[mcardoso]

I ran my 0.75kW without a resistor for about 3 years. Ran fine, but I kept the accel and decel reasonable.

[maxspongebob]

Take another video. I want to see my pulleys in action. Also, glad to see that it is making improvements.

[steelcogs]

@maxspongebob I'll get some video tomorrow once I can also try out a 3L belt. At just under 5000 RPM it's at about 85dB which isn't bad considering all my measurements before were at 3500 RPM. I believe the motor and pulley by itself was running at 75dB at that same RPM. I'm pretty sure there is too much preload on the bearings and I think it could be creating too much noise because of it. Surprisingly they weren't getting hot though. Going to take the spindle into work to press it out and double check the bearings are ok then re press it in. Then hopefully this time get the preload right.

@rcheli So you have the resistor and line reactor then, correct? Off the topic of spindle noise but I did some messing with my motor today and got the gains I guess what I would consider dialed. However (and this has always been the case), I can't seem to get the correct RPM to output vs. the voltage input. Did you have this problem too when you set yours up? I've confirmed I have the correct voltage coming through and I'm using a 1:1 pulley ratio so the GEAR_NUM should be set to 5000 for 5000 RPM at 10v. I'm wondering if this is a noise issue since I don't have a line reactor.

[rcheli]

I'm running LinuxCNC and the PID on the spindle took care of getting to the exact speed (+- 10~20 rpm). I do remember a setting where LinuxCNC asked what the speed was it at 5 volts (1500rpm) and it based everything on that. My pulleys are 1:1.8. The motor will get up to just shy of 3000 so my spindle makes it to just shy of 5400. I usually run it 5300. I forget what GEAR_NUM I'm using but I'm going to hook the PC back up to the drive and get the info tonight. I really do to get this all recorded and documented.

[rcheli]

Here are the parameters that I am using with my DMM DYN4 1.8kW servo that I am using as a spindle motor. It is geared 1:1.8 to get a target spindle max speed of 5400 RPM.

Attachment 413848

[steelcogs]

@rcheli Those settings are about what I could handle too. I was able to get the speed and main gains to 30 but I might back them down if there is a heat issue. I'm jealous of that linuxcnc feature (among others) since I've been having issues with RPM not being what it should since day 1. Originally it followed the RS232 commanded RPM with the "test movement" area perfectly. Now it's about 60 low of that and 100 low of what's commanded via analog. I always thought the encoder was supposed to help compensate for issues like this. DMM told me to use the analog deadzone settings but they changed literally nothing. May have to send them another email since 0-10v control should be pretty straightforward. I was also thinking you may have a more quiet setup just because your motor only spins at 3000 RPM and mine goes to 5000 so I don't see any reason it wouldn't be louder.

@maxspongebob Ok here are some videos of your pulleys running on a 2L belt. Apparently I ordered too big of a 3L one so I'm gonna stick with 2L for now. One video shows the motor and pulley alone, then the other shows the whole assembly. The motor/pulley are roughly 75dB at ~5000 and the whole thing is roughly 85dB at ~5000. The video makes it sound way more horrendous than it actually is. It's very much more bearable than it used to be and at a higher RPM.

https://drive.google.com/open?id=1UM...Tclzx2Vy4-gKKk

https://drive.google.com/open?id=1Fx...Ol8YoJ0sqFGa7z

All in all I started at over 90dB at 3500 RPM and now I'm at a bearable 85dB at 5000 RPM so It's much better. I did readjust the preload so the spindle is much smoother but it didn't help the noise any. It's tempting to reswap the bearings back to AC again (with proper preload) just to hear the difference but from what I can tell most people seem to say tapered roller are more quiet. That could be wrong though.

[mcardoso]

Got a little more time to run in and test the spindle on my G0704 with a servo / belt drive conversion.

