[N4NV]

I did not want to hijack JD's thread any further. I am starting a conversion of my stepped pulley J-head Bridgeport mill to CNC. At the same time I am converting a Taig lathe to CNC. I got a lot of information from JD's thread and much of what I plan to do will be similar to his. I don't really have a good reason for the conversion other than the entertainment and educational value of the project.

I purchased my mill about 8 years ago for $500. It was pretty worn out. I had to shim the Y-axis gib to get it close to being useable. I could get the slop in the table down to .002, but then it would be tight at either end of the travel. I could get decent result if I locked the table in the direction I was not milling. So, to start this project I sent the table, saddle, and knee out to a machine rebuilder to have them ground, scraped and fitted. No point in starting a project like this if the mill is this worn.

I am not going to install ball screws at this time. The machine rebuilder guarantees I can get .002" backlash when he is done (and less than .0005 slop in the table). I can take care of this with software. If it turns out that I am not getting the results I would like, I can always add them later.

I am really starting from scratch. I found 5 60VDC motors at Triangle machine (on of the places that the "MythBusters" shop). It took about three days to get all the specs, but I think they will work for this project. They have no encoder or shaft for an encoder so I will have to add that. The brushes and bearings were shot on most of the motors. The bearings are standard R10ZZ. I found the brushes at Arrowhead Electric Company.
http://www.carbonbrush.com/brushes.htm
They were about $6 each and it takes 4 per motor. I paid $50 each for the motors, $120 for all the brushes, $40 for bearings and $54 each for three encoders.

The first picture is my mill stripped and waiting for the parts to return from the machine rebuilder.
#2 is boring the hole for the encoder shaft in the end of the motor shaft.
#3 shows the armature before I turned it.
#4 is turning the armature.
#5 is the finished armature.

To bore the hole in end of the motor shaft I mounted the motor in my lathe and centered it as best I could. Then I added power to the motor and used and end mill in the tail stock to bore the hole. I did this before I replace the bearings in case I caused them stress from the procedure. Next I pulled the bearing and turned the armature. After turning the armature you have to undercut the armature sections so there is a little gap between them. I installed new bearings, assembled the motor, installed new brushes and tested them. They all worked great. Today I am going to work on installing the encoder shafts and mounting the encoders.

A word on encoders. I purchased some Renco encoders, but after talking to Mariss at Gecko, I found that they drew too much current. The Gecko’s can only deliver 50 mA to the encoder. These Rencos required 125 mA. I purchased some US Digital E5S encoders that will work with the Gecko 320.

Vince

[N4NV]

I installed the encoder shaft on three of my servos today. I installed the encoder bases as well. When I ordered the encoders, I got the adhesive base option as the mounting holes in the base would not line up properly. It takes 24 hours for the adhesive to fully cure so I will install the rest of the encoder body another day.

I moved on to my power supply. I found a 120 to 43 volt transformer at Triangle Machine. That should give me 60 VDC after rectification, which is what my servos are rated for. I purchased one 7,500 uF capacitor at Triangle, and a friend had two more. I purchased two 50A bridge rectifiers on eBay. I hooked it all up and found out that the bridge rectifier was made out of compressed smoke . Only two of the terminals were marked, one + and the other AC. I guessed wrong on the other two. I looked up its data sheet on the internet and found the correct way to wire it. It is a good think they came in pairs. The second time nothing dramatic happened. I measured 59.8 VDC on the output of the capacitor bank. I connected one of the motors and it ran great. What is a good value and wattage rating for a bleader resistor for these caps?

I'm running out of things todo. It will be another two weeks before I get the parts back for my mill. I need to get it running to start making things.

Pic #1 shows the alignment tool that you have to buy separately from the encoder to ensure that you have the base centered properly.
Pic #2 shows two motors, one with the encoder shaft installed, and the other with the shaft and encoder base.
Pic #3 shows the power supply hooked up and running one of the servos in the vise.

Vince

[Eric]

Where is Traingle Machine located?

