Quality of cut

[richards]

After nine months of using my alpha, I'm still having concerns about cut quality. Yesterday, I spent most of the day tuning and testing the shopbot to find the best combination of move speed, cut depth and RPM. After a lot of testing, it looks like the factory ramp settings gave the best quality with version 3.4.1 software. Also, speeds of 1-ips or 5-6 ips give almost identical quality (3/4-inch MDF, 1-flute straight 3/8-inch cutter, Porter-Cable router 16,000 - 19,000 RPM).

The photos show the cut quality.

Image 1 (1-ips, 16,000 RPM - Only CONVEX portion is important):
4926

Image 2 (5-ips, 19,000 RPM - Only CONVEX portion is important:
4927

Image 3 (1-ips, 16,000 RPM - 45 degree):
4928

Image 4 (5.5 ips, 19,000 RPM - 45 degree):
4929

The tests were repeated at various sections of the table giving similar results. All bearings/rollers are tight - too tight to turn by hand.

It's probably not clear in the photos, but the CONVEX sections in Images 1 & 2 are almost perfect. The 1-ips 45-degree cut is rougher than the 5.5-ips cut (except in the part where the machine is ramping). By the way, the 45-degree cuts were made using 3-inch ramps, so the sections shown were basically at depth when the actual cut was made. The 3/8-inch 1-flute straight cutter used gives similar results to the 3/-8-inch spiral downcut cutter that I often use. The lighting for the photos was set to emphasize the problem. In reality, one or two swipes with 150-grit sandpaper usually removes the 'ridges'.

The two questions that I have are:

1. Are you getting similar cut quality with your alpha.

2. Would a spindle significantly improve the cut quality?

[richards]

Sorry guys, but I'll keep asking until someone responds. Just a little help please to let me know whether 'this is as good as it gets' or not.

[evan]

I've got a PRT upgraded to ALPHA w/ a PC router.
Your cuts look about the same as what I've been getting. However I've noticed that my cuts are not as good as they use to be because of wear on the PC Router. I know that the spindles have vertually no run out, where as the PC router doesn't have that high of tolerances. That alone would imply a spindle will give you a better cut. However some of that chatter may also be eliminated by stiffening the table. I plan on upgrading to a spindle within the next couple of months. There are basicly two reasons I'm planning to upgade:
1. SPEED
2. improved cut and less sanding.

By the way I cut alot of RICHLITE wich is very dense, hard and heavy so it puts alot of wear on both bits and router.

Hope this helps.

[paco]

Hey Mike!

I've made similar observations with my tool that was equiped with a PC 7518... I improved some with slower finishing pass... and since about a few days, I'm experimenting this REALLY GREAT (so far) 5625-20 Milwaukee!!! VERY MUCH less chattering/vibration on heavy loads; I bought a dial indicator along this buy and check both the PC 7518 and the 5625-20 Milwaukee... the milwaukee had about 0.001" runout MAXIMUM and the PC 7518 may have about 2 to 3 times (0.002-0.003") as the 5625-20 on the chuck... this to say that I think that it might be more about a balancing of the whole motor thing that is the cause of the poor results that I (and others at it seem) got from the PC... In fact, just by running the PC 7518 unloaded one can feel the vibrations while the 5625-20 is free of this little dance even up to 21 000!. I've check both tools with and without a bit in the collet; one test was with a bit in the collet and the other was to check the chuck itself... the PC collet added some more runout while the Milwaukee collet did'nt... the 5625-20 collet are MUCH better designed looking more like ER spindle collet (without snap ring to keep collet and nut assembly together) made of only two pieces and not openned. This Milwaukee is DEFINITLY LESS NOISY with MUCH more speed setting available that the user can dial while the router is actually running! It does'nt heat up at the chuck as the PC 7518 do.

The only drawbacks that I have found so far is that it is a little more expensive (though having a FIVE YEAR warranty rather than one) and that it is not as "plug and play" as the PC 7518; it need some non-invasive modifications to fit readily the ShopBot PC holder.

That said, I do think for sure that a spindle would be the number one choice for close to perfect cut... if you can afford it!

