[bevinp]

Dear OP,
Applying the principle of Ockham's Razor (a 13th century philosopher would proposed the most simple answer is the most probable solution), I suggest that GWizard will calculate a high feed rate for 0.001" per flute given th 0.005" DOC. Have you tried a slower feed?
Very few of us would have the knowledge, tools or inclination to go down the path of Fourier Transforms to identify possible harmonic peaks, particularly since having found a consequent peak the solution would involve RPM or feed.

[tmarks11]

So, according to my calculations using the value of 5.4 as explained in the other thread, this would give 5000RPM / 5.4 / 38ipm => 24.4 "ripples per inch" or about 0.040", which is fairly close to the value you estimated.

I was just calculating based upon drawing one line per rotation of the spindle (which results in lines way too close together for what he saw). But I think you could be on to something with the "interference pattern" calculation.

[tmarks11]

Or by direct measurement using an eddy current sensor Eddy-Current Sensors Overview - National Instruments on the spindle with output fed into an A/D (e.g PC. sound card or National Instruments A/D card) ....I also believe in Minimum Effective Action. ....The O'scope time domain display can only show so much info. The FFT transform into the frequency domain of the same signal will give one so much more insight

The problem with such posts here is probably a majority of the audience understands exactly what you are saying, as a bunch of us are engineers. We also understand that you are specifying access to lab tools that aren't found in most garages. Even buying cheap on ebay, you are buying specialized tools that will sit on the shelf forever after a one time use. Most would-be machinists don't have oscilloscopes and function generators on their garage bench (show of hands? Just me and Don and Ray?)

So your proposal really isn't the simplest approach, since the OP doesn't have these tools on hand, he will spend more time and money getting ready to find the answer, instead of actually focusing on the solution. I guarantee that experienced machinists would be able to solve his issue without vibration sensors, oscilloscopes and Fast Fourier Transforms. Just because these things are in your toolkit of solutions doesn't mean that they should be applied to every problem.

Most of the audience here is probably more familiar with the term "inductive sensor" instead of calling it an eddy-current sensor.

I am surprised that you haven't mentioned using a spectrum analyzer and vibration sensors, so you can display the results directly in the frequency domain overlaid with rotational motion of the spindle, without need for hand calculations.

Personally, I would use a $25 phototach, shoot both the spindle and the motor shaft. An oscilloscope is an awkward solution. Measures the spacing with calipers and a 10X magnifying glass (both of which should be in every machinist's toolbox). Turbostep helps on the math, and we are done, saving ourself $500-3000 in unnecessary tooling.

[popspipes]

You are onto something, Don. I think I'd low-pass filter the part by rubbing it with a resiliantly-mounted (i.e. rubber sanding block) piece of abrasive paper. This would tend to remove the higher-frequency components of the surface without disturbing the basic shape.

Randy

I like your approach to the problem Randy!!

[G59]

The problem with such posts here is probably a majority of the audience understands exactly what you are saying, as a bunch of us are engineers. We also understand that you are specifying access to lab tools that aren't found in most garages. Even buying cheap on ebay, you are buying specialized tools that will sit on the shelf forever after a one time use. Most would-be machinists don't have oscilloscopes and function generators on their garage bench (show of hands? Just me and Don and Ray?)

Right on the money.

Well said.

[popspipes]

This looks like a fairly bad case of what was discussed in this thread:
http://www.cnczone.com/forums/tormac...ml#post1140470
At the end of this post I described a simple calculation that should help you determine whether my assumption is indeed correct. If you can't follow my description simply post your feed rate and spindle speed and I'll calculate the number of "ripples" per inch that expect you're seeing.
Step

Edit:
This the kind of finish that you can (sometimes) get after balancing:
Attachment 253564

This thread was a good one, regarding spindle balance, it cured about 90 percent of the problems I had.

[Hirudin]

I happened to run across this article on CNC Cookbook, thought I'd pass it along...
CNC Milling Chatter and Stable Milling Speeds

[David C. Allen]

Instead of finding equipment from the past why not employ things of the present. A "smart" phone may have "free" vibration analysis applications that may be loaded on the phone. The app in the attached image will measure and record X, Y, & Z axis.

Attachment 253628

An yes I have an old Tektronix Oscilloscope

David

[Don Clement]

Right on the money.

Well said.