[N4NV]

Triangle Machine is in San Jose, CA on comercial court, which is off of comercial street. Take 101 to 13th street north, turn right on comercial street, left on comercial court.

Vince

[N4NV]

I finished my servos today. Mounted the encoders and a 3/0 Bell box extension to protect the plastic encoders. I am stuck again for now, I ordered three connectors from US Digital to pug into the encoders, but I can't figure out how to attach my cable to them. I might need a special tool. US Digital calls the connectors a finger latch connector. It looks like the audio connector on the back of a CD or DVD drive in a computer. If I have to get premade cables, I'll be another week before I can test the servos.

I received a Gecko G320 today and a cnc4pc parallel port break out board. Once I know that everything will work electrically and I can drive my servos with Mach 3, I plan on updating to the GREX 100 and Mach 4. I will order another two 320's as well.

The machine shop called today and said my mill parts will be ready to be picked up at noon tomorrow. It looks like I will have more to work on for the next couple of days. I ordered an electrical cabinet to mount the power supply, GREX, Gecko's etc. It will come in next week.

I uploaded three pictures that show the encoder electronics and body installation.

Vince

[jdelaney44]

Vince, nice job on the encoders!

Your lathe is going to be good for working the screw for the Z drive. Wish I had one that big when I was turning mine down.

Best,
-jd

[Eric]

Vince.

Try going over to Halted Specialties on 3500 Ryder St. in Santa Clara (408-732-1573). They have tons of small motors over there with encoders. Maybe you can find one with a cable attached that will work with your encoders. If you need any electrical help, don't hesitate to ask.

Where did you send your machine to get reworked?

--Eric

[N4NV]

I have been going to Halted almost every day for the past two weeks (along with the trips to Triangle Machinery). They don't have that type of connector. US Digital sent me a PDF that shows how to make up the cable to the connector. You need a special tool, but I think I can do it with some small needle nose pliers.

I had my mill ground and hand scraped at D&L precision, 2821 Fair Oaks Avenue,
Redwood City, CA 650-444-5208. Tell them you heard about them from Vince. Jason runs the place. They are out of the shop about half the time. Most of the big machines they rebuild are done in place. I picked up my parts today. They are still in the back of my truck. I am waiting for my wife to get home and give me a hand. When I get them mounted back on the mill I will post some pictures. Jason said they were in pretty bad shape. He quoted me $700 for grinding the table, saddle and knee. He said it took him two or three times longer than usual to get them flat and square. I guess most machine shops rebuild them before they get that bad. It was another $540 to have them hand scraped and fitted, which included shimming and hand scraping the gibs. I should be good for .0005 now.

Vince

[N4NV]

My wife got home and helped me get the trailer out of the garage so I could have room to work, then she left for a meeting. I put the mill together by myself. I uploaded some pictures of the process. It went together a lot easier than it came apart. I still have a lot of little items to install like the DRO scales and the oiler system. It came out great. There is no play that I can feel with the gibs adjusted. Unfortunately, wear showed up in other places. I found that my lead screws are worn in the middle. I have adjusted the half nuts and am able to get .003 backlash, but the table and saddle get tighter at either end of their travel. Without the lead screws, I can push the table or saddle the entire length of their travel with one finger. . I may have to go to ball screws after all. That would really blow my budget and could delay me for a while.

The entire assembly process took me 3 hours. I installed the table twice. The first time I had almost .200” of backlash on the table. I could see it was coming from the half nut in the dog bone. I could not tell exactly what was happening so I pulled the table. I saw that I had applied pressure to the half nut when I installed the lead screw. This pushed it together and allowed the nut to slide back and forth in the dog bone. I reinstalled the lead screw, this time holding the nut in tension. It worked and I the backlash dropped to .003.

Vince

[jdelaney44]

.003? I'd give it a go. The software can suck up .003. You can just call it .0015 on the average.

You will have to be careful about transitions from feeding into the cutter and climb milling where the cutter is pulling the work in.