[davidallen]

have you tried making the same cuts by hand, using the same router and bit? that might help isolate the problem to either the router or 'bot.

da

[rjguinn@optonline.net]

Mike,
I installed a second Y motor on my older PR & noticed a significant improvement in cut quality.
There are virtually no tooling (chatter) marks. I read that others have done the same to various SB models. I've always suspected that 1 Y motor was the machines' biggest fault. Everything is torquing to 1 side until the slop is removed from the rollers on the side opposite the motor. This action is worsened during ramping & directional changes.Try separate straight X & Y ,diagonal & circular cuts to isolate the problem using the default ramp values. I hope this helps.
Jeff

[richards]

Thanks to all who have responded so far.

I've verified that the problem is not caused by this particular Porter-Cable 7518 by trying other brands right after I got the alpha. A DeWalt, Bosch, and 2nd Porter-Cable all produced similar results with the Bosch being the best (but not significantly).

The problem is almost identical on both the X-axis and the Y-axis. The photos have been cropped to exclude the straight-line portions of the test; however, the jitter for each axis individually is very similar to the 45-degree cut.

The cuts were relatively light, 3/8-inch deep, with a 3-inch ramp going into the cut. All parts of the cuts, including ramped portions had about the same amount of jitter. On the smoother cuts, where the move speed and the RPM seemed to match, the jitter extends about 1-1/2 inches from a 'direction change' and then smooths out until about 1-1/2 inches before the next 'direction change'. However, playing with the VR command to change the ramp values always produced MORE jitter than I got with the default factory settings.

On most cuts, the amount of jitter isn't very important because it is easily sanded away. However, I've been asked to make a variety of jigs in MDF that need to be practically perfect where the variation caused by sanding away the jitter marks would make the jigs useless.

Perhaps I've reached the limits of the machine/router/cutter/hold-down combination. If that's the case, then I'll stop trying to get a better cut. Every machine, whether it's a cabinet saw, a drill press or a Shopbot, has a certain amount of 'tolerance' inherent in the machine's design. Knowing IF the 'tolerance' level has been reached is probably the root of my tests and of my original post.

[sawkerf]

Check your turn-buckles. Mine were way too tight.
I loosened them and correctly adjusted them. Now my cuts are far better.Kip

[bleeth]

Mike: If the design of your parts allows it set up for a clean-up pass and see how that works. Also, unless your dust vacuum is doing a great job, you may be getting a bit of deflection from the dust buildup. Try an upcut spiral climbing on the object side of your cut.

Dave

[fleinbach]

Last year when I receive my ALPHAPRT120 I needed to cut a piece of MDF into 6 inch strips. In the process of doing this I wound up with two piles of strips. One pile contains strips measuring 5 7/8 inch while the other pile measured 6 1/8" I wanted to find out what went wrong. For the next two weeks I did some rather extensive testing to find a weak link. It didn't take long to rule out software.

Next I felt it would be best to look at all mechanical parts of the ShopBot for looseness or free play. In the end after many hours of testing I feel that I have determined the weakest link. That being the Y-car attachment to the X-car. I believe the most positive solution to the problem is described above by Jeff. That is adding a second Y motor to reduce the flexing of the Y car. There is one lesser expensive treatment that I believe could minimize this flexing but adding a second Y motor would probably be the ultimate solution.

[gerald_d]

Frank, to test your theory, why don't you lock both sides of your y-car to the gantry and then take a cut in the x-direction? Before I would add a second motor, I would investigate a spring-tensioned "hold-down roller" on the other side to emulate the hold-down effect of the motor on the first side.

Has anyone noticed a difference in cut quality when the y-car is in the middle of the gantry as opposed to when it is near the ends? The reason being that the gantry will have more flex in the center compared to the ends.

[richards]

Problem solved. Kip's advice about the motor tensioning springs (turnbuckles) was one thing that I hadn't tested. Mine were TIGHT. I loosened them totally, tightened them until the spring was barely under tension, and then tightened them 2-1/2 turns more. The cut quality improved dramatically!

Next, I played with the VR settings again. The default Move Ramp Speed was .4, I varied that to .6, .8, and 1.0. For straight x or y axis cuts, the .6 setting gave excellent quality with little or no difference at .8 and 1.0. On the 45-degree diagonal cuts, the .6 was a little rough but .65 was almost perfect.

My specs are 3/4-inch MDF, 3/8-inch 1-flute straight cutter, 5.5 ips, 0.375 inch depth. As a sanity check, I ran at 1.0 ips and got nothing better than I had at 5.5 ips. (The Porter-Cable starts to slow down at higher speeds and/or greater depths).

