Alpha Report - Almost NO Chatter Marks

[richards]

I just finished gearing my Alpha 3:1 with a home-made belt-driven gear box on the X and Y axes. Before installing the gear box, I ran a text having circles, perpendicular lines, and lines at 45-degrees, 30-degrees and 60-degrees. As expected, the unmodified Alpha had the usual amount of 'chatter'. The same test run on the modified Alpha had almost NO 'chatter'.

Other than installing the gear box, all I had to do was modify the Unit Values. Nothing else! The Alpha still jogged at 25-ips, but I quickly cut that back to 15-ips, since my gear box is just a prototype and I didn't want to shake it apart on the first day.

With this one simple modification, I think that I can live with the Alpha and the Shopbot software without going the Gecko/Mach 3 route.

There's not time to take pictures of the test today. I'll try to have some to post by Monday.

[lto]

This sounds very interesting, I'm hoping you will find it to be the solution to your chatter problem.

Keep us posted; there's alot of us out there that face the same challenges.

Hopefully someone will market an upgrade kit.

[evan]

Mike,
Thank you for sharing with us. I like this path for an upgrade, it seems much more straight forward. I'm looking forward to seeing your test results on Monday.

[richards]

This image says it all:


7702

The cut on the top is with the 3:1 gearbox. The cut on the bottom is plain old Alpha. The general roughness to both cuts mirrors the state of the V-rails, which means that I either have to have them ground or not. Frankly, if it weren't for the extreme angle of the light, you wouldn't be able to see any flaws in the 3:1 cut.

3/8-inch downcut spiral.
4-ips
13,000 RPM

NOTE: The material was 0.75-inch MDF cut to a dept of 0.650-inch. The cut shown is a 45-degree segment. The line is the trim router bearing mark made after the cut was finished. I left the 0.10 of material to make sure that the vacuum would hold the material firmly in place.

[joe]

Congratulation Mike,

It looks from here you've sure made a good improvement in the Alpha. Why couldn't Shopbot do all of this for us?

Keep up the good work.

Best wishes,
Joe

[pfulghum]

Mike,

I hate to rain on your parade, but my cuts with stock alpha seem to look very close to your top photo. I will call you tomorrow to compare setups and the like. After we have our system configured exactly with exactly the same cut file, I would like to compare pictures. What time would be good to call? It will have to be after 4pm.

-- pat

[mikejohn]

I have also reported no chatter marks on a 'new type' PRT.
The fact that some (a few?) of us get no chatter is interesting. I just wonder why this is.
I am pretty certain my Shopbot is not set up any better than any one elses.
Am I doing something different that I don't recognise?
If someone can devise a series of tests, I am willing to try them out, but I am using a router, so I can't cut very deep.

...............Mike (the other one)

[richards]

Joe,
What an individual does is often much quicker than what a company does. After all, companies, by their very nature, have to be responsible for a lot of things to a lot of people, while an individual can try to solve one problem at a time.

Pat,
On Sundays, I'm usually finished with Church after 4:00 pm. You can almost always reach me at 801-254-3900 or 801-712-1210. (Some people don't like to give out telephone numbers, but I've never had enough telephone abuse to worry about.) If you'd like, I can send you the SBP file so that you can try things on your own computer.

Mike J.
You slipped in while I was typing. You've made a very important point. It seems that different machines give different results. If I were a pessimist, I would blame quality control at Shopbot, but, in fact, I think that I'm actually a realist. There are so many factors that contribute to the quality of cut, that simply saying that using a gear box will improve everyone's cut quality would lead to false hopes. If you've followed my posts, I think that you'll agree that I've pretty much followed a systematic process of elimination to fix every major flaw. A few things could possibly be directly assigned to design 'flaws' but, on the whole, most of my 'problems' could be attributed to the way that I'm using my machine and to the actual equipment used in my shop.

[jhicks]

Mike, are this cuts run in a climb or conventional direction? Have you compared between the two? I notice that my chatter is usually worse in climb (on the piece/inside cut)especially in softer materials like gator foam. I does improve with speeds slowing down but still produces a better finish on conventional than climb even on MDF.
Is that a 2 flute spiral upcut? Have you tried Super O-Flutes? We like the Super O-Flutes better for smoother cuts but whatever works is the right answer. Good to see you're making progress on this option.

[robtown]

Mike, wood is not as much of a problem as plastic, I can tweak out my 3 year old PRT and get similar results in wood, it's plastic cutting that gives me trouble, and I can usually get good results with a bit of tweaking there. Of course to be fair I run at speeds MUCH slower than I see others here run, volume isn't my business model.

[richards]

Jerry,
The cut is in the conventional direction with 3/8-inch 2-flute downcut spiral cutter (which is what I normally use when cutting melamine coated partical board).