+/- .005 is actually OK for a whole lot of applications. So you are going to be OK for many things. Not sure what you are going to make with it.

When you can afford the new screws it will just get better.

-jd

[CNCfun&games]

Vince,
If you decide to go for a Ball Screw Conversion, check out Hiwin Corp., I purchased their Bridgeport
Ground Leadscrew Kit w/ Yoke for $950 +$30 (S&H)which was less than most of the Rolled
kits I surveyed in 2003. They are located in San Jose, CA. (408)943-0290
I've been using the screws in manual mode for about 3 yrs. & they are excellent.

On the Z axis retrofit you can minimize your alignment problems if you machine 2 brackets to mate with the stepped reference points on the quill STOP screw mounting Bosses(see picture)
Also are a couple of pictures of the Z Quill/Leadscrew Mtg Block and my Y axis plate.

Regards....Will Smith

[N4NV]

I had a conversation with Mariss yesterday about the various tasks I need to do to get my servos up and running. During the conversation, he mentioned that not all DC motors can be good servo motors. He said on servos the commutator was twisted like a barber poll. I remembered that mine were not. He said a way to tell if it was a good servo was to run it at a very low voltage until the motor is only turning a few rpms. DC motors have something called torque ripple which with cause the motor to turn in a jerky fashion when at low rpm. I started to worry that I might have wasted money on my servo conversions. I could have sworn that the factor rep for the motors told me they were servo.

As soon as I got home I put one of my servos on an adjustable DC power supply. At 1 volt (my motors are rated 60V) it turned too fast to count the rpm. I turned down the voltage to 0.3V. At this voltage it turned at 2 rpm, and smooth as a gear motor. There was no sign of torque ripple. I turn down the voltage as low as my power supply would go, 0.26V and the motor slowed to under 1 rpm. I could just make out the torque ripple. It looked like the second hand on a spring wound watch. I was amazed that it would still turn at such a low voltage.

I looked up information about DC brushed servo motors and found that there are several ways to combat torque ripple. One is by skewing the armature windings. Another is to increase the number of poles or brushes. If you look at the pictures of my motors above, you can see they have both skewed windings and 4 brushes.

Just for the fun of it I hooked up another DC motor I had laying around. When I got to 1V on this 90V motor, it would turn in 15° jumps. A very noticeable display of torque ripple.

Vince

[N4NV]

Vince,
Also are a couple of pictures of the Z Quill/Leadscrew Mtg Block and my Y axis plate.
Regards....Will Smith

Where did you get your timing pulleys and belt? I am going to have to have to do my Y axis like you did. My servo motor is too long to place under the knee.

I talked to Hiwin today (they are in Fremont now) and they could not figure out what I needed. I have the Bridgeport power feed on the X and this changed the length of the screw. Looking through their documentation, I could not find a diagram that would fit the power feed. Rockford did have a screw specifically for the power feed.

Vince

[CNCfun&games]

Vince,
I buy my HTD Pulley's and Belts from: www.sdp-si.com
Pulley's 5MM-15 mm wide(28 tooth) A 6A25M028DF1508
(56 tooth) A 6A25M056NF1512
Belt 6R25134150 ( yours may vary)There are belt /pulley formulas
in their Catalog.

Since you are discarding the power Feed why can't you use the standard
HIWIN CORP B3205X-42 Ball Lead Screw? (see enclosed invoice)
Regards....Will Smith

[N4NV]

I ordered my timing gears and belts on Saturday and picked up some 1/2" aluminum plate for the motor mounts. Hiwin is sending me a quote for their ball screws. I spent Saturday and Sunday putting my power supply into an enclosure. I solved the encoder connector problem. Using a microscope, I was able to solder the wires from a Cat 5e cable to the connector. I hooked up my Gecko 320 and the parallel port break out board and started up Mach 3. All seemed to work fine as I went through the start up procedure in the Gecko manual. When I hooked up the motor, something is wrong. The motor jumps back and forth about 1/32 of a revolution continuously causing constant faults on the 320. It is not making a "singing" sound like the manual says it might, but more or a low growl. The gain, current limit and damping trimpots are set to the suggested positions. Adjusting them has no effect. I disconnected the printer cable to see if it was coming from the computer but nothing changed. A call to Mariss is in order, until then, any suggestions?