I ran the spindle up to 5000rpm for 90 minutes and here are the (relatively) steady state performance metrics:

Lower bearing temperature: 149*F
Upper bearing temperature: 153*F
Head casting (middle): 143*F
Motor (Top): 158*F
Spindle Nose: 129*F
Tool holder (End of Collet nut): 90*F

Motor running amps: 1.4A (12A = 1800W)
Heat loss into spindle assembly: 208W
Spindle efficiency: 89% (includes all motor, belt, and bearing losses)
Noise Level: 80dBA at 3" microphone 90 degrees to machine, 73dBA at 18" microphone 90 degrees to machine

All in all, spindle runs hotter than I would like but not dangerously so. Noise level is very acceptable and no squealing, ticking, rubbing, or other bad sounds are heard.

[steelcogs]

@mcardoso

I ran mine the other day for over an hour at just under 5000 RPM, can't remember the exact time, and the spindle was MAYBE warm. The servo was pretty damn hot though. I couldn't find my laser temp sensor to check exactly what it was, but I imagine if I tuned the servo a bit better it'd be cooler. It didn't fault out on me so it shouldn't be a major issue. My noise is about the same as I last posted but it's so manageable to listen to compared to where I started. My last thing at this point is getting the RPM to be not 200 off of what it should be. Then I'm moving onto a PDB and enclosure!

[mcardoso]

@steelcogs

Awesome! I'm jealous of your cool temp, but I am not tearing mine down again Servos are meant to get HOT. It shouldn't hurt them and the drive should protect the motor as well. I'm running mine as a Step/Dir spindle and have a 1-2 rpm variation at steady state and maybe 10 rpm when entering a moderate cut (which quickly rebounds to nominal). Are you running yours as an analog servo or a Step/Dir? Just got my PDB done and it is so nice to have. Thinking about an enclosure...

Are you running tapered roller bearings or angular contact?

[steelcogs]

@mcardoso

Yeah I know the drive will throw an alarm if it gets to a dangerous temp. I'm using analog as my controller only accepts analog or PWM and the DYN4 only outputs to analog out of those options. The testing I did I concluded that using the test control constant RPM without a load on the motor it is dead accurate. Using the test control constant with it setup to the spindle the RPM is less than it should be. Then, when connecting it via analog it is even less RPM than it previously was. Interestingly, when I command 5000 RPM though, it jumps way past and goes to ~5500. I'm thinking it's either an EMI issue or just wrong tuning. I emailed DMM but haven't heard back when I usually had them reply pretty quickly. Getting ready to email them again here pretty soon.

I'm also running upgraded tapered roller bearings. I never got the preload right with AC and if I had the patience to take it apart (and re-press the bearings) I'd see what the difference was with noise and heat.

[mcardoso]

@steelcogs

I work with servos and tuning every day, so I'd be happy to help if I can (although DMM's tuning is a tad strange so their comments might be better). The no load test does not surprise me since the load inertia is 0. This gives the servo a huge dynamic range to correct for errors. When you ran the second test using their software to set the speed reference, you found the rpm to be lower than desired. This is 100% where tuning the motor will get it to almost zero error. This would be the first place I started and would work to get a near zero steady state error. Keep a notebook of values that you have tried and how well they worked. From here, I would start playing with some test cuts and seeing the speed drop and rise as you enter and exit a cut.

Once you have the performance where you want it with the software speed reference, you can start using the analog input. This way, any error that shows up once you've switched to analog you know is coming from the signal, not the tuning.

