G0704 CNC AC Servo Rebuild (Picture Heavy)

macardoso

H-M Supporter - Silver Member
H-M Supporter - Silver Member
Joined
Mar 26, 2018
Messages
2,724
Hi All,

About 5 years ago I completed a CNC conversion of a Grizzly G0704 milling machine. Like everyone at that time I chose to go with the Hoss Machine style conversion as there were not many options on the market. I was not comfortable designing my own system at the time since I was just starting out in the hobby. The conversion was not glamorous, but it worked and I've tended to use the machine more than just fixing it up. The original purchase and conversion have been more than paid off in fun side jobs. I have literally thousands of great parts that have come off of this machine.

Fast forward to today, I am graduated from college and working for an industrial automation company (specifically in servomotor applications and control system design). I have been extremely fortunate to be able to get my hands on hardware leftover at work that I would have no justification to purchase on my own. Off and on over the last year I have been designing and building a complete system overhaul for the machine. I'm primarily focused on ease of use and reliability of the machine, but there should be huge performance increases as a result. I know everything you are about to see is complete overkill for this size machine, but I left it flexible enough to be transitioned onto a larger machine should I outgrow the G0704.

System overview:
The new control cabinet is a 24" x 36" x 8" sealed enclosure housing 6 AC servo drives.
Control 1.jpg
Designed to be run from a Ethernet Smoothstepper and Mach 3/4, the Allen Bradley Ultra 3000 series drives power AB bulletin MPL servomotors.

  • X Axis - 0.75 kW 5000rpm Max, theoretical rapid 1000ipm, continuous thrust 350lbf, peak thrust 800lbf. High resolution multi-turn absolute encoder (~2M count/rev)

  • Y Axis - 0.75 kW 5000rpm Max, theoretical rapid 1000ipm, continuous thrust 350lbf, peak thrust 800lbf. High resolution multi-turn absolute encoder (~2M count/rev)

  • Z Axis - 1.80 kW 5000rpm Max, theoretical rapid 1000ipm, continuous thrust 900lbf, peak thrust 2400lbf. High resolution multi-turn absolute encoder (~2M count/rev)

  • Spindle - 1.80kW 5000rpm Max, Continuous torque 4.18Nm, Peak torque 11.1 Nm, 1:1 gearing, High resolution single turn absolute encoder (~2M count/rev)

  • 4th Axis - Prewired for up to 1kW. I have a high torque Alpha Wittenstein SP series gearbox which I plan to make a 4th axis with in the near future.

  • 5th Axis / Spare - Prewired for up to 1kW.

The panel is wired for 40A @ 240VAC. It theoretically could pull 70A, but I don't foresee running all 6 drives at 100% at the same time. The control voltage is 24VDC which is coupled to the 5V Smoothstepper through optical isolators. All I/O points are broken out to M12 industrial connectors which include DC power and common to allow direct connection of 3 or 4 wire sensors. I have also purchased a pendant.

Mechanically, all of the original conversion hardware will be removed and replaced with new heavier duty components that are already built (not trying to make this a build log per se). The spindle is getting the most work as the bearings have been dying a slow death over the past year. The new design will be a totally sealed belt drive (noise reduction) and feature a pneumatic drawbar. I have started this process and will share lots of pictures in the next post.

Rebuild plans:
  1. Rebuild spindle
  2. Install spindle motor, belt drive, and drawbar
  3. Remove old CNC components
  4. Install axis motors and mounts
  5. Install electrical cabinet
  6. Install limit/home switches
  7. Install wireway and cable routing
  8. Install touchscreen computer
  9. Migrate to Mach 4
  10. Validate system functionality
  11. Performance characterization
  12. Mach 4 Pendant Configuration
  13. Mach 4 LUA Scripting for servo drive serial communication (Diagnostics, true rigid tapping, absolute homing, gear hobbing, etc.)
I'll follow this post up with some info on #1 since I'm nearly done with that part.