Vince

[jdelaney44]

I had a similar problem but it was from the VFD noise finding it's way back up the mains, through the PC power supply, and then out LPT1. Fixed it with ferrite beads as you have read.

I see that you are missing the capacitors on the power side of the Geckos. Very important. I had issues without these as well.

Is the encoder wire shielded? CAT5e is good, but shielding is still needed and the shield needs to be grounded on one end only. Usually the control end. The shield needs to end as close to the G320 as possible in my experience. Marriss posted something about this at DIY-CNC.

My motors don't sing either. They are totally silent. Singing is dithering. Growling is either BIG dithering or some other kind of periodic error compensation or interference.

Does the growling sound anywhere close to 60hz? I'm looking at that big old magnet you got going with that tranny and thinking it's messing with the Gecko.

These reliance motors I have, have some indexing at low voltage as you suggested with your motors as well. So, I'm thinking it's not the motors.

-jd

[N4NV]

Mariss said that cat 5e was supposed to work for the encoder. He also said that I may or may not need the cap. I am online ordering the cap from digikey now. I have the motor leads in shielded cable. I don't have a VFD in the enclosure. The growling sound (and movement) is random and not at any particular frequency. Sometimes the motor is turning slightly forward and sometimes slightly backward.

Vince

Vince

[jdelaney44]

Hmmm, he said otherwise on the DIY-CNC post. He said shielded cat 5e on that one. In theory the twist should do it.

OK, well, what you are describing was EXACTLY what I was experiencing. It took a combination of the caps AND the ferrite beads to nail it. I also shielded the wires from the break out to the G320. But, every set up is different.

Cool. Well it will be interesting to see if the same things fix your problem that fixed mine.

Best,
-jd

[N4NV]

I called Mariss today. He said since I ran Cat 5e cable to the encoder, the most likely cause of my problem was I ran the A&B channel down the same pair. He said that Cat 5e is as good as a shielded cable. He added that since I just replaced the brushes, the motors need to be run for 20 hours in both directions.

When I got home and checked, sure enough both the A&B channels were on the same pair. I had to chuck that connector and make up another with the A&B channels on separate pairs. I hooked up the drive and motor and tried again and it really helped. The motor was still hunting back and forth a little, but not as much as before. I ran the motor on straight DC and tried again. It ran a little better. I am now letting it run over night.

It has been 20 years since I used and oscilloscope so I don’t know if it is my scope or how I set it up, but I could not get the screen to look like the sample in the tuning section of the manual. I played with the gain and damping a bit, but without the proper display, I could not tune the motor. I will try again tomorrow after the motor has run for a day.

I have been holding off using my mill until I can get the oil system installed. The machine came with zerk fitting which someone in their great wisdom filled with grease. Disassembled the mill, cleaned it, reassembled it and installed my own one shot oiler. I did not know about metering device then. I ordered some Bijur meters, pn B1114 to install on my oiler manifold. These are a size 2, which D&L said were the size to use. When they get in, I can finish assembling the mill and start on the drive mounts.

While I am waiting for all the other things to come together, I spent some time fabricating the mount for my Gecko’s. I am using the same type system as JD. When I get something worth pictures I will post some.

Vince

[N4NV]

After spending an hour on the phone with Mariss, I think I found the problem getting the motor to work with the drive. My computer has 3.5V logic and the parallel port interface card I am using is made for 5V logic. I don't want to spend much for an interface card because I am going to change to a Grex 100 as soon as Mach 4 is stable. Does anyone have a suggestion for an inexpensive interface card that will work with 3.5V logic?

Vince