Here's You Sign #98 - Pennsylvania USA

[smreish]

Odd....double the debounce, 1/2 the RPM achieved. So, what freq is 8000 and 140000 respectively on your VFD?

[pblackburn]

Sean,
132 Hz is 8000 and 233 Hz is 14000. Sorry I did not get to post an update last night. Spindle ran for 25 minutes until it faulted. I have also ordered a different cable made for VFD applications. The one I have is double insulated but with braid only so I ordered LAPP cable that is braided with shield and drain. I would like to get the debounce back under down closer to zero. The carrier freq was increased because of audible motor noise. Had to increase to 5kHz to smooth out motor as I mentioned in a earlier post. This will ultimately change when the new cable is installed. I was trying to find where the noise was most prominent, the supply or the motor leads. Cable coming from Allied. Have checked local vendors and they all said the best price was from Allied as they could only get in a larger quantity than I needed. I was hoping that the braid only would be sufficient but I need a good base to start the troubleshooting with now.

[KenC]

The noise is generated at the motor lead, in a poorly designed VFD, the supply lead will pass the EMI/RFI noise to the mains. Form the mains, will contaminate everything that are connected to it. but if you have you VFD power lead separated from the control panel & PC mains, you don't need the "reactor" Which I call EMI/RFI filter. this filter is no rocket science, its basically 1 or 2 coils + a few capacitors.

[pblackburn]

Ken,
I know we experience the effects of noise where I work at. All our machines in one room have woods vfds. They affect a high speed overhead door that is 20 feet from the closest press.

Understood about the EMI/RFI filter, that is my next step if these steps do not work. I do not think I will have to go that far. The problem will not allude me forever and will rear its ugly face. I just need a solid base. The drive is a Hitachi WJ200. It is mounted separate from the panel in its own enclosure that is vented. The power does come from the machine control panel. The problem here is time. At work everything is there you find and fix the problem because of stock supplies. Here at home, it is purchase, wait, wait, receive, try, troubleshoot, evaluate, order more parts,.... Vicious cycle.

[pblackburn]

Well I was looking for ground issues. Was reading cnccookbook about possible solutions. It was mentioning opto-isolated circuits. The board I have is but I thought I would review the manual anyway. Well found some possible problems, did not have the gecko power supply tied to ground but the one I never saw before was to not have the parallel port metal of the connector to touch the enclosure metal but needs to be isolated on a non conductive surface. Correcting these tomorrow.

[pblackburn]

Went over the grounds, made isolation plate for db25 connectors, grounded gecko power supply, installed zero phase reactors from Automation Direct. Ran spindle for 2 hours yesterday at 8000rpm with debounce at 20 (800µs) with no problems. Tied in GS1 and GS2 safety circuit for tool change. Have been running test programs today to see the lowest I can set the debounce. 5(200µs) no problems but 1(40µs) will not, get limit switch triggered on spindle stop. Trying 3 now. Will have new VFD cable tomorrow to add additional shielding. Screenshots to follow. Will try to upload a lot of pictures but may run out of allotment.

[pblackburn]

Found the information for the brain and modbus here at Mach Support Forum. Using a serial 232 to 484 converter as recommended by Danilo from ebay.

[pblackburn]

The VFD enclosure I used a shell of an industrial computer. It allows for free air flow while protecting fingers from getting into high voltage areas. The bottom had two 12 VDC fans, I removed them and have 2 24VDC fans to place in there if the VFD gets too hot. The cover is clear polycarbonate with an access hole to the keypad. The emergency stop is wired into 3 and 4 with the middle set of switches above the terminal set to on for the Safety Function that is built into the drive to work. The wiring is source logic, when the current is broken to GS1(3) and GS2(4) the VFD shuts down the output to the motor safely to protect the drive from damage. It will not run even if a signal from modbus or other is present, however if you send a run command and then release the estop it will turn immediately accelerate if it is in run. I am using it for tool change safety since I have to manually change tools. Both the supply feeds and the motor leads are wrapped around the Zero Phase Reactors I got from Automation Direct. I also used 5mm snap on ferrites on the motor leads and ferrites removed from pc cables on the supply feeds in both the vfd enclosure and the main panel.

The wiring for the top of the spindle was a bit of a pain. There are 4 pins and 1-3 are used and 4 is not used, in fact, 4 is not even wired, not even to ground. I had to split the cable insulation up farther and snake out the ground through the braiding and re-wrap the foam protective layer and the foil shielding. I then pulled the braid down and over the clamp for contact. Hitachi for some reason wants the braid grounded on both ends. This is something new to me since I was always taught to only ground one end of any shielding material. The bottom picture is just to see it. I picked up the leds from ledwholesalers for $6 and they work with 24VDC the color index is 5500 and they are bright. So bright that recording a video with them on is almost impossible but as for you seeing while is cutting, it is perfect.