Rob,
I agree that plastics can be a bear. I cut the gear box mounting plates from Derlin at 1-ips with a 1/4-inch upcut spiral cutter. I had to put a pause in the program on every 4th hole to stop the spindle and clean off the debris.

[richards]

I've had several emails basically stating that even the better cut still looks sub-standard. To that, I'll have to agree. I would like to see a better cut, but the 'pattern' that is clearly visible is almost a perfect mirror match of the pattern on the V-rails.

When I first got the machine, I placed a long carpenter's square along the rail and then shown a stong light through the gap between the rail and the blade of the square. What I saw then, and what I still see, is the pattern that is shown in the edge of the part.

Up until now, I've been fighting chatter marks that didn't relate to the V-rollers and the V-rails. Now that chatter is under control, if I still want/need a smoother edge, I can have the rails dressed or I can even replace the rails and install linear guides/blocks. For the time being, I'm just going to enjoy cutting parts without chatter marks.

[richards]

I may have been a little premature in my glowing praise of adding a gearbox to my Alpha. Today, when I tried cutting more gearbox brackets using Delrin, I couldn't cut small circles. My 0.75 diameter circles with almost 1/4-inch out of square, even though my cut speed was only 1-ips. the 2.37-inch pocket for the Alpha's front ridge, even though cut 0.008 inch oversize, didn't even start to fit the stepper. The frustrating part is that I could move the cutter to any point on the table and return to my zero point by giving the J2,0,0 command - and not have any measurable deviation from zero.

Belt tightness was checked, loosened, then tightened again with no measureable difference. Finally, I restored everything to factory configuration and the circles cut properly. Right now I'm a little too tired to think things through properly. If any of you have any idea how my machine could cut out-of-round circles and still return to the zero point, please let me know.

[joe]

Mike, I admire you. Tomorrow, I'm sure you will saddle back up and charge those windmills again. What a guy.

He's on a mission.

J.

[richards]

Perhaps I should have mentioned that the chatter test part is about 12X12 inches. The part cut with the gear box installed was almost exactly the same size as the part cut with the stock Alpha. There may have been about 1/32-inch difference between the two parts. Since the test piece had arcs, perpendicular lines and angles of 30-degrees, 45-degrees and 60-degrees, as well as a hole cut from the middle of the piece, I thought that I had pretty well covered the bases. Obviously, I left something out.

With the gear box installed, I could see definate hesitation in some of the moves. Since the cut speed on the Delrin was only 1-ips, the machine's jerkiness surprised me. (I've rarely cut slower than 5-ips since I installed the spindle and have never noticed the hesitation that I saw last night.) Cuts on an angle (M2,45,45) didn't exhibit that hesitation at all, making me wonder if the higher unit value (3819.xxx) broke something in the arc/circle routine.

[odulfst]

this me be slightly off topic or somewhat related.

my prt-alpha cuts great, even w/ an 18" Z. I think all the tireless alignment paid off. However, quite often i'll be cutting large files (read ~500,000 lines) that i got from mill wizard, and I'll have the cut speeds all set how I want, and it will be ramping fine. But then sometimes it won't slow down before a direction change causing a massive jerk, shaking the machine, and reducing the quality of my part. Any idea what causes this? It's completely inconsistent. I can sit there an watch, and one pass will be just fine, and the next will have a bunch of jerky movements.

Maybe the same crappiness that's causing my problems is also involved in your crappiness.

[dingwall]

Tony, check your code for J3 commands that shouldn't be there. I noticed whenever I get jerky movement, there are short J3 commands between arcs. Sometimes it's one or two, sometimes it's many.

I use Vector so the problem could exclusive to Vector but it might not be.

[paco]

Hey Tony!

I've observed similar behavior when I change the speed on the fly... maybe I should let support know about that...?

Next, I would suggest to make some test with the ramping settings; mostly '3D Ramp Threshold' and 'Slow Corner Speed'. Setting '3D Ramp Threshold' to higher (up to 600) than default (100) will make the 3D moves more aggressives while, say 50, will make 'em very gentle (too much if you ask me, but if you machine aluminum...).
'Slow Corner Speed' seem to affect direction changes on 3D files; check it out with some air cut tests.
How is the rest of your ramping settings? Have you been changing 'em?

[bleeth]

Paco:When you change speed on the fly using the < and > keys the move speed changes but the ramping stays what is was for the original move speed. If you pause the file (S) and insert command and reset move speed then the ramp is adjusted accordingly. This should help your jerkiness. (I assume you are raising the move speed)

Dave

[harold_weber]

Dave, if I look at the attached drawing from the command reference manual,
7703

it sounds like what you are saying is that using the < and > keys will NOT adjust the speed at the end of the acceleration ramp, or the speed at the beginning of the deceleration ramp. It sounds like you are saying that using the > and < keys ONLY adjusts the constant speed portion of the curve called "motion at full move speed". So there can be a discontinuity in speed at the end of the accel ramp, and another at the beginning of the decel ramp. Is that right?