As promised by the title, some pictures:
06.PNG
The freshly completed Hoss Machine style conversion (ca 2012)
01.PNG
The very first control panel, literally just stepper drives and a power supply
00.PNG
Got a touch screen monitor
04.PNG
Upgraded control panel (ca 2013). Ethernet Smoothstepper, CNC4PC breakout boards, StepperOnline drives, and a DMM Tech DYN3 AC servo for the spindle.
05.PNG
As installed (24" x 24" x 6")
07.PNG
Servo spindle motor with belt drive and plastic cover
02.PNG
Pretty much what it looks like today, except for some cleaned up wiring.
08.PNG
Machining some valves for a robotic bartender I built with some buddies in college.


I hope you will follow along with me as I complete this project. I have been very grateful to this entire community for the knowledge that is shared, and I hope to be able to give back even if it is a fraction of what I have learned!

Mike
 
Last edited:
Very interesting and putting A/C Servos onto a G0704 is indeed some overkill but "anything worth doing, is worth overdoing" :)
 
Nice Job! :encourage:

Looking forward to more
 
OK, so I am part of the way through what I listed as step #1 above (really more like step #437) which is rebuilding the spindle. I've been uncomfortable working on the spindle in the past as it requires a lot of care and precision as well as some tools that I don't have (arbor press, bearing pullers, etc.). That being said, my current setup has significant notching to the bearings and runs very hot and loud. They need to be replaced.

Prior to starting work, I made a set of tools to disassemble the spindle, press the bearings on and off, and get things adjusted correctly. If anyone wants a rough set of drawings for their own use, I'd be more than happy to make them up. I think the job was much easier with these.
IMG_3553.jpg
Pin spanner for removing the lower bearing retainer
IMG_3554.jpg
Deep pin spanner for adjusting and removing the spindle locknut (sometimes called an adjuster nut)
IMG_3556.jpg
Various rings to press bearings
IMG_3557.jpg
Tool to knock out the upper quill tube assembly


I began by removing my old AC servo, belt, and pulleys. From there, the drawbar could be removed and the entire lower section of the spindle drops out. There was a touch of surface oxidation on the spindle housing, but otherwise it looked really good. At this point I could tell that the angular contact bearing pair on the spindle itself was the source of the notching, and the upper quill tube bearings spun smoothly (I still replaced them since I had purchased new bearings). I used an aluminum disc and a deadblow to knock out the quill tube.

I removed the spindle head casting to fill a few holes with epoxy. I have designed the new belt drive to be "air-tight" to reduce the audible noise from the system.
56971895231__0CE5A755-A584-47A6-BEA3-0EC73D4FE3F7.JPG
IMG_3537.jpg
IMG_3542.jpg
There was a little porosity due to using JB Kwik, but it won't affect the final product.

Next I worked on removing the quill tube bearings. Many youtube videos show people popping these off with a pair of screwdrivers no problem, but my experience was different. After an hour of getting nowhere, I put it away for the night and went to Harbor Freight to get a 3 jaw gear and bearing puller. I made a quick tool to go on the end of the quill tube so the bearing puller had something to push against, and viola, the bearings came right off. I highly recommend the $20 to have one of these in your tool box for times in need.
IMG_3555.jpg
Quick tools don't have to be pretty.

Since I don't have any kind of press, I use my lathe to apply pressure to press parts together (pressing against the side of my toolpost). Pressing the bearings for this project is about the most force I'm comfortable applying with this method. Using a machined aluminum disc and a piece of scrap tubing, I was able to carefully press on the new bearings and reseat the snap ring.
56971896605__83A8ED69-C900-455C-8C6D-68C4CBE31A38.JPG
Before
IMG_3558.jpg
After

I'll save my angular contact bearing woes for the next post. Night!
 
Yesterday I had planned on spending the afternoon and evening to replace the angular contact bearings and reinstall the spindle components into the casting. I wrote up a procedure for how I was going to assembly the bearings such that I would never press across the balls of the bearing.