The final setting of the Mach3 debounce was 4 (160µs). After multiple tests this is the lowest I can go. I have run test programs (dry runs) at every speed from 8000rpm up to 24000rpm.

[pblackburn]

The tank is a stainless steel with a clear polycarbonate top. The flow sensor is on the return. You can get them on ebay for $150.00 or less but new are much more. It is controlling a miniature relay wired to an alarm. When the flow gets below the preset, the alarm goes off. It is a Mallory Sonalert that is 120VAC. The flow switch does not have any interaction with Mach3 as many of you recommended. Thanks for all you input on that. It just makes its annoying sound momentary when the power is first turned on. The pump I used was a Tramm-Rio 1100. I am using distilled water currently. I spent many days talking with different chemical representatives and waste water personnel. All of the discussions made my head hurt. I learned the best coolant next to water is mineral oil. Will not grow bacteria and has a good thermal conductivity. But with the application as it is and the special hose needed for it, I decided against it. From what I was told, water in its purest form is a great thermal conductor but the additives from antifreeze and other coolant mixtures will lower the thermal conductivity. I did not see that as a problem but was more concern with microorganism growth that will occur in any coolant mixture and the handling and disposal of such coolants. In the end, a marine biologist help me decide what to do. Distilled water with regular changes monthly. He said to add a teaspoon of bleach a day or two before the change and just let is circulate for a day. The concentration of bleach is so low it will not be detrimental to the stainless steel and the would clean the water of any organism if any have started to grow. He even said that he did not foresee a problem with just treating the water this way and running longer but it was up to me. This way I do not have to worry about containment if a spill occurred and the water can be discharged down any drain and not need a special holding container and disposal concerns. The tank holds 7 gallons of water. $5.00 per month.

The other picture is what I did to get rid of my ground loop. I made a isolation plate out of nylon mounted in between the panel and the db25 connectors mounting plate.

[barry99705]

A lot of the liquid cooling computer people put a piece of silver coil in their cooling loops. It will kill off the bugs that can grow in the water. http://www.frozencpu.com/products/11...oir_Strip.html

I've also see people just use real silver coins, not that plated crap we call money now, but the old style Susan b's.

[KenC]

WOW! That's extensive work put into the spindle coolant & VFD enclosure compare to what I have.
Thanks for sharing.
I would suggest empty a small 50ml bottle food colour into the coolant, help a lot in enhancing visual impact.
AFAIK, copper is also a good biocide.

[pblackburn]

Now back to figuring out my previous problem of the lines not equally spaced. I have noticed it first on the diagonal photovcarve but now I see it does do it on the standard x move then y move. y+ movement is moves less after an x+ movement but moves more when a x- movement has been completed. I also noticed that on 3d the roughing pass is larger than the finish pass. I saw this at Vectric Forum.

Could it be a setting in mach3 config that is causing this? Any Mach3 users wish to throw me some knowledge it would be appreciated.

[danilom]

what kernel speed in mach do you use? on some computers anything above 25khz would give results like you described

[pblackburn]

Danilo,
Interesting you mentioned kernal speed. Made me look back over my notes. I was running 25k but for my setup the excel sheet with calculations on it say I should be at least 36.7k. In the beginning I had problems with the original computer not pulsing correctly so I switched to a new one. I never tried setting the kernel back up. I ran a driver test and it says it can run 45k. I think I should increase to that and tune my motors again.

A lot I have read recently says to set the CV setting to >85° for better control of 3d work. Is this advisable? Also to have the Z axis motor set to half the speed of the X and Y motors. I have notices that if the Z speed is higher, I get multiple shadings of the 3d image about an inch out. Still trying to understand it.

[sailfl]

Pete

Do you have links to the threads that talk about the kernal speed? Could you share?

The other controls for CV are important.... look ahead and the degree. I would also like to look at the threads you are looking at.

I recently increased my motor speed but I have not retuned. I will do that again. That is a good idea.

Thanks

[pblackburn]

http://www.mechmate.com/forums/showt...req+calculator

http://www.mechmate.com/forums/attac...0&d=1207125382

As far as tuning, it is mentioned in the Mach3 Install Manual

http://www.machsupport.com/docs/Mach...all_Config.pdf

[pblackburn]

Nils, a lot of the threads I saw were at vectrics forum. There are also mentions at cnc zone, sawcreekmill and others. Most of the research was done on another computer that I will not access to for 2 weeks. I was searching for mach3 cv settings toolpath and other key words on google. Sorry I can not be more help on that but it was really easy to find information about cv and roughing and finish passes. I know most said to make sure all the feedrates were that same for the 3d bits and the z axis should be at have the speed of the other axis. I do not quite understand this but I will try it.