[bleeth]

Harold: What this means is that as you increase move speed the ramp time remains constant. Therefore the ramp angle, as charted, would increase. If you increase speed too much you simply are not giving enough time to change speeds. (Think of burning rubber off the line or screeching to a halt).

Dave

[harold_weber]

Dave, so you are saying that using the < and > keys is going to change the parameter that Ted Hall is calling the RAMP RATE, is that correct?

[Brady Watson]

Tony/Paco/Harold - Try playing with ramping (VR) and change the Slow Corner Speed value to something between 25 and 40. The SCS affects both 2D AND 3D moves and reducing this value (which is a percentage of move speed) will eliminate the 'banging' you are experiencing. The other value to tweak is the Minimum Distance to Check. Default is .15, I found that .08 is a good setting all around for MY machine (which I get into below). I believe that this defines the look-ahead distance for ramping and the values work opposite of how you think they do. Harold - No. < & > just increase/decrease on the fly speed and does not change ramping dynamically. Hit 'S' as Dave points out and <I>nsert command MS. When you resume, ramping will be recalculated, the stack cleared and refreshed with new ramping values for the speeds you changed. (You don't want ramps set for 2 IPS to be run when going 6 IPS)

Mike et al - I have been running an Alpha with 7.2:1 gearboxes on it since April. This is something that I special ordered from ShopBot. It has better mechanical resolution (1527.xx UV) than the standard Alpha and let me tell you...GOBS of torque. You cannot move any axis if you push with all of your might like the standard 1:1 Alpha. I think it has about 150-160lbs of cutting force. The real advantage of course is the fact that you gain mechanical resolution with the gearboxes. This has an amazing effect on smoothness. What used be an embarassment to me when cutting plastic on my PRT, is not an issue. The gearbox version really does a beautiful job on plastics of all types, in addition to all other materials that I cut. The Oriental drivers for the geared steppers are setup in such a way that I can also configure them for ultra-high step rates (IE @ 4,000 UV) with a reduction of top speed. In 'standard' mode, the tool jogs 30 IPS, cuts with full torque up to 12 IPS.

Let me just add a word about all of the upgrade and Mach3 hoopla that has been going on lately. I own Mach and have tested it against the SB3 system. I did not find that it cut any smoother than SB3 using the EXACT SAME hardware. For example: My PRT started out as a standard 1/4 stepping box (like everyone else's) and I wasn't thrilled with the smoothness of it. I tried running Mach3 in combination with Gecko 212 drivers, and 70v power supply and the result was a faster smoother tool. I was pretty exited about this. Just to play fair, the Geckos and 70v PS were plugged into an Alpha main board and the results were pretty much identical to the Mach3 setup I ran. The 2 common denominators here were 1) The Geckos with 2.5 times the resolution, plus increased current handling and 2) The 70v power supply which moved the torque curve up a bit with the stock PRT 3.6:1 motors. From reading some of the reactions on this board I take it that some think that Mach3 is the best thing since sliced bread. It isn't. You can do the same thing with SB3. The Geckos and power supply provide the increase in performance. Let's be clear about that. If you own an Alpha, then Mach3 is going to do nothing for you. Keep your Alpha hardware!

To be fair, ShopBot recognized the need for increased speed in the near past and aswered it with the Alpha tools. Cut quality was better than a PRT to boot. Everyone was crying for speed like baby birds (myself included!) and ShopBot stepped up and answered the call. Lately the cry has been about smoothness. ShopBot recognises this and I am confident that it will also be resolved in the future. ShopBot has been pretty good with implementing improvements requested by their customers in the past, and I see no reason why smoothness wouldn't be addressed by them as well.

-Brady

[lto]

Thanks Brady,

Would you replace all the motors, or only the X and Y?

This upgrade is a significant investment at over $1000.00 per motor, so I've been postponing to see what else developes. Would you upgrade if you had a standard Alpha?

It were nice if we could trade in the standard motors.

[evan]

Brady,
So to get this straight: Are you running a stock Alpha with the addition of 7.2:1 gearboxes, or have you upgraded to different drivers as well? The beefier power supply and Gecko drivers were on your old PRT not Alpha, correct? Who makes the gearboxes? How much? Did you upgrade each axis or just the X & Y? Was mounting the gearboxes fairly straight forward or was there serious modification required? Inquiring minds want to know...

Thanks,
Evan

[Brady Watson]

I bought the entire package directly from ShopBot. I kept my PRT and paid for an additional license so that I could keep it. When I upgraded, I bought a new Alpha gantry and instead of buying the upgrade with standard 1:1 motors, I paid extra for 7.2:1 geared Alpha motors with matching Alpha drivers. These are different drivers than the 1:1 use. The motors look just like the gearboxes on a PRT, but they are Alpha motors. All axes have 7.2:1 on them. You'll have to contact ShopBot for pricing for your tool.