First step was to freeze the spindle shaft in the freezer and heat the bearing at 250*F in the oven (limit for phenolic cage material). I waited about 30 minutes then dropped some spacers on the spindle shaft, chilled it further with a gas duster, and pressed the bearing on. This went quite easily and the bearing was seated very flat, confirmed on a .0005" indicator and my surface plate. Next I heated the spindle housing at 350*F and chilled the upper bearing. When I went to press it on, it just dropped into the housing, no press required. After it cooled, I could remove the bearing by hand. The housing measured a few tenths oversize on the bearing so the fit was a tight sliding fit. There was no radial play on this bearing. I revised my plan to include pressing this onto the spindle shaft after the shaft and lower bearing were installed in the housing.

The next step was to heat the housing to 350*F again and chill the spindle/lower bearing assembly. After 30 min I set everything up on the lathe to press it on, but when I went to slide the shaft in, the bearing got stuck crooked by about 5 degrees in the threaded portion of the spindle housing. The thermal shrink kicked in and I was screwed. SH*T!!!. I took a deep breath and though about how I was going to remove the bearing. I tried pressing it from the rear, I tried pressing it in further from the front, hoping that it would straighten out, but neither worked.

After an hour of no success, I decided to whack the spindle shaft laterally with a deadblow to align it again (and thus dooming the bearing to the garbage can. This did free the shaft from the housing, but not in the way I had hoped. The races separated on the bearing and all the balls dropped out. SH*T!!! From this point I could see the bearing was stuck slightly in the bearing bore of the housing, and jammed into the gully of one of the threads. It looked like a 1 : 1,000,000 chance. Anyways, when I went to HF, I bought a set of 2 jaw pullers, a set of 3 jaw pullers, and a pilot bearing puller (planned to return whatever I didn't need for the job). I didn't think I would use the pilot bearing puller, but boy did it save the day. After a few minutes of fiddling with the setup, POP, the outer race was removed. Now the inner race was still thermally shrunk to the spindle shaft and was nearly flush with the nose of the spindle, I certainly wasn't going to be getting the bearing puller on there. I carefully reassembled the cage and balls and basically hammered the bearing back together (reattached the outer race to the stuck inner race). Using an aluminum ring from before, I was able to press the inner race off as well. WHEW!

So at the end of the day, there was no damage to the spindle shaft or spindle housing, but the angular contact bearing was toast. I have a new one on order and I will be making some tooling to make sure that the spindle is aligned when it is being pressed into the housing.

I hate working on spindles...
 
Also, question for everyone. I had previously purchased 7005C and 7007C angular contact bearings (15 degree contact angle). This time I mistakenly purchased 7005B and 7007B angular contact bearings (40 degree contact angle). Does anyone see an issue with using the 40 degree ones? I understand they trade some of the load rating from the radial direction for more axial, but could this be an issue? They are still rated for much higher speeds than they will be run at, and I can't imagine that the loads I will be applying to them would be anywhere near their limit in either direction?
 
While I am waiting for the new spindle bearing to arrive, I am doing a bit of re-wiring on the electrical cabinet. Since I finished it a year ago, I have decided to add another 3.4A 24VDC supply and a time delay relay to power the industrial PC running this system. I found the PC to be sensitive to the swing up on the power supply voltage when it first turns on, so a 1 second time delay will wait for the voltage to stabilize before closing the power contact to the PC. I'm using an AB 700-FS timing relay for this purpose.
26fabff1827290353cc39f13705c8c57c0336712-medium.png
In addition, I am adding 2 DPDT relays to allow Mach 4 to mechanically disconnect the signal lines from the smoothstepper to the Z and 4th axes and connect these axes directly to the feedback from the spindle drive. If I can figure out the serial communications to the drives, I should be able to set the gearing ratio on the fly to do true rigid tapping and gear hobbing. That being said, it will be one of the last things I try to get working once everything else is finished and working, but I'd rather have the wiring in there already.