Nowadays expensive lab equipment is not needed as was the case back in the 90s. What is needed is an A/D card (soundcard) FFT software, PC, and sensor (eddy current). The PC is already there with my Tormach. FFT software is available and inexpensive. Eddy current sensors are cheap or easy to build as a DIY. This technology can predict when machine components are failing before the failure occurs. e.g. bearing failure. Military helicopters are using this technology in real time to predict mechanical failure on a real time basis. I am surprised that CNC companies haven't incorporated this technology as real time maintenance with the eddy current sensor built into the spindle. It's not really about the expense, because there is no real added expense, just the knowledge of what the technology is. Knowledge is power.

Don Clement

[cobrakai]

Somehow I missed the replies in this thread. Thank you all for the input.

I will revisit my g-code and see exactly what my cutting parameters were.

[Hirudin]

This FFT stuff sounds very cool but I'm completely ignorant of it. I don't even know what terms I should search for on Google. Does anyone know of a beginner's website or video?

[aarggh]

Nowadays expensive lab equipment is not needed as was the case back in the 90s. What is needed is an A/D card (soundcard) FFT software, PC, and sensor (eddy current). The PC is already there with my Tormach. FFT software is available and inexpensive. Eddy current sensors are cheap or easy to build as a DIY. This technology can predict when machine components are failing before the failure occurs. e.g. bearing failure. Military helicopters are using this technology in real time to predict mechanical failure on a real time basis. I am surprised that CNC companies haven't incorporated this technology as real time maintenance with the eddy current sensor built into the spindle. It's not really about the expense, because there is no real added expense, just the knowledge of what the technology is. Knowledge is power.

Don Clement

While I did train in electronics early on, and have my beloved and trusty old Tek 465M CRO, along with a mix of HP/Tek MDA's/Logic Anaylsers, etc, in the workshop, i think for most people on these forums it's enough of a struggle just to simply get to grips with the intricacies of gcode, CAD/CAM, and simple machine design and operation. Using FFT wouldn't be on my list of machining troubleshooting options, and I'm sure it wouldn't be on most others. For the very basic machining most people do, on machines averaging from very light-weight to mid-range at best in many cases, a brute force method of varying machining parameters is usually what will work. ANd this is how they'll achieve the experience along the way.

cheers, Ian

[cobrakai]

Okay, here are the exact cutting parameters:
.375" x 1.75" 3FL solid carbide uncoated (yes it's a long one, maybe this is all just harmonics)
.005" WOC
.375" DOC
5000RPM
40IPM
Climb milling

GWizard told me to use 43IPM for a .0007" chip thickness (fine finish pass).

Ripples are .040" apart according to my super accurate eyeballs with a caliper.

Based on what I've learned here, I don't think I should go for a thinner chip because this can cause rubbing. In fact, I might want to up it to a .001" chip, but I will try running it at an overall slower speed. So next time, I will try:
2500RPM
32IPM

As for analyzing machine. I could try seeing how well my spindle and motor is balanced and measure the runout.

[aarggh]

As for analyzing machine. I could try seeing how well my spindle and motor is balanced and measure the runout.

And that my friend could be the start of a very slippery slope depending on how OCD'ish and finicky you are! :-)

cheers, Ian

[popspipes]

I would try a shorter cutter, preferably just long enough for the job. I have never been a fan of long cutters unless its absolutely necessary.

[cobrakai]

Yeah... you're right. I bought it for one particular job where I needed to face a 1.75" long side. Turns out I would have been better getting a reduced shank end mill. Stiffer and equally as capable for that part of the job.

[G59]

Is there any audible squeal from the cutter when doing your finish pass?

[cobrakai]

Actually no, it just hummed right along. I have definitely heard the squeal before, but not this time.

[G59]

Well, then it's not from the cutter flexing, as if it was, those marks would be wider on the bottom than on the top of the part.
Spindle run-out?
I'm just throwing things out there as I have no idea. Like I said, I've seen this before even on my machine which is dc servo driven, so no chance it could be the AC ripple causing it.
Maybe a different endmill with steeper flutes?
I use HSS for in Aluminum, just wondering if this effect can be reproduced no matter what cutter you use?

[BAMCNC.COM]

Climb mill for roughing (it pulls on your ball nut), leave .005-.008" on the wall, it depends on what type of endmill you rough with, size and material. If it's a small HSS endmill leave .008", if roughing with a big Carbide endmill then leave the wall at .005"

Conventional mill for finishing, it pushes back on your ball nut, and will provide a superior final finish, I usually run this at 50% ipm of what it was Roughed at

Get a GOOD Variable helix endmill - takes the harmonics out of it. If in Aluminum, get a ZrN coated endmill - I use Helical Brand for all my Aluminum

Simple advice and this assumes you have your Gibs tight and use a proper way oil