• Scam Alert. Members are reminded to NOT send money to buy anything. Don't buy things remote and have it shipped - go get it yourself, pay in person, and take your equipment with you. Scammers have burned people on this forum. Urgency, secrecy, excuses, selling for friend, newish members, FUD, are RED FLAGS. A video conference call is not adequate assurance. Face to face interactions are required. Please report suspicions to the forum admins. Stay Safe - anyone can get scammed.

G3616 Conversion.

jcdammeyer

John
Premium Member
I'm using the Grizzly part number because the House of Tools one is just too long. And it's big brother with the horizontal mill, the G3617 under a different part number, is still available from Grizzly.

I'll try to make this thread about the conversion process of this sort of mill. Bare in mind that I have a foundry and that part of my day job involves PC board design and software. So what I might imply is simple, is because it's simple to me.

I have a JGRO (MDF and roller skate bearings) CNC router with stepper motors. I don't really like them. So noisy but they are easy to install and work with.

Initially I put a 500 oz-in motor with 3:1 reduction on the knee. The sloppy fit of the hand crank and almost dropping it on my toes once was the initial reason for the conversion. I ran it with a Gecko Stepper drive and 60V power supply. It became very clear that 500 oz-in wasn't enough so I upgraded to just over 900 oz-in. Oh and I used one of my prototype ELS controllers as the stepper control for motion. No more hand cranking.

Controlling the knee wasn't nearly as simple as one might think. it's a right angle drive to a bevel gear set to turn the ACME screw that lowers and raises a very heavy assembly what with the rotating table arrangement (w/o a horizontal drive capability). And one set of bearings were just cast iron plus there was a lot of backlash on the gears. Running this shaft at high speeds would clearly result in a lot of wear.

So the first step was to place a ball bearing at the bevel gear end and adding a bearing at the crank end. End play was set up at the bevel end for best contact and as little backlash as possible.

So I replaced the iron bushing with the carrier shown in the KneeBackBearingHolder photo.

At the front end I added a casting that held a bearing. It was just short enough to fit on the lathe for boring the slip fit for the ball bearing. The closeup shows two bolts holding it in place and the cover just sitting above on the table.

A second casting held the motor.

The pulley slides over the shaft and 4 bolts anchor it on the 4 flats where the original handle fit so badly.
 

Attachments

  • KneeFeedPartsSmall.jpg
    KneeFeedPartsSmall.jpg
    146.8 KB · Views: 0
  • KneeBackBearingHolder.jpg
    KneeBackBearingHolder.jpg
    115.5 KB · Views: 0
  • BoringKneeBracket.jpg
    BoringKneeBracket.jpg
    155.7 KB · Views: 0
  • KneeBracketCloseup.jpg
    KneeBracketCloseup.jpg
    106.5 KB · Views: 0
  • BearingHolderInstalled1.jpg
    BearingHolderInstalled1.jpg
    78.1 KB · Views: 0
Those knee modification photos were done in May 2010. Things were relatively stagnant even though I'd had enough of the machine apart to at least get an idea of what I was dealing with. The photo of the Y axis bracket and thrust bearings was from April 2008.
YAxisBearings.jpg
I had modeled what I needed.
YAxis-Assembly2.jpg
The pattern posts were turned on my Gingery Lathe using the taper feature of the ELS. Once screwed in place auto body filler was used to make a fillet.

YAxisPattern-1.jpg
It was September 2017 when I started boring the bearing hole for the Y axis.
YAxisMachining-6.jpg

Here are the castings posing with the motor and the HP_UHU Brushed DC Servo driver. I had a small stub arbor I'd turned on the lathe to hold the pulley in place. I used the ELS to drive the servo just to make it turn back and forth.

