Need Help to Improve Surface Finish

[TomS]

Your comment about untidy wiring got my attention.

Cable dress can be important when it comes to reducing magnetic interference. You have a situation where you have conductors carrying high current to your steppers, and conductors that are relatively high impedance signal lines. Magnetic coupling between them can cause problems (ask me how I know). Shielding in the form of copper braid really doesn't address this kind of interference. Good cable dressing, and keeping high current lines separated as far as practically possible from signal lines, is the way to do it. The cable dresing should minimize the area between your drain lines and signal lines. See my illustrations below

Example 1, bad lead dress (lines far apart)
signal line ---------------------------------------------------------------------------



Drain line ---------------------------------------------------------------------------

Example 2, better lead dress
Signal -----------------------------------------------------------------------------
Drain-----------------------------------------------------------------------------

Example 3, Even better lead dress (this is hard to draw so I won't): make a twisted pair using the signal and drain lines. Induced voltages tend to cancel. I've even seen this approach used in high frequency integrated circuits.

Reason: the larger the spacing between the signal and drain lines, the more magnetic flux can pass through the loop that is formed, producing more induced voltage.

Ideally you'd also route your cables so the high current lines are at right angles to the signal lines but that could be difficult to do.

Sorry, I knew my luck would rub off.

You'll get 'er.

Sorry, I knew my luck would rub off.

You'll get 'er.

This is wierd. I clicked Reply on your post and it brought me to Homebrewed's post. Yeah, I'll get it done just frustrating.

Homebrewed - I like your idea about twisting the drain wire and signal wire. Being that I've got some additional wiring ahead of me tomorrow I'll give it a try.

 

[TomS]

My plan for today was to have test data to share. Unfortunately not. I stripped out the unused X and Y Enable - wire and rerouted all of the drain wires. Also added a ground wire from the BoB to the star ground. Ran individual ground wires from the BoB/ESS enclosure and the motor driver enclosure then terminated them at the VFD ground lug. Powered up the system and the Y axis driver fault light is on. Checked all the wiring making sure the termination points were tight then double checked again. Made sure I hadn't crossed a wire and all is good there. Gave up for the day out of utter frustration. Tomorrow I'll swap out the Y driver with the X driver. If the green light comes on then it's the driver. If no green light then it's back to figuring out what changed. Sure would like to be making parts.

I swapped the X and Y driver wires at the BoB. If it was a wiring issue I would have expected the X axis driver fault light to come on. It didn't. The Y axis driver fault light is on so it looks like a bad driver. Looks like I'll be ordering a new driver. The price isn't too bad, about $40 to $45, but it's coming from China. Probably going to take 2 to three weeks to get here.

 

[jbolt]

If I recall you used the DQ860MA drivers for the X & Y? Both of mine eventually went bad. There are rectifier diodes on the back of the PBC that kept failing. I went to the KL-11080 from Automation Technologies. It is similar to the DQ2272 I have on the Z axis.

 

[homebrewed]

This is wierd. I clicked Reply on your post and it brought me to Homebrewed's post. Yeah, I'll get it done just frustrating.

Homebrewed - I like your idea about twisting the drain wire and signal wire. Being that I've got some additional wiring ahead of me tomorrow I'll give it a try.

I've never heard of it causing a problem like you're describing but it's possible for the different steppers to talk to each other as well. If your wiring configuration for the steppers has changed since your initial build, you might want to take a look at it with this in mind. You wouldn't want to bundle all the stepper cables together because that would be worst-case as far as coupling is concerned. Bundling each stepper's wires together should be OK because the currents should -- more or less -- cancel out (i.e., you've got the same current flowing IN as OUT so the mag fields should cancel). You actually end up with a dipole type of magnetic field, which drops off in intensity much more rapidly than the field from a single wire.

I recall a story one of my electrical engineering professors told long ago. He was contacted by a mining company that was having problems with a new 3-phase electric motor they had gotten. The motor was running as though it had insufficient voltage. The company was testing it before they hauled it into the mine, and he noticed they had coiled and stacked the (very long) wire bundles on top of each other. On a hunch he told them to separate the coils: and the motor began to operate correctly. The problem was the mutual inductance between the phases, exacerbated by their close proximity.

 

[spumco]

The problem was the mutual inductance between the phases, exacerbated by their close proximity.

Interesting, and is a parallel to what I was taught in the Navy about extension cords. We were told not to use extension cords or welding cable coiled up but to spread them out to reduce heating. Of course, we didn't want to do that on some big old 00 welding leads when we only needed a couple feet for the job, but you pretty much say Yes, Chief and get on with it. I get my NDE inspectors to do the same thing now when we're doing mag particle inspections running 600 amps through giant cables. You can measure the voltage drop in a 50' cable when it's coiled vs. spread out.

I also was advised by someone here or on CNCZone to twist my VFD conductors inside the conduit to help cancel out noise. Twist the three leads, but do not twist the motor ground with the bundle. Each conductor is looped through an individual ferrite ring right at the VFD end, and then the twist starts. Everything is run inside metallic conduit with ferrite and so far I've had zero VFD noise issues.

Not that that's an issue with Tom's current woes, but I figured I'd mention it anyway since we're swapping electrical pixie stories.

 

[TomS]

Received my new driver today and got the mill up and running. I ran a quick test with the mag base mounted to the column and the DTI on the vise jaw as in the previous test. Needle movement was the same as the first test. Tomorrow I'll run the test as jbolt suggested.

