmas presents!!

-Robert

[rcmcdee]

Just curious... I've seen reference to a brake on the Z axis motor for power off conditions. Why is there no mention of this in your build?

Parts are ordered per your BOM and soon I will posting our build record.

Many thanks again for you diligence.

Robb

[DDgitfiddle]

Just curious... I've seen reference to a brake on the Z axis motor for power off conditions. Why is there no mention of this in your build?

I do not currently have a brake on my Z-axis. So far, it has not been necessary. Depending on the Z-axis design, two main factors will change: 1) the friction in the mechanism and 2) the weight of the moving parts of the design. As long as the friction forces are higher than the gravity forces, there is no NEED for a brake. In my system, so far, the friction has been pretty high, and I have to push the Z-axis down pretty hard to get it to move with the power off. I am using the term "friction" a little loosely here to mean any resistance to motion.

I think the main factor here is the ball screw. There is a 5mm pitch on a 14mm diameter (I think) screw. If you theoretically unwrap the screw, you get a ramp, and in this case the ramp is pretty shallow. The result is that a downward force can't be translated into very much rotational force (torque). The friction in the bearings, seals, ball nut, and even the resistance of the motor windings all resist the rotation, so it takes alot to get the ball screw turning. A 10mm pitch would be easier to move, and would be closer to needing a brake. With lead screws (without the low friction ball nut), you can get "self-locking" behavior that can never be back driven at all because of the combination of screw pitch and friction. That is a nice feature for some safety critical applications (think hospital beds or other similar things.)

The 5mm is a good choice for this machine because it increases Z-accuracy (more rotation needed per Z distance moved) and reduces the torque needed to move. The tradeoff is that it needs more motor speed to go fast. Z-axis speed isn't as critical as X and Y, though, so again, a good tradeoff in my opinion. Choosing a faster motor with lower torque could increase that speed, and it would probably still have plenty of torque. The ironic thing is that the ClearPath motor that meets these criteria is MORE expensive, not less. Oh well.

If you REALLY wanted to make sure the system was absolutely safe, I imagine you could fit a brake, but it would have to fit between the motor and the ball screw. You would need to find one the right size and it would need to somehow extend the motor shaft. The shaft on the ClearPath motor would not be long enough to reach the coupling if you add a brake.

Good luck with the build! I'm interested to see how it goes. I'll try to keep up and offer help if you run into any problems.

-Robert

[FineLineAuto]

I do not currently have a brake on my Z-axis. So far, it has not been necessary. Depending on the Z-axis design, two main factors will change: 1) the friction in the mechanism and 2) the weight of the moving parts of the design. As long as the friction forces are higher than the gravity forces, there is no NEED for a brake. In my system, so far, the friction has been pretty high, and I have to push the Z-axis down pretty hard to get it to move with the power off. I am using the term "friction" a little loosely here to mean any resistance to motion.

I think the main factor here is the ball screw. There is a 5mm pitch on a 14mm diameter (I think) screw. If you theoretically unwrap the screw, you get a ramp, and in this case the ramp is pretty shallow. The result is that a downward force can't be translated into very much rotational force (torque). The friction in the bearings, seals, ball nut, and even the resistance of the motor windings all resist the rotation, so it takes alot to get the ball screw turning. A 10mm pitch would be easier to move, and would be closer to needing a brake. With lead screws (without the low friction ball nut), you can get "self-locking" behavior that can never be back driven at all because of the combination of screw pitch and friction. That is a nice feature for some safety critical applications (think hospital beds or other similar things.)

The 5mm is a good choice for this machine because it increases Z-accuracy (more rotation needed per Z distance moved) and reduces the torque needed to move. The tradeoff is that it needs more motor speed to go fast. Z-axis speed isn't as critical as X and Y, though, so again, a good tradeoff in my opinion. Choosing a faster motor with lower torque could increase that speed, and it would probably still have plenty of torque. The ironic thing is that the ClearPath motor that meets these criteria is MORE expensive, not less. Oh well.

If you REALLY wanted to make sure the system was absolutely safe, I imagine you could fit a brake, but it would have to fit between the motor and the ball screw. You would need to find one the right size and it would need to somehow extend the motor shaft. The shaft on the ClearPath motor would not be long enough to reach the coupling if you add a brake.

Good luck with the build! I'm interested to see how it goes. I'll try to keep up and offer help if you run into any problems.

-Robert

Robert,

Great analysis. That was the thought process that we used knowing that some users would install heavy spindles on the unit. Unless your spindle is more than 75 lbs, you don’t have to worry about it.

[ger21]

If you REALLY wanted to make sure the system was absolutely safe, I imagine you could fit a brake, but it would have to fit between the motor and the ball screw.

When a brake is required, it's usually a servo motor with an integral brake.

[DDgitfiddle]

Robert,

Great analysis. That was the thought process that we used knowing that some users would install heavy spindles on the unit. Unless your spindle is more than 75 lbs, you don’t have to worry about it.

I can say first-hand that you got it right. I'm not sure I would want to see a 75lb spindle on this system...seems a little overkill.

-Robert

[DDgitfiddle]

When a brake is required, it's usually a servo motor with an integral brake.

That would definitely be the easiest way to do it. I've custom fit brakes to motors in the past and it is usually difficult to find exactly what you need.

-Robert

[difalkner]

I agree, Robert. Ours is right at 20 lbs. and it seems like a beast.