YAxisMachining-4.jpg

This short Youtube video shows the X and Z in motion. Gecko drive on the Z axis, HP_UHU on the Y axis, MachineKit (LinuxCNC) running on a Beaglebone Black with a Cape specifically for MachineKit. Short G-COde program moving the two axis back and forth.
The Stepper motor power supply is hung on the wall on the left. The 105 VDC servo supply, Beaglebone and BoB are out of sight to the right.
This was May 2018 now 10 years after I first disassembled the mill to figure out how to power the knee

John
 
Hi john , thx for posting. Good pictures/descriptions

I was the member who purchased the similar mill (more like the g3617/g0757 with the horizontal spindle) from @Janger

It too is converted to cnc but using the Dow feed instead of the knee

I too do foundry work and am down various paths of the gingery journey

Question, did you convert to ball screws ? And what software are u running? I just converted to Linuxcnc using a Mesa 7i76e Ethernet board. So far so good!




Sent from my iPad using Tapatalk
 
Hi john , thx for posting. Good pictures/descriptions

I was the member who purchased the similar mill (more like the g3617/g0757 with the horizontal spindle) from @Janger

It too is converted to cnc but using the Down feed instead of the knee

I too do foundry work and am down various paths of the gingery journey

Question, did you convert to ball screws ? And what software are u running? I just converted to Linuxcnc using a Mesa 7i76e Ethernet board. So far so good!

Sent from my iPad using Tapatalk
I originally intended on feeding the quill. Even made a bracket to which I've attached the scale for the Shumatech DRO.
QuillClamp.jpg
You can just see the scale between the power switch box which I moved away from the mill to make room.
DRO_SwitchBox.jpg

I'm running a licensed copy of MACH3 on my CNC router but to be honest I'd also then have to get one for the mill so that's one cost I'm not sure I want to have. Second, the DC Servo motors run 150 ipm on the X axis with a 250 line encoder (x4) so although MACH3 can do the step rates, do anything else on the machine that causes a PC hiccup on the step pulses and the motor drivers instantly go into FAULT.
I've borrowed the USB smooth stepper from the JGRO CNC router and that problem goes away so another cost is $180US for the Smooth Stepper.
I'm using a PMDX-126 Break Out Board which, like the PMDX-125 on the JGRO CNC router, has mounting points for the Smooth Stepper so you can see I did originally intend on using MACH3.
But for $89 and minimal shipping because a friend brought it back from the USA I've also got a MESA 7i92H which is really from a hardware cable perspective the same as the SmoothStepper but for LinuxCNC.
So I set the surplus Lenovo bought from BC Asset Recovery for $75 to be dual boot with both Windows XP and LinuxCNC. As long as I use just the parallel port I can dual boot and run either.
Slowly as I switch back and forth between the two I'm starting to like LinuxCNC a bit more. Perhaps a lot more once I install touch plate probing and the module that adds wizards. I really like the wizards on MACH3.

Ball screws will have to wait until I have the spindle conversion and TT auto tool changer done. That and a shroud around the system so I can use either the flood coolant pump or a mister.

And I'm not done yet with the cart that holds the electronics and will have an extendable arm for the keyboard and monitor etc.

I'll go down today and take a photo of the current setup. Don't have anything recent.
 
One more little change to this mill that really can apply to just about any of them. The work lamp that came with the mill failed about 4 years ago due to condensation rusting away the wire connections inside the pin socket for the halogen bulb. I removed it with full intentions of fixing it asap. In the photo of the DRO above you can see that the lamp is still missing.

As a fix I used some surplus LEDs from a 2009 project and 3D printed up a ring light holder along with a box to hold the LED driver also from that same project. It's held in place by friction but unfortunately the larger chuck and tooling creates a shadow exactly where I didn't want one. That was September 2016 so I've been using it for 3 years.
LEDLightTrial1.jpg

Last month, after turning a shouldered adapter ring I finally got around to fitting in a diode, capacitor and LM7812T regulator to bring the 24VAC lamp power down to 12V without a lot of heat on a secondary mounting plate in the lamp base. A new color coded cable up into the lamp arm and now I have a proper work light again.
LEDWorkLight.jpg

Project #42 complete. Only 42 more to go.
 

Attachments

  • LampStuff2.jpg
    LampStuff2.jpg
    111.9 KB · Views: 0
  • LEDHolder.jpg
    LEDHolder.jpg
    89.5 KB · Views: 0
  • LEDLightTrial2.jpg
    LEDLightTrial2.jpg
    380.1 KB · Views: 0
That’s all bloody awesome. Did you say you had the axis running at 100+ ipm? What motor to lead screw ratio are you using?