An unexpected surprise was that the cleanest cut was the straight Y-axis cut.

Bottom line. The alpha is running better than ever with only a slight trade-off in speed due to the increased Move Ramp Distance setting.

Y-Axis:
4930

X & Diag:
4931

[richards]

My left-handedness is showing again. The VR variable that I changed was Move Ramp Distance.

[fleinbach]

Mike

Great to see you've made some progress. I believe the reason you're cuts in the Y direction are better have to do with what I described above and below. The Y-car is pulled askew by the cutter in the Y direction so it won't be noticed when moving in that direction. Most error due to this Y-car deflection affects cuts that are primarily moving in the X direction.

Gerald
I sorta did what you suggested but in reverse. I clamped a homemade gauge(I don't own a dial indicator) on to the X-car to measure the deflection of the Y-car on the opposite side of the motor. When cutting in the X direction the cutting forces pulled the Y-car askew sometimes over 1/8 of an inch. I made the test with both 1/4 inch and 1/2 inch bits at varying depths from 1/8 inch to 1 inch in a single pass. Under 1/8 inch there was just a slightly detectable off-center track. Once you cut deeper than 1/4" the bit tracted off-center by over 1/8 inch. Using the same parameters both the 1/4 inch and 1/2 inch traveled off-track by the same amount.

I will be making a modification to my Y-car to compensate for this. This modification will certainly be less expensive than adding a second Y-motor but probably not as effective. I believe my modification could cut the movement in half while a second Y-motor would probably stop it completely.

[richards]

Frank,

I've not had to worry too much about having the shopbot 'torqued' too much by the forces involved in making a heavy cut, since I'm limited to making shallow cuts by the Porter-Cable; but, I've worried about the problems I might face when I finally get a spindle. Reading about those of you who can cut all the way through a sheet of material at high speed makes me envious, but then knowing that it comes with it's own set of problems makes me hesitate about throwing more money into the machine before I've learned how to deal with today's little problems.

[gerald_d]

Here (http://www.scapenotes.com/notes/messages/2020/4760.html) is the result of a back-to-back test this morning using ShopBot control software (DOS) and drivers (pre-Alpha) versus Mach2 software and Gecko drivers. The straight cut is at about a 20 degree angle across the table.

I won't bore you with details of cutter type, router speed, material, yadda, yadda because that was all identical from one test to the next.

During the 80 mm/s move speed test, one y-car roller was bouncing on the gantry rails due to lack of hold-down rollers while cutting the radiused corner.

[paco]

Gerald,

goods pics. What are your appreciation of Gecko/Mach so far?... from thoses tests.

[benchmark]

Gerald,

What is the stepping rate of the Gecko's set at.

Paul

[gerald_d]

After spending only 20 minutes with Gecko/Mach connected to the existing table, setting ramps, directions, "unit values", etc, and then doing those two cuts only, I would say that Gecko/Mach is better, but it is not the total cure. The operating requires a lot of mouse work for the computer and I will have to do a lot of customising before it is friendly for a high-speed action, lots of sawdust operation.

The stepping rate was the bottom end 25kHz, but for 80 mm/sec with a "unit value" of about 26 steps per mm (I have no gearboxes), the 25kHz is more than adequate.

[dingwall]

Paco, "The only drawbacks that I have found so far is that it is a little more expensive (though having a FIVE YEAR warranty rather than one) and that it is not as "plug and play" as the PC 7518; it need some non-invasive modifications to fit readily the ShopBot PC holder."

What were these modifications you mentioned? What's the diameter of the Milwaukee body?

[paco]

Hummm...

I'm really curious about how it goes for your setup Gerald; keep us inform... do you use the SB motors you had previously just without the gearbox?

Sheldon,

one need to remove the adjustment screw (easy) and I'd suggest to free the cord protector from the body (just slide it on the cord until warranty is over... even more easy). The 5625 body is about 0.040" smaller than the PC 7518; shiming is needed for the provided SB PC mounting. Docs for specs and details are available to download at the MK WWW site... come back if you need more help; I think I'll will post the instructions I sent to otherS here since many seem to look to go with the 5625... I've check with some who mail me for infos and they are still happy with the tool...

I'd just sent my MK 5625 for a minor issue, contacting MK service/support by e-mail, and they are VERY good so far!