[sailfl]

Pete,

I am relooking at CV, Debounce and other issues to insure that I have my machine tuned correctly.

I have my Z set at 150 and more recently set my x, y at 600. I have always had my Z at 150.

Thanks for the info.

[pblackburn]

I have been chasing this and do not know where to find the solution so I am hoping that maybe one of you can assist me. The roughing toolpath is always a little larger than the finish toolpath even though they are both off of the same oval vector. Also there is profile shading (for lack of correct terminology). The machine I use has steppers, the velocity of the x and y are 500 and accel is 50 and the z is 200 and accell is 45. I have read ahead 200 lines and stop cv on getter than 75°. I do not think it is missing steps because I can start stop rehome and resume at a line number and it will be right were it stopped. I can get the roughing pass to be within the finish path if I put in a negative boundry vector offset but I don't think that is correct. The multiple profile effect that is happening on the finish toolpath I have not idea what to look at. This was cut at 30 ipm. This happens on every piece of 3d I cut. V carve is not affected. I have read multiple posts at other sites and was looking for some insight into how to correct this.

[Surfcnc]

Run the code in exact stop and if the problem goes away it is probably the CV settings.

This link is ok.
http://www.machsupport.com/documentation.php has a doc on CV.
Might help.

Ross

[pblackburn]

Good Day Ross,

Thank you for the response.

I have read this article many times before. I was trying to avoid an exact stop only scenario. I thought that was bad for the machine from what I read. I know cv affects even the z axis. Would lowering the CV to 45 help? The multiple ghosts are annoying. But the ghost all created by the finish toolpath. You can see that here. Photo

[Gerald D]

Pete, I am afraid that the photo doesn't help me understand what you are trying to explain.

[smreish]

The slight sheen I see in the photo could be a couple of things.
The far right of object is the roughing area, the left of center is the finish pass area.
Are you sure the part isn't flexing up (bowing) and causing uneven 3D profiling?
This was a very common problem when I was doing a lot of 3d foam cutting and the material slightly climbing up towards the tool bit and causing the appearance of a deeper cut while in reality it was the base material moving.

[pblackburn]

Sorry Gerald, I am technically deficient when it comes to words to express things at times.

Sean,
The roughing cut was my mistake of not setting the boundry offset in aspire to compensate for the larger tool diameter.

The ghosting of the 3d model that is happening 3x is another story. I just changed my plunge rate to 10 on the Z and set my velocity to 140". The ghosting is a the problem now. Still cutting slow at 33 ipm. Need to find the ghosting problem before worrying about speed. Any input is helpful. This is my first cnc. I may repair them but understanding the fine details is something else of the HMI is something else.

[Gerald D]

Are you seeing ghost lines/images to the left side only? What is your cut direction?

[pblackburn]

It is on both sides but closer to the model on the left. Lowering the plunge gave more detail but still have some ghosting. I am using raster along x.

[Red_boards]

I did have a problem where the Z was bottoming out so I was losing steps and my finish pass was not hitting material, but I think your finish pass is too deep?

In post #204 you mention tool diameter. This was my first thought. Different tool for rough and finish passes?
I suggest trying the same tool for both passes. If the problem disappears, then it is tool specification in the software.

[Gerald D]

dragon.jpg

There is no ghosting in the yellow areas?

Does the change in the picture on the red lines mean anything? (This is what I first thought the picture was all about)

The dragon casts no shadow until the cutter reaches the blue scribble?

My impression is that there is no problem in the middle band of the picture, and that you see effects just off the middle band......

Does your finishing cover the whole oval area of the picture, or is your finishing on the dragon only? Did you sand anything before taking the photo?

[pblackburn]

Red and Gerald,
I ran the program again but with the Z plunge at 10 ipm. The ghosting is diminished but not totally. No it was not sanded in the picture. Was trying to achieve a carving that minimizes sanding in hard to reach places. You are right the problem is in the middle of the picture. On the second run the 0.125 cutter was removing almost 0.100" of material across the hole dish. I only see this with dish carving so maybe the plunge will have to be set very low to stop mach from moving until it has gotten closer to the bottom of the Z until continuing in the x direction. Still trying to wrap my head around the mach controls.

[Gerald D]

Originally Posted by pblackburn

. . . You are right the problem is in the middle of the picture. . . . .

I was thinking the opposite