-Brady

[patricktoomey]

I wouldn't expect Mach to cut any different on the same hardware as Mach just delivers the commands to the controller which does the stepper driving. It's the controller that's giving everyone fits due to low step resolution. I think the main interest in alternate controlers is getting increased resolution without having to shell out $4,000 plus for new motors and gear boxes. That amount of money is not practical for a PRT upgrade if the controller can be swapped out for less than $1,800 to get the same or better results.

[Brady Watson]

It's the motor driver that gives resolution, not the controller. The controller hardware can influence the step RATE, but not step resolution, generally speaking. Communication speed also has to be good when increasing resolution, or it will choke at high speeds.

If you have a PRT, then you can go with the botrods setup and get a speed increase with an increase in resolution, or run the Ascension for the same performance with gcode.

This upgrade is NOT the same as an Alpha with 560 Oz motors or the geared version. The PRT upgrade regardless of what software pushes it still runs 305 Oz motors. The increased cutting speeds that these new upgrades offer fulfill what most of us with a PRT have dreamed of for the past 5yrs...A little more speed and a little more resolution. You are not going to cut reliably @ 10 IPS on a hopped up PRT without a whole lot of ramping and good communication speed.

Upgrading a PRT to Mach3 or Alpha board + Geckos is not the equivalent of an Alpha. Hence the price differential. I've got both a hopped up Gecko/Alpha boarded PRT and an Alpha. The Alpha is top dog in the shop.

-Brady

[richards]

Brady,

Your setup is very interesting. Ted and Frank, at Shopbot, have been extremely helpful the last few days. (They've always been helpful, but they really went beyond normal service while I've been playing with a belt-driven gearbox on my Alpha.)

There are two specific items of data that would be helpful right now, before I cut up another $200 piece of derlin for the gearbox:

1. I'm still running stock 20-tooth gears, but 30-tooth were highly recommended for use with a gearbox. What are you running?

2. The stepper drivers on my machine are still set that the factory setting of 1,000 steps per revolution, but it was strongly suggested to change that to 500 steps per second (again because of the gearing.) What is the setting on your drives?

I'm not trying to second guess Ted and Frank, since they know a lot more about the machine than I'll ever know. I've asked the questions simply to know what you're using today on your production machine.

[Brady Watson]

Mike,
1) I run 30T pinions on mine. They have a keyway machined in them to deal with the increase in torque.

2) I run the 500 setting for normal operations & the 1000 setting for super hi-res up to @ 4IPS max(rarely).

-B

[evan]

Brady,
I'm confused... you stated; "I paid extra for 7.2:1 geared Alpha motors with matching Alpha drivers." Are these different Alpha drivers than the ones that come with a standard Alpha? And for that matter are they the same steppers that come with the standard, just with the gearbox attached? And is that the same gearbox that was on the PRT steppers?

Thanks,
Evan

[patricktoomey]

Brady, I'm seeing the difference now, there are so many parts to this puzzle it gets confusing. I have a PRT so I've been thinking in those terms as an Alpha upgrade doesn't make sense for me at this point. My second machine may be an Alpha in a coniguration more like yours. Gotta build an addition on my shop first though.

[richards]

Brady,
Thanks for the quick reply. (Just as I posted, one of my sons called to be rescued from his car problem.)

Based on what you're doing and what Ted and Frank suggested, I think that I'll postpone the 2-stage gearbox until I've tried things again with a 3:1 box. This time, I've had a machine shop bore out the pulleys properly and I've redesigned the housing so that there should be no play of any kind within the gearbox. If I still have problems with circles, I'll set the steps per revolution to 500 instead of 1,000. That should get me to a working machine that can be used while I play with gear ratios and number-of-teeth on the spur gear.

Thanks.

[Brady Watson]

Evan,
Q1 - Yes. Completely different.
Q2 - No. Completely different.
Q3 - No. Alpha specific.

Mike - No problem, glad to help. Let us know when you get it running the way you want it to.

-B

[evan]

Brady,

So what you are saying is you do not have a Shopbot Alpha. The only things your machine has in common with an Alpha are the rails, the gantries, controller (SB3), and the color blue. The drivers are different, the steppers are different, and the pinions are different. As you said,” It's the motor driver that gives resolution, not the controller.” So it would seem to me you have a ShopBot “Something” but not an Alpha. I don’t mean to be nitpicking but you special ordered a custom built machine that is by your account a far superior machine to the Alpha. Congratulations! It sounds like you found away to achieve what the rest of us are working on, smooth chatter free cuts at reasonable speeds.

Evan

[paco]

If you ever get some spare time this weekend (since you got plenty of... 8-D ), get the digit. cam. and show us all what you get from your "custom" Alpha tool. Set it on macro.