I've been doing a lot of reading while waiting for the replacement driver. There are a few settings in CamBam (my CAM software) and the Mach3 post processor that may need to be tweaked. In CamBam there is a arc fit tolerance and auto arc fitting setting. In the Mach3 post processor I found arc center mode which can be absolute or incremental. There is one absolute setting and four different incremental settings. Default is incremental (C-P1). One CamBam user has been using absolute for years without issue so I'm going to give that a try. Then there is arc to line tolerance. The default is .01. I did find a reference that .01 is the default for working in metric. Couldn't find a reference to the default for inch mode but .01 of an inch is much greater than .01 of a millimeter. Looks like I need to make a change.

For info my X and Y power supplies are set at 60 volts output. I've read that higher is better but how high is too high?

I'll post more test results tomorrow.

 

[spumco]

Good to hear you're making progress. I had the same problem when I switched from Mach to UCCNC; the default settings are for metric units. My machine profile imported correctly, but all the variables not in my profile caused a few pucker moments until I went through every single thing and set it up for proper units.

As far as the driver supply voltage - it's too high when the smoke comes out If you're still using the DQ860MA driver, it's rated to 80 volts. If I recall my internet stepper theory lessons, power is a function of voltage, and torque is a function of current (all else being equal). Increase the voltage, and the power (torque x speed) goes up. The low-speed maximum torque won't change, but the point where it starts dropping off will be at a higher RPM.

The way to increase low speed torque is to raise the max current. The downside to this is motor heating - so just set the driver to the motor rating and be done with it.

(Hopefully someone smarter will chime in and confirm my hillbilly stepper motor hypothesis.)

I suggest you leave them at 60v and get your machine sorted out. Once you have the surface finish problem fixed and you've cleaned up the copper octopus in your enclosure, then you might try bumping the voltage and seeing what feed rates you can achieve in a standard cut before you start losing steps. Increasing the voltage will probably permit you to take more aggressive or faster cuts, assuming your spindle isn't the limiting factor. Again, this assumes your power supplies are capable of maintaining 80v to the drivers at max current draw.

Cant wait to hear how The Jbolt Method (TM) works.

-S

 

[TomS]

Good to hear you're making progress. I had the same problem when I switched from Mach to UCCNC; the default settings are for metric units. My machine profile imported correctly, but all the variables not in my profile caused a few pucker moments until I went through every single thing and set it up for proper units.

As far as the driver supply voltage - it's too high when the smoke comes out If you're still using the DQ860MA driver, it's rated to 80 volts. If I recall my internet stepper theory lessons, power is a function of voltage, and torque is a function of current (all else being equal). Increase the voltage, and the power (torque x speed) goes up. The low-speed maximum torque won't change, but the point where it starts dropping off will be at a higher RPM.

The way to increase low speed torque is to raise the max current. The downside to this is motor heating - so just set the driver to the motor rating and be done with it.

(Hopefully someone smarter will chime in and confirm my hillbilly stepper motor hypothesis.)

I suggest you leave them at 60v and get your machine sorted out. Once you have the surface finish problem fixed and you've cleaned up the copper octopus in your enclosure, then you might try bumping the voltage and seeing what feed rates you can achieve in a standard cut before you start losing steps. Increasing the voltage will probably permit you to take more aggressive or faster cuts, assuming your spindle isn't the limiting factor. Again, this assumes your power supplies are capable of maintaining 80v to the drivers at max current draw.

Cant wait to hear how The Jbolt Method (TM) works.

-S

I'm taking it a step at a time. I want to specifically identify the root cause so I can pass it along to others. The "copper octopus" is gone. I ran the test today to see if the wiring changes had an effect. They didn't so now it's on to the next test.

Thanks

 

[jbolt]

Here is what kept failing on my DQ860MA drivers. Exploding diodes! The diodes are only a couple of bucks and easy enough to fix but after the third one I gave up. Wantai to their credit offered to fix them under warranty but failed to understand that the shipping charges back and forth to China would exceed the replacement cost. They did give me the US spec for a replacement diode.

 

[homebrewed]

Here is what kept failing on my DQ860MA drivers. Exploding diodes! The diodes are only a couple of bucks and easy enough to fix but after the third one I gave up. Wantai to their credit offered to fix them under warranty but failed to understand that the shipping charges back and forth to China would exceed the replacement cost. They did give me the US spec for a replacement diode.

View attachment 239887 View attachment 239888

If the diodes are blowing themselves off the board they are severely overstressed! The important parameters are the diode type (speed rating), breakdown voltage and current handling capability. I found a teardown log here: http://blog.bouni.de/blog/2015/02/03/taking-a-dq860ma-stepper-driver-apart/ that provides a link to the diode data sheet. The diodes are "ultra-fast" type with 2 amp/50 amp average/peak, 280/400V RMS/peak ratings. The key to finding a US replacement is the speed -- they MUST be fast-recovery for a PWM application. Of course, you want the current and voltage ratings to be equal to or better than the OEM.

BTW from the same teardown log I note there are some indications that the driver manufacturer may be using counterfeit chips -- the sanded-off surface of the controller IC either says it's counterfeit or the manufacturer did not want their board to be reverse-engineered. If it IS counterfeit, they also may be using diodes that were pulled from other boards. If so, they may or may not be fast-recovery. That could explain the early failure of the diodes.