When my light failed I put an ikea lamp on the swivel arm. You circular spindle led is a good idea I want one now for my mill.

What is the white ring material you mounted the leds in?
 
Last edited:
I think Alexander Kevin d and myself will want to see this mill of yours. We all owned the same model with a Cnc conversion. You’ve done an excellent job.
 
That’s all bloody awesome. Did you say you had the axis running at 100+ ipm? What motor to lead screw ratio are you using?

When my light failed I put an ikea lamp on the swivel arm. You circular spindle led is a good idea I want one now for my mill.

What is the white ring material you mounted the leds in?

The lead screw is 5 TPI (0.2" per rev). The X axis is 3:1 reduction while the heavier Y is 4:1 so only runs 120 ipm. But then it's shorter distances. The motors are: KL34-180-90 brushed DC. They were ordered at the same time and were supposed to be identical but it turns out 10 years later that the winding resistance between the two is different and the X doesn't draw as much current. The Y axis motor warms up more. So I think the X has less torque but that doesn't seem to be an issue.

Anyway. Supposed to be constant torque 226 oz-in (1.6 N.m) peak torque 1125 oz-in (8.0 N.m). With 90VDC and 7.8A peak 40A a Gecko drive (I have two in a box) doesn't have the voltage/current spec. So I went with the HP_UHU kit. Then addedHenrik Olsson's dsPIC upgrade module to get rid of erroneous encoder noise.

More on the motor side of things in a bit. I've been dealing directly with a factory in China for AC Servos after seeing how nice my Harmonic Drive works with the STMBL Servo drive kit.

Clear 3D printed filament. Comes out kind of white.
 
I think Alexander Kevin d and myself will want to see this mill of yours. We all owned the same model with a Cnc conversion. You’ve done an excellent job.
Thank you. I just wish I could get it all done faster. Shop is always open for tours. I try not to have the mill dysfunctional for any length of time.
 
Here's a photo of the mill as it is today. The machine cart still needs side panels. The wiring still needs to be cleaned up and I need to take remove the power feed which is laying on the X axis table. The Ethernet MESA card isn't plugged in at the moment. Just straight parallel port.
And yes, I took the side panel from the spindle pulley cover over to the local paint shop and had them color match the paint so the cart matches the mill. That's the anal part of me at work.
 

Attachments

  • MillSystem.jpg
    MillSystem.jpg
    261.4 KB · Views: 0
Speaking of motors. While I was looking around for a 3 phase 2HP motor (1.5kW) I realized that an AC Servo, 1.8kW and 3000 RPM might be a nice replacement. A bit more searching and I came up with Bergerda in China. I bought two motors and drives.

Although initially I'll be running the spindle with 0-10V from PWM so I can use the single parallel port for both LinuxCNC and MACH3 in the long run I'll be using step/dir/enable to control the spindle. The drive takes either along with RS485 although I'm still not sure about how resistant to noise it is for that.

In addition to the Spindle Motor I also bought an AC Servo with the same specifications as the DC Brushed motor. It too was cheaper than ordering a replacement DC motor and encoder from the US to replace the one that draws too little current.
D8X_5544-TwoServos.jpg
The one on the right with the 10 year old noisy US Digital Encoder is the DC Servo for the X axis. I created a second casting and machined it to fit the Bergerda AC Servo.
D8X_5536-AC_ServoMount.jpg

Here's the AC servo executing some G-Code commands to move slowly in one direction and return quickly in the other. No lost steps or position. Very quiet. I had to configure it to pretend to have only a 250 line encoder rather than the 2500 that it comes with so that the LinuxCNC motion was the same between the two.


The step rate for the full 2500 line (10,000) pulses per rev would be too high for the parallel port. Both the MESA and SmoothStepper can do that. But it also needs to be differential pair step pulses in stead of just TTL at 500KHz stepping. To be able to do that I needed a translator board. One that would work for the STMBL Harmonic Drive A Axis and also for Bergerda AC Servos if I change over.