May I suggest plastic...

Thanks in advance.

[Brady Watson]

Evan...No. I have the following:

1) A PRTAlpha with 7.2:1 motors operating on Alpha gantry & the LEFT most side of a 5 X 16' ShopBot table. This is a 100% ShopBot purchased product.

2) A PRT with Alpha board, Geckos and 72v PS running stock PRT 3.6:1 motors on modified/welded PRT 5' gantry on the RIGHT side of my 5 X 16' ShopBot table.

Both independent machines share the same rails and table. They are 2 distinct systems & gantries running on the same rails. Future plans include 2 smaller 5X8 tables, one for each.

Paco,
I'll see what I can do. Today is the 1st day in 7 weeks that I have had 'off'.

-Brady

[rhfurniture]

Brady,
Unfortunately Shopbot do not list these enhancement options. I will have to email them for a quote.
Running at slower speeds (1.5" ps move speed and below), how would you judge the cut quality difference between your 7.2:1 alpha and gecko'd prt? I assume you are saying that both of these had significant cut quality improvement over your previous stock 3.6:1 PRT, even at slower speeds.

R.

[Brady Watson]

Both cut very well at slower speeds. Aside from my personal judgement of the quality being 'good enough' from each machine, it is hard to quantify it...Which really seems to be the problem here since identical hardware tried by others often yields different results. Even bone stock PRTs seem to yield different results from machine to machine. At the moment I only have one spindle...so a true side by side cut comparison is out at the moment. (Spindle was run on rodded PRT before going to geared Alpha)

All I know is that as far as the PRT goes, I am pretty happy with it at the moment. All I really wanted to achieve with it the past couple of years is an increase in smoothness and about 1 or 2 IPS more of cutting speed without having to resort to exotic hardware/software. The Alpha board/Gecko combo has achieved this and more.

I got the geared Alpha because I liked the new gantry, which is light and strong and the speed, torque and overall smoothness that it affords. I often get jobs that require me to run 8 to 10 IPS for 12 to 16hrs in order to make deadlines. The Alpha is made to do that. The hot rodded PRT really isn't, but can hold it's own reliably at about 1/2 the speed of the Alpha. So...as a guy that started out running a 'diesel' PRT via DOS with a ceiling of 2.5 to 3 IPS max, I am pretty happy with the PRT at the moment! The geared Alpha is everything I wanted the PRT to be from day one. I don't think that it has gone into 'Alpha mode' ever since the motors have so much torque. You literally CANNOT overpower any axes by pushing with all of your might. You can however do this on a PRT, rodded PRT and standard Alpha without much difficulty.

-Brady

[joe]

Brady,

Now I'm really confused.

For us non technical guys, are you suggesting new motors and drivers? I'm not too technical. I can change out a light bulb but I'm not gonna crack open the control box. All this geko, enteositer, gurd stuff isn't for me. I bet there's lots of us old guys out here that feel the same.

Does Shopbot have a cost effective solution coming? Are you working for them? Are your findings sanctioned by Shopbot? Just what are you suggesting us to do, if we want better performance?

I'd like to see is an imperical test. Strait forward comparisons, photo's if possible, of your solution compared to the trusty old PRT and the Agek and to the Ascension. Do you have access to the Ascension or have you given it a test hop?

I still think the plain old PRT is a great buy, even if it needs a update. IF the PRT comes without the motors and controll box that would be even better. I could get someone to put it all together for happy routing.

J
www.normansignco.com (http://www.normansignco.com)

[jimmya]

Joe, most of your post have been negative against SB. I am sure they see your points!
Jimmy

[Brady Watson]

Joe,
What model ShopBot and control system do you currently run/own? This is a broad topic and clarification would be helpful for all in the discussion.

Let me state clearly that I am neither endorsing nor suggesting what others should do. My posting is simply to SHARE what I have done personally with various systems and what I have found works best for the type of cutting that I do on MY specific machine. ALL of the upgrade options out there (Ascension, AGek, homebrew) are valid, viable and offer an increase in performance and speed over the factory PRT offering, which is now about 6yr old technology. Add state of the art electronics, and you will get an improvement. (Add electronic fuel injection to your carburated truck & you will get an improvement) They all do what they say they will, and a big BRAVO! to ALL of the guys who spent countless hours and dollars to make it happen on their own accord.

I would love to have countless hours to test all of the different combos out there and give you an unbiased comparison of them all...but like you I have to make a living and time is not a luxury I have in surplus ~ Nor do I feel that I have to prove or disprove anyone's setup. I will say that CNC cutting is what I do for a living, and combining my talents with the helpful sharing of knowledge by others, I have pieced together a solution to breathe new life into my 5yr old PRT. One that doesn't require me to run g-code, different part files or change my design software or have to learn a completely new set of commands to do my job. Same basic hardware/electronics as the AGek and Ascension with minor differences. I personally wanted to do this myself rather than buy something already done.