If you haven't done PID tuning before, my plain english explanation is this. The P parameter acts like a spring. The further you are away from your velocity setpoint, the harder the drive pushes to get you there. You can have a P only controller, but it is common to get oscillations (think spring again) or steady state error (once you are pretty close to the setpoint the force driving you to get closer diminishes). Typically you would also want to add in some I gain (integrator). This gain acts slower than the P and gradually increases its contribution to the force as error has existed over time. I gain is good for closing up that steady state error. Think of a slinky where you drop one end from an outstretched arm. Lets say you desire to have it coiled in your hand but it was dropped and now bouncing up and down. The force of the spring is the P gain (and the potential oscillations that come with it) but you can see it will never get back into your hand due to gravity. You would need to add some additional force upwards to correct the error to zero. That is the I gain. Finally you might find that as you change speeds there is some overshoot. This is typical of PI systems and also where the D gain comes in. The D gain looks at the rate of change of your velocity and typically counteracts the other gains. So when the P and I gains want to dramatically accelerate the spindle to a new velocity, a touch of D gain will dampen the acceleration and reduce overshoot (at the expense that it takes longer to reach the new speed). All 3 gains have pros and cons and there is no perfect tuning formula, so set a "good enough" goal and just stop when you get there.

My typical tuning method is to set all gains to Zero. Then increase the P gain to the point of instability and reduce 10-25% (this is done without cutting forces). Then increase the I gain until your steady state error diminishes (this is first done just changing speeds, but also should be tested under cutting loads), finally increase the D gain slightly to reduce overshoot to a reasonable value. I believe the DYN4 also has a low pass filter. This is a good tool to battle resonance and instability and permits you to get higher gains than would be possible without it. Try tuning at first without it (record what you are doing) then increase the low pass filter and start over. DMM has some weird names for the gains, but the principles remain the same.

Once this is good, play with the analog signal.

[steelcogs]

@mcardoso

Thanks for the detailed explanation. I do actually have a little bit of experience PID tuning with racing drones from a couple years ago. However that knowledge is limited to how changing the values on the drone's controller affects how it feels to fly, so you explaining it in reference to servos is helpful. I'll give retuning a go when I get a chance to match the RPM and isolate any error to the analog control.

[vger]

Just my 2 cents....
I worked doing vibration analysis and balancing for many years and found that often times noise came from quite unexpected sources in various machines. One unit I worked on was quite deafening loud when in operation at normal speed, but the noise dropped out as soon as the speed dropped by about 5%. Ended up being a panel on the machine that was resonant at the rotation speed. Stiffening the panel with a bit of steel angle did the trick.

Locating the source of the noise may require a bit more than just an SPL meter. You might look at an instrument that displays a spectrum of the noise. There are some apps that do FFT even on iPhone or iPad. Once you know the frequency of the loudest (highest peak) then you can start deducing the source. Moving the microphone around near different parts of the machine while observing the spectrum may help.

Another method of input to the analyzer would be an accelerometer instead of a microphone. The accelerometer can be placed in contact with the machine measuring the vibration at that point. Places like bearing housings, are a good bet to start. If you find there are high frequencies well above the RPM, there are actually databases of bearings that will tell you the frequencies they generate if any damage is present. Terms like “ball/roller pass frequency” etc.

There are even guidelines to tell you if a bearing is under greased and over greased! Yes, too much grease can cause issues like overheating as well.

Beyond bearings, things like varying belt tension caused by either worn belts or pulleys can cause vibration and noise. Imagine your belt tension is changing be a few pounds at the speed of rotation! A very slightly bent shaft can be the culprit.

And then there is mechanical balance. If it’s out of balance and it’s rotating, it’s gonna vibrate.

Steve

Trapped on a little island off the coast of Cancun.

[ChrisAttebery]

My spindle was pretty noisy when I converted to Hoss' L belt drive. The 3D printed bushing he designed reduced the rattle between the spindle and quill drive but it was still pretty loud IMHO. Last year I swapped the L belt pulleys for a set of Multi V pulleys that I turned. I removed the internal quill drive and attached the spindle pulley directly to the spindle. The spindle pulley has a journal that fits into another sealed 6007 bearing that I installed into a spacer at the top of the head. The upper bearing keeps the spindle from flexing under load since the stock upper bearing on the spindle is 5-6" below the top of the head casting. This setup is much quieter than the original L belt setup I used.

Here's a link to my post with the pulley upgrade if you'd like to look at it.
https://www.cnczone.com/forums/bench...ml#post2154122