StepDir2STMBL.jpg

This little board takes TTL Step/Dir/Enable in and presents an open collector FAULT out. Power from 24V in for the regulator to 5V and on the RH side either the RJ-45 connector for a short Ethernet Patch Cable to the STMBL drive or the Molex 0.1" header to a DB-25 for the Bergerda. There are two jumpers that select either all RS422 for the STMBL or RS422 for Step and Open Collector for DIR/ENABLE. The Fault signal is on a separate screw terminal block to be paralleled with the others. And there are two small 2 pin headers that provide an open collector to ground signal whenever enable is FALSE. ie. this is used to reset the HP_UHU drives when they have a fault. The STMBL resets when ENABLE is brought FALSE as does the Bergerda.

For now I'm going to stay with the DC Servos because I've put all the work into installing the controllers etc. But if I can justify shipping costs by buying a bunch more then I might just retrofit everything. Closed loop is so much nicer than open loop.
John
 
I was discussing 3D printing for the workshop with a friend the other day. The question came up about when to 3D print it and when to use CNC to mill it from metal. Ultimately the answer is probably "It depends". The limit switches on my Milling Machine Conversion are mounted onto what were sections of card guide holders from old STD Bus computer cabinets. As such they fit the small surplus DEC Printer (also surplus) limit switches.

But for the Knee home/limit switch I needed something more than just adjustment in one direction and the metal bar was too narrow. So I milled some of it and 3D printed the other part.

Not liquid tight but Normally Closed so a broken wire will create a fault.
 

Attachments

  • ZAxisSwitchInstalled-1.jpg
    ZAxisSwitchInstalled-1.jpg
    445.5 KB · Views: 0
  • ZAxisSwitchMounted.jpg
    ZAxisSwitchMounted.jpg
    283.6 KB · Views: 0
Now that I have the CNC part working reasonably well it was time to remove the power drive permanently. The same casting as the LH side was used and bored for a 37mm ball bearing. The shaft just floats inside it as it's really just there to keep the shaft from whipping while turning.
RH_BearingHolderS.jpg

I didn't want to cut the ACME shaft off just in case I might want to go back to a fully manual mill. When I order ball screws for the conversion I'll get them sized so the shaft cover isn't needed. Meanwhile I 3D printed a cover for it.
RH_BearingCoverModel.jpg
Afterwards I used the lathe to shorten the inner bearing clamp and then 3D printed an adjustable extension that slides onto it so I can glue it in place to press on the outer race to hold it in place. Just so it doesn't walk out when I least expect it.
RH_ShaftCoverS.jpg
But for now another project is complete. I can remove the power wires from the power feed and set it in a box for whenever.

Still need to counter sink the socket head screws so more than one thread goes into the casting. Minor detail. The two bolts are much longer and there's no axial thrust on these anyway.
 
No progress on the conversion other than the control cabinet has been wheeled into the woodworking shop area and lifted, with the help of my son, on the table saw so the electronics are all at eye height. Now to go through the list of little things like wire carriers and cabling etc. Stuff that's no fun when crouching on my knees.

Since my son was visiting for Christmas I used him to also remove and replace the X table on the mill. I'm sure it weighed over 200 lbs. I wanted it off so I could photograph the leadscrew and see what sort of oil passes were there. I have a single shot oiler pump that has been sitting in the box for probably 10 years.

Kind of disappointed at how the channels are carved. It uses a drip over the edge method to lubricate the dovetails and the tapered gib. Which means the far end doesn't get a lot of oil. Going to have to figure out how to plug it, drill an oil hole down into the dovetail and a passage way plus slots into the tapered gib.

The project list keeps getting longer rather than shorter.
 

Attachments

  • XAxisBedway.jpg
    XAxisBedway.jpg
    47.4 KB · Views: 1
So a quick update. I can now run the mill via Parallel port with both LinuxCNC and MACH3 and drive the Bergerda AC Servo I bought for it. It's a long story but my MESA 7i92H Ethernet interface cannot do STEP/DIR on Parallel Port #1 and I don't yet have a BoB for PP2. It can do PWM/DIR on pins 14 and 16 on PP1 as can MACH3. In fact MACH 3 can also do STEP/DIR on PP1.