So...let me re-cap in saying that this type of PRT upgrade/conversion is not necessarily for everyone. (and several of us 'geekier ones' on the board have stated this numerous times) I am not sure how tech support works with it and the end-user's ability to troubleshoot a problem efficiently. My PRT is out of warranty and I felt that there was no harm in upgrading it. The Alpha is under warranty and if something goes wrong, I pick up the phone and ShopBot helps me fix the problem. Nobody is going to help my fix my homebrew PRT. I don't know what support structure is in play with other upgrade paths, so I cannot comment on that.

-B

[joe]

Jimmy,

I just re-read my previous posts. Can't say I agree with you, but good to see your posting.

[joe]

Brady,

Thanks for the prompt and very imformative reply.

I'm most supportative of the company. However it's difficult to mention other controll systems, good points, without sounding negative. It should be stated how much we owe Shopbot. The PR and PRT has not only changed out lives, it has changed the industry. Everyone I know wishes her the best. But there needs to be changes. Constructive criticism isn't negative. We want the company to come out on top.

I am still running the old PRT, which I originally purchased new back in 2001. This unit was sold to StClair Cast Stone some time ago and they are letting me use it untill my new unit arrives. We are considering an update for their unit in order to get cleaner and faster cutting. They cut letters, with a v bit, in reverse, to be used in concrete casting. We are needing smoother cuts to reduce the necessity of sanding.

We will be moving into our new building in a few weeks and may look to a second CNC. The way it stands now, that could be another PRT/Ascension. I haven't seen any difficulty learning Mach. That's probably because Alan has it set up for Shopbot. You are correct though, learning Mach from scratch could be tough. To tough for me. I'm not that smart.

J
www.normansignco.com (http://www.normansignco.com)

[richards]

I'm not ready to publish results from my "new" belt driven gearbox, but preliminary results are "great"! Because of the problem with cutting small circles with my first attempt, I going to delay a little while before proclaiming that a 3:1 belt-driven gearbox is perfect for the Alpha. Until then, if you're like me with serious problems with chatter, please be patient. The solution seems to be almost at hand (at least for the standard Alpha model). Testing takes time and I'm old enough to not want to hurry too much any more.

[mikejohn]

This week was the 25th anniversary of the IBM PC.
Without this PC introduction, the world would not be where it is today computerwise.
IBM PC computers no longer exist.
They made the introduction, then we got more, bigger, better and cheaper from others.
Maybe this is the way CNC routing is going.

................Mike

[robtown]

"They made the introduction, then we got more, bigger, better and cheaper from others.
Maybe this is the way CNC routing is going."

Maybe ShopBot is the Dell of the CNC world...

[Brady Watson]

RH,
You won't find the 7.2:1 option listed on the pricelist or options page...It was specially ordered and can probably be considered a prototype of sorts at the moment since I ordered it as a 'custom tool'. I say this in case some of you were thinking of calling and asking about the 7.2 'option' as if it were an off the shelf deal. (probably should have said that at the beginning)

Mike,
Like my Alpha, I think that the 500 setting is right for 'all purpose' cutting and the 1000 setting for super high res cutting. Not sure what the multiplier or UV would be for the 1000 setting off the top of my head though using your gear ratio. Either way, if you do use the 1000 setting, don't expect it to cut very fast...and make sure your comm speed is good.


-B

[rhfurniture]

Brady,
Given my trade, I am looking for the best cut quality possible. Using slow speeds, sharp cutters, and a well tuned & strengthened prt, I have got it pretty d*** good, better than my fixed head pin router was, but not yet nearly as good as I get off a 5" dia block at 6k on my 40 year old woodworking spindle/shaper. I am looking for the next step, and need pricing/improvement assesments as part of that. I can see that yours was a custom machine, but SB would surely be willing to provide it for others, and I am interested in this possibility.

R.

[richards]

Back to the belt-driven gearbox on my Alpha - Things are working very well. I'm using a 3:1 ratio, single-reduction gearbox with 20-tooth, 20-pitch spur gears. I just multiplied the Unit Values by 3X to get the proper move distance with a 3:1 reduction unit.

Speed tests using the M2 command showed absolutely no problems at speeds upto 12-ips (maximum move speed on my machine) in repeated long multi-axis moves at 30-degres, 45-degrees and 60-degrees and then returning to the 0,0 point. The J2 command gave the same repeatability at 20-ips. (The belt-driven gearboxes have significant mass, so I cut the Jog speed from 25-ips to 20-ips; however, I tested jogs all the way up to 30-ips second without any problem.

Cuts were much smoother than the standard Alpha and even smoother than the photo posted earlier in this thread. There are still some minor marks that will require sanding, but cuts are the best that I've seen on any Shopbot so far.