David

[DDgitfiddle]

I had a very bittersweet day yesterday. I finally had everything in place to start my first real 3D project, and the time to work on it. Everything went pretty well, and then this happened:



I was getting ready to start the final operation to split the bowl from the rest of the material, but I wanted to test the tool path in air first. I made a program change that put the safe start location way to low (more on this in a second), and it drove the spindle straight through the bottom of the bowl. I can neither confirm nor deny that there may have been some very colorful language. Just to make myself feel better, here is what the inside look like before the incident:



So as I am getting over my loss, I have a few questions about some things that happened along the way:

1) On some of the first roughing passes, I noticed that the machine was creating facets instead of smooth circles. Not a big deal for roughing, but I wanted to figure out why. The facets were about 1" long and you could see that the motion was slightly jerky. I looked at the G-Code and the distances between steps were MUCH less than 1" so I figure the problem might be in the motion controller. I adjusted the tolerance values and increased the control frequency from 100kHz to 200kHz and the motion got alot smoother and more circular. This made me a little more cautious about running programs before testing them, however. Still there were some sections of the roughing passes for the bottom that were very jerky. I found that these sections had G-code that alternated between G1 and G2 (or G3) commands (alternating every line) and they were on the last pass of a particular Z level nearest to the finished workpiece. The other passes on these levels would be all linear moves and sounded very smooth. So my question is this: Is this jerky motion related to this switch between G1 and G2 codes, or is it related to something else in my setup? Is the machine just finishing a step too quickly and stopping before it gets the next command?

2) I tried putting the control frequency at 400kHz, and found that my spindle no longer turned on. Any clue what might cause that? I am using a relay output on the MB2, and it SEEMS to be firing. I am using a standard output for frequency control, so that may be part of the problem. I haven't tried debugging this yet.

3) I used a "Scallop" tool path (from Fusion 360) for the finishing passes with a 1/2" ball end mill. This strategy circles around the part at a particular level then transitions along the surface to the next level. I used a 0.05" step between passes, and this is measured along the surface, not vertical or horizontal. I noticed that there are several ridges that line up along the Y-axis. The ridges are essentially little bumps in the same spot on each pass. To me, this implies that there is something mechanical going on...in this case in the X-axis. I suspect that there is some sort of inconsistency in the x-axis rack or bearings. Alternately, it could be something in the motors, but I would expect that to happen every motor revolution. The ridges look to be at the same pitch as the gear rack. The ridges are not that tall, and could be sanded pretty easily, I'm just curious if anyone has seen something similar.

4) I roughly know what happened to cause the "incident," but I really want to try to understand the root cause. I have been trying to "air cut" programs while I learn more about the system. To do this, I have been raising the Z-axis to a point above the work piece, trying to make sure that I clear the workpiece. The problem that I run into is that I routinely get an error message from UCCNC indicating that the program will exceed the maximum Z-axis soft limit. Each of my programs starts with G90 G0 G53 Z7.5. This command tells the controller to enter "absolute coordinate mode" then to rapid to an absolute Z height of 7.5. In general, that would put the bit above most work pieces, a safe place to be before starting the spindle. In general, the Z height gets zeroed to the top of the stock, which puts the zero at a some height above 0 in machine coordinates. UCCNC seems to check that the the maximum Z move plus this zero location in machine coordinates is lower than the maximum absolute Z. For example, when you put a fairly long bit in (say 2") and a fairly tall work piece (say 3"), and set the zero at the top of the workpiece, the Z-axis could easily be at 4" or 5" in machine coordinates. Adding 7.5" to this would exceed the 11.4" maximum on my machine. So by reducing the "Z7.5" to Z6.5, I could gain a little space and be able to run the program.

The problem is that when UCCNC runs this move, it sends the Z-axis to 6.5 in MACHINE COORDINATES. This is inconsistent with the calculation that it is running which would send the axis to 6.5" above the Z-axis zero position. The rest of the program runs in "absolute coordinate mode" relative to the axis Zero positions in X, Y, and Z, so why not this particular move? In the "incident" above, I had set the program to Z3.9, forgetting that this is what happens. The result was plunging the bit straight into the part without the spindle running. Is there there a setting that I am missing or something I should do differently in the program to prevent this? Should I change the programs to not move in Machine coordinate? (i.e. G0 Zx.xx instead of G0 G53 Zx.xx?) This would assume that the Z-axis zero was properly set ahead of time, which it would need to be to get a decent part. The UCCNC post processor in Fusion 360 had this initially and it has caused me headaches ever since.

Thanks,
Robert

[difalkner]

The bowl looks nice, Robert. Or, at least from the top - ugh! I need to try one of those.

In Fusion 360, do you have 'Smoothing' turned on (Passes tab)? I wasn't using it initially but when I started the facets went away and the file size, along with the lines of code, shrunk by about 40% and that is significant. Now I use 'smoothing' on every job.

I'm using Mach4 but when I cut in the air I just raise the bit higher above the work piece than the ultimate depth of cut and then tell Mach4 that is zero. When I run the file, even if I run it to completion, the bit won't reach the work piece. When I'm ready to run the job I lower the bit to the work piece and set the real zero. But I'm not familiar with UCCNC and how this would look in that program.

David

[ger21]

1) What are all your CV Settings set to? They can make a huge difference on how fast and smooth your machine does 3D cutting. You can probably set the maximum error a lot higher than you think, and it should then run a lot faster and smoother.

2) My guess is that the pulse width is too short for the VFD at 400Khz.
As you increase the pulse rate, the pulse length gets shorter.

Changing the kernel speed won't have any affect on speed, if you don't change your velocity and acceleration settings as well.

3) I'd need to see some pics to understand what you are talking about.

4) Z zero in machine coordinates should be the very top of the Z axis travel, and your g-code is telling it to go another 7.5" above that. When you zero the Z axis to the top of your stock, you are setting the Z axis work coordinates.
I'm not sure where that value is set in Fusion. I just did some quick testing in Fusion, and my code is G90 G53 Z0.
Personally, when I start using CAM in Fusion, I'll be editing my post to get rid of those moves.