I didn't have much spare room on my 11" section of DIN rail, so I am removing all the single tier terminal blocks and replacing them with these 3 tier ones.
1.jpg
This gives me about 240% increase in wiring density and just barely enough room to fit everything. With the addition of the jog pendant, power drawbar, and a 3 button operator station, I have used every one of my remaining 13 inputs and 4 outputs I saved as spares. But it all fits :)
2.jpg
Completed 3 tier wiring, you can see how much space it saved. I haven't done any dressing of the wires; I promise I'll make it look pretty
4.jpg
The original lower density terminal blocks.
 
crikey, very impressive work! I haven't pressed on any spindle bearings, but I've pressed on a bunch of bearings before and getting them on/ in straight is always tricky. The best way I've found (where possible) is to use a threaded rod with end plates that locate both the bearing (sized to fit part of the ID of the bearing) and the housing (sized to fit the ID of the housing). That gets the bearing started as close to perfect as possible. Once it starts going on straight, you can swap out the bearing plate for one that allows you to press the bearing on all the way. The screw nature of the press means you go real slow and can correct any errors as you go, rather than getting part of the way and realising the bearing is cocked.

looking forward to the rest of the build!
 
Well I had a pretty productive weekend on this project.

While I waited for a replacement bearing to arrive, I did some more work on replacing terminal blocks.
IMG_3569.jpg
I was able to fit all the new relays, and have 18 spare points should I need them.
IMG_3571.jpg
I also got a little help from Zoey (she likes to bite my hand when I go digging for wire ferrules in the bag)
IMG_3573.jpg
After I spent half an hour dressing the wires.

I still need to land connections between the empty block on the left (the MPG pendant) and the breakout board for the Ethernet Smoothstepper, and make the connections to the new relays, but it is getting close!

IMG_3569.jpg

IMG_3570.jpg

IMG_3569.jpg

IMG_3572.jpg
 
I had watched a youtube video last week about another G0704 belt drive conversion where they warned that the inside diameter of the quill tube had significant runout. I measured mine at .006" which was a lot higher than I wanted. I removed the rear bearing using a bearing puller and covered the front bearing in a generous helping of masking tape:

View attachment 286101
View attachment 286102

I then spent 45 minues indicating the part to run concentric and coaxial to the spindle (major PITA with so little material to hold onto and all machined surfaces had 12 thou or more of runout). I skimmed the bore until it cleaned up.

View attachment 286103
View attachment 286104
View attachment 286107

After this machining, I had about .0005" of runout when running on its own bearings. I was plenty happy with that.

View attachment 286108

Again I used the lathe and some tooling plates to gently press the lower bearing back onto the quill tube

View attachment 286109
View attachment 286110

Since I had enlarged the bore and managed to ruin both flanges on my previous spindle pulley, I got to work machining a new one:

View attachment 286112
(Pulley stock on the left, ruined pulley in the middle, and the old steel L series pulley on the right)

I used a split ring tool to hold the pulley on the teeth without damaging the flanges. This allowed for work on both the bore and the hub OD in the same setup.
View attachment 286113
View attachment 286115
View attachment 286116
I indicated this into true and corrected axial runout using a lead "tapper"
View attachment 286117
View attachment 286118
View attachment 286121
Since all my parts were designed from a 3D model of the G0704 that I purchased, there were a few sizing errors that I had to correct. In particular I tightened the pulley bore to spindle spline fit to a light hand press.
View attachment 286124
View attachment 286125
(CCGX insert, picture doesn't do justice for how good the bore finish was)
View attachment 286126
Shop kitty #2 needed to see what was going on.
View attachment 286127
Finishing hub OD to prep for pressing on an aluminum ring (effectively making the OD larger than available from SDP-SI)
View attachment 286129
The finished precision aluminum ring, fitted to the newly bored quill tube.
View attachment 286131
And pressed on with some Loctite 271.

IMG_3615.jpg
 
Last edited:
Back
Top