The problem that I had with the original belt-drive design and small circles must have been due to the wobble that I was experiencing with the hacked together gears. On that machine, I bored the gears on my drill press. Even though I thought that I had the ability to bore the holes on a drill press, they ended up being anything but concentric to the gear's center. On this latest design, not only did I bore the holes on a lathe, but I stiffened up the whole assembly considerably. I'm guessing that the belt tension was loose then tight then loose, etc. as the shaft turned, causing the Alpha motor/driver to automatically compensate instead of missing steps. On a small circle, the compensation on multiple axes probably caused the out-of-round condition. I'm just guessing at this, but, because I could always return to the 0,0 point, even though the small circles were anything but round, it makes sense. In any case, the new gearbox has had no problem cutting small circles properly.

To sum things up, I'm getting cuts that are acceptable to my most picky customer. There was no loss of speed (most of my cuts are in the 5 to 6-ips range). Torque is excellent (3X better than the original machine). Accuracy and repeatability are both excellent compared to the original Alpha. The best part of all, to me at least, is that I didn't have to abandon the Shopbot software. (The question of whether Shopbot software or Mach software is better is not important to me. The important point is that PartWizard generates tool paths for Shopbot software but not for G-code; therefore, my preference is to use AutoCAD LT with PartWizard and then cut the parts using Shopbot software.)

[evan]

Mike R-
Congratulations! Thank you for your perseverance in this matter. I am very interested in following in your footsteps. Would you be interested in sharing or selling your design of the gearbox? Or manufacturing gearboxes for those of us that don’t have the equipment to do so? Did you put a gearbox on all three axis? Once again Congratulations!

[richards]

I just finished some tests comparing the 1,000 steps per revolution vs 500 steps per revolution setting on the Alpha stepper driver control. The cuts were almost identical. In fact, some areas looked a little smoother with the 500 setting. I think that would have to be attributed to factors other than the resolution setting (such as the roughness on the rails).

If I've done my math correctly, at 1,0000 steps per revolution, I should be moving about 0.001047 inches per step. At 500 steps per revolution, I should be moving about 0.002094 inches per step.

The formula that I used is Pitch Diameter X pi / gear ratio / steps per revolution. Since I'm using a 20-pitch, 20-tooth gear the pitch diameter is 1-inch. PI is 3.14xxxx. The gear ratio is 3:1. So 1 X 3.14159 / 3 / 500 = 0.002094 for the 500 steps per revolution setting. The other setting is 1 X 3.14159 / 3 / 1000 = 0.001047.

A curiosity is that, at the 500 setting, and using a 3:1 gear ratio, I'm only getting an effective gear ratio of 1.5:1 when compared to the factory settings, and yet the cut is much superior to the cut gotten with factory settings. I'm beginning to wonder if torque is playing a significant factor rather than steps per inch. It almost seems that the higher torque of the geared system is better able to hold everything steady.

As of today, either the 1,000 steps per revolution or the 500 steps per revolution setting look good with the 3:1 belt-driven gear box.

I forgot to add that the Unit Value for the 500 step per revolution setting needs to be set at 1/2 the value for the 1,000 step per revolution setting.

Evan,
You slipped in while I was composing. I only put gear boxes on the X and Y motors, since I only do 2-D work. As for selling plans or kits, I'm still deciding whether it would be worth the trouble. The design is simple, but tolerances have to be tight (as I found out with my failed first attempt). Also, I built this gearbox out of polycarbonate so that I could cut it with the Alpha. (Other tests used Derlin or Delrin, which is better, but twice as pricey.) Kits should be fabricated out of metal, in my opinion. So, in short, I'm wearing out a spreadsheet to see whether a kit is feasible or whether this would have to be a do-it-yourself project.

[Brady Watson]

Nice work Mike!

FYI ~ Initial tests on my end using Mach3 and some generic 1:1 drive 600 Oz steppers showed some cogging and chatter. I felt that it should have been geared between 5:1 and 8:1 to get the best usable performance out of them. 1:1 moved at an ungodly speed on the top end, which is cute for a day or so...but cut quality is what pays the electric bill


If you by chance are going to revamp the plastic box (or others who want to tinker) you may have luck checking out some of the components offered by both Boston Gear and SDP/SI.

-Brady

[evan]

Thanks for the quick response.
Looking forward to more updates and plans are good too.

[pfulghum]

VERY interested in following... I hate the chatter I get with my Alpha.

Mike,
Can you post a picture of the "magic box"?
How big is it?
How is it connected?

[richards]

Pat,

My digital camera seems to be on the blink at the moment. I tried to take some photos earlier and got nothing. Oh well, what do you expect from a Nikon D100? I'm just kidding about that. It's a good camera, but I don't use it enough to remember what to do.