[gerald_d]

Paco, the first few PRT's had direct-drive motors - we have one of those.

[richards]

Gerald,
You saved me a great deal of time by running those Mach2/Gecko tests. I was thinking of temporarily kludging up my alpha tomorrow and driving the alpha's existing motors directly through the PMDX breakout board using Mach2 software. The purpose, of course, would be to eliminate the existing controller board as a major cause of the 'jitters' shown in your photos.

As an aside to those who may be wondering what all the excitement is about Geckos and Mach1/2/3 software, G-code, etc., some of us, at least me, have an application in mind that just doesn't fit, hand in glove, with the Shopbot. Mine is a horizontal boring machine with capabilities to do more than drill 8mm holes. Rather than ruining a perfectly good Alpha, I thought that I would try another approach. So far the Mach2 software, Gecko drives, PMDX breakout box, PMDX power supply interface board, toroid transformer, existing spare 34 size stepper motors, etc., etc., etc., have been an avenue to explore interesting possiblities. However, in my case, the possibility of the 'other' electronics and software actually replacing the Shopbot is almost NIL. My alpha does everything it was advertised as being able to do - and it does it without complaint. I can see the possiblity of building a small custom machine with ball screws and servo motors (which would give much greater precision) for those jobs that need a perfect edge without post-routing hand work.

[bleeth]

Mike: Should you decide to go ahead with your experiment, you may find somewhat different results then Gerald since he is running without holddowns or a spindle or the somewhat stiffer gantry and carriage construction of the alpha. Although my PRT comes in the middle between yours and Geralds, your results would still be very educational.

[richards]

I might try the experiment some time in the future, when I've gained more experience with the Mach software. After spending about thirty minutes looking at the Oriental Motor ASD30E-A driver manual and then looking again at the suggested retail price of the drive/stepper combination and knowing full well that most of my designs work - on the second try after blowing up the first prototype - I think that I'll wait just a bit.

The PH299-03 series stepper that I've had sitting on the shelf for more than ten years are no comparison to the ASM911AA alpha steppers that Shopbot sells. If you compared just basic motors, the newer PK296-F4.5A has just a little less torque than the ASM911AA motor and the PK299-F4.5A has just a little more. Of course, neither PK series motor has an encoder or any kind of feedback circuit like the Alpha - I'm simply comparing speed/torque ratings. If my horizontal boring project works as expected, I'll probably upgrade the steppers for that project from the PH series to the PK series. At that point, it would be fairly easy to mount those motors onto the Alpha machine to run the test. (In other words, I'm chicken. Blowing up a $150 PK motor and/or a $170 G212 driver board is easier on the wallet than frying a $1,200 alpha step motor and driver.)

[gerald_d]

bump

[stevem]

Finally got around to testing.

.75" MDF,4" circle, .1875" 2 flute up cut spiral, 19,000 RPM, 3 IPS.

There is a backlash divot at each quadrant. The spiral marks are clearly visible. Looks like the RPM was too slow.


4932

[gerald_d]

Steve, this is the first time that I see spiral/helical marks. And this gets me excited about what Mach/Gecko can do! Seriously, I like what I see there.


"Quadrant Divots" don't worry me because that is obviously mechanical backlash or flex - stuff that I understand and can chase down.

So far I have only seen vertical (not helical) chatter/scallops that are much coarser than what you are showing. Showing that it is movement of everything except the cutter teeth. Now you are showing something that is clean enough to actually show the cutter teeth marks. I think it is a breakthrough in the understanding of cut quality with ShopBots!

[jhicks]

Just a quick question about all these tests. Where these cuts made in conventional or climb direction? Does it matter? I've seen some pretty nice cut outs with similar spherical edge marks using an O Flute on MDF and plastics generally cutting down .250 per pass trhough 1/2" and 3/4" leaving a final pass web of around .080" at 15,000rpm on a spindle with a move speed of 1.7"/ sec to 2.0"/sec but am always interested in improvement. Seems a lot of testing was done here but is the answer as simple as spring tension being the key variable?

[paco]

Are Bench top tool any bettrer as regarding this issue?!

I observed similar "toolmarks" (as Steve's above pic) from HSS spiral up bits... less with SC ones...

[wemme]

I have an issue with back lash. will modifying the ramps help deal with this?

[stevem]

Wemme, no.