So I created a module that accepts ENABLE/STEP(PWM)/DIR on a 4 pin connector and optically isolates the signals. It outputs the PWM to a small Chinese PWM to 0-10V analog output. It also translates the TTL signal into RS422 for high speed STEP/DIR out to the Bergerda AC Servo Drive. Here's a close up of the module.
PWMSpindle-1.jpg

And here's it all connected and from either LinuxCNC or MACH3 I can run, using PWM, from about 200RPM to 2950RPM. With STEP/DIR I can run 0 RPM all the way to 3000 RPM which will work nicely for power tapping.
PWMSpindle-2.jpg

The motor is 1.8kW. It´s quite a bit smaller than the 2HP (1.5kW) single phase motor that came with the mill. Once I cast the new bases for the motor mounts I'll post the results. Hopefully no more belt changes.
 
PeterT mentioned that my mill would be a good VFD candidate. I like the idea of no more changing belts but also am not ready for CNC. Can the servo controller be set to a "manual" mode using a switch for forward/stop/reverse and pot for speed adjust? Does the controller have a tach output? A .5hp or .75hp version might work well on the lathe as well.
 
PeterT mentioned that my mill would be a good VFD candidate. I like the idea of no more changing belts but also am not ready for CNC. Can the servo controller be set to a "manual" mode using a switch for forward/stop/reverse and pot for speed adjust? Does the controller have a tach output? A .5hp or .75hp version might work well on the lathe as well.

This AC Servo motor can be configured to run with 0V-10V. My initial testing was with a pot and 10V supply. I varied the pot and the motor speed varied. A different input is used for direction. There's also an enable input that when off allows the motor to free wheel.

The encoder on the motor is 2500 lines and in quadrature that's 10,000 pulses per rev. However the controller has a pair of scaling parameters that can take that value and output that scaled value out to external hardware. So you could for example divide by 250 and multiply by 30 to get (10,000/5000)*30=60 pulses per rev or 3000 pulses per second at 3000 RPM.

If it's turning 240 RPM then that's 4 revs per second * 60 pulses per rev = 240 pulses per second.

I've not tried the scaled output yet. Once you do decide to go CNC you can remain with that configuration and just use a small PWM to 0-10V module like I did there or use step/dir signals and have the CNC software set the speed precisely.

I have a smaller 400W version that I want to use as either the Mill X/Y axis or possibly the Lathe Z axis. I have had it running on the mill.
Pretty boring 1 minute video but shows the speed and precision once you get the parameters right. The nicest part is not needing an external power supply. Runs on 220VAC.

I already have a standard GS-1 three phase VFD on my lathe. But if I like the behavior on the X axis I will probably order a couple more. Shipping from the factory (Alibaba) is much more expensive than the AliEspress slow mail but we could combine an order.
 
If I'm reading this correctly the servo has an on board controller. You connect the main power but speed is controlled by a 0-10v input with fwd/stop/rev controlled by separate inputs to the servo.
 
If I'm reading this correctly the servo has an on board controller. You connect the main power but speed is controlled by a 0-10v input with fwd/stop/rev controlled by separate inputs to the servo.
Yes. Look back at the photo of the motor and pulley and the tan box just behind it. The little PC board takes the PWM, DIR and ENABLE from my PC running either LinuxCNC or MACH3. The tan box takes 220VAC in. The little green connector on the side of the box is teh 0V-10V in. The Tan DB-25 cable runs to the adaptor board I made up that translates the appropriate STEP/DIR/ENABLE signals and the black DB-25 cable runs to the motor encoder.

In this case the tan box has a cooling fan and the motor is 1.8kW. I can move the green connector and the DB-25 to a smaller box for the 400W motor and as long as the Servo config parameters are the same it works identically. Configuration is done with buttons and LEDs on the front panel. Once you are happy with the setup you can save to EEROM and then after that on power up it runs that set.

The Servo drive also can run with MODBus but I did have troubles with electrical noise but that was sitting on the table saw 10 feet from the PC in the other room Don't know how much of that was my wiring.
 
Back
Top