The box replaces the entire Shopbot stepper motor mounting bracket. It is roughly the same size, height and width. However, it is about 3-inches thick instead.

Basically, it takes about 15 minutes per motor to remove and install the belt drive. The Alpha stepper is entirely removed from the stock bracket and the spur gear is removed from the stepper. A new XL gear is attached to the stepper and the stepper is bolted to its new mounting plate. A base plate is mounted to the machine in place of the stock stepper plate. The base plate has two shielded bearings, two collars, a large XL pulley and a 20-tooth spur gear (similar to the original spur gear but with a 1/2-inch shaft opening instead of a 14mm shaft opening). The motors are never electrically disconnected from the machine.

The cost for parts (mostly from MSC) came to about $450 for three boxes. One-half inch thick polycarbonate for the three drives (2-X axis, 1-Y axis) cost about $75. Delrin costs double that. Aluminum and mild steel costs about the same as Delrin. XL gears for three mounts cost about $200. Spur gears cost about $45. Bearings and collars cost about $75. The shafts were from other projects, but they would probably cost about $40 for three.

Cutting time using a 1/4-inch "O" flute cutter was about 2 hours. Hand drilling for tapping took another 30 minutes. Boring out the gears took about an hour. Assembling things took about an hour. Trips to the local Home Depot added another half day. Add all of that up and you'll have an idea of the scope of the project.

In short, making a prototype had better be a labor of love, because it will cost double in time and materials what a ready-to-use device would cost. And if you're like me and have to go through two complete prototypes before being somewhat satisfied. Then add to that a third set of Delrin parts that were ruined because of sloppy circular holes (when I discovered that I had a serious problem with version 1 of the gearbox) and you'll see that building a new gearbox took the better part of a week and the better part of $1,500 for parts. However, that being said, I've been through this process many times in the past when I've designed and built process computer controllers. This particular design went fairly smoothly with very few 'gotchas'. The results were well worth the effort.

[richards]

Brady,

I agree with your suggestion for a 5:1 to 8:1 gear ratio if using Mach3/Gecko/Steppers. If I went that route, and used a 30-tooth spur gear rather than the standard 20-tooth, I would end up with a top jog speed of about 20-ips with the 5:1 box and a top jog speed of about 12-ips with the 8:1 box (assuming 45,000 pulses per second from Mach3 and the Geckos set to 2,000 steps per revolution). However, a 5:1 or 8:1 belt-drive gear box would require a multi-stage reduction unit - meaning lots of bulk. (I know. I designed a 4:1 unit and decided to try the 3:1 unit first because of the bulk of the 4:1 unit.)

On the other hand, by sticking to a 3:1 or 3.6:1 ratio, a single stage reduction unit is practical. The smallest practical XL pulley that can be bored out to 14mm is 20-tooth. The maximum size pulley listed in any of the catalogs that I looked at is 72-tooth. The 60-tooth pulley was a better match in physical size to what I wanted to do - and it still had a workable ratio.

The end result for me was that a single reduction unit could be built using the stock Shopbot software, the stock Alpha controller, the stock Alpha steppers, and stock Alpha speeds. (Except that I reduced my jog speed to 20-ips instead of the 25 to 30 that I've used in the past. The machine seems to be able to easily handle 30-ips, but moving the gantry at 30-ips with the 3X torque made me want to be more cautious.) In short, I ended up with basically a stock Alpha in all respects, except that I've added three-inches to the thickness of the stepper motor mounting plate.

TO EVERYONE
Please don't think that adding a gear box of any kind will give perfect cuts. There are still marks that will need to be sanded out - and I suppose that there always will be. The significant advantage on my machine is that unexplained 'chatter' has been eliminated on most cuts and significantly reduced on others. Smoothing or replacing the V-rails would eliminate most of the remaining imperfections, because they appear to be a mirror image of the rails. But, even if I did that, there would still be other things to tweak, each one being more minor than the previous, until searching for things to fix would become futile. At the present, I'm just happy to have most of the 'chatter' problem behind me.

[frank_hav]

Just some food for thought.

I have a new Alpha and also have some chatter, but I also have a $120,000 Dollar Osnrud which also gives me chatter. Sure I have to push the chip load more with the Onsrud, but I can buy 6 Bots for the price of the Onsrud.

Frank

[odulfst]

Wow, stop checking the forum for a couple days and look what happens.

That belt driven gearbox is very interesting, but would a regular gearbox (or smaller pinion) give the same/better/similar results? It seems to me that I should be able to purchase a gearbox similar to the one that came on the z axis of my prtalpha, and just bolt it under my motors.

When i was reading Mike's post about the performance of his belt drive vs. stock, it made me think that the control software isn't taking advantage of every step that the motor has, or maybe that it makes calculations w/ less steps and runs a smoothing algorythm (or no smoothing...) I think most of the chatter marks I'm experiencing are driver related or software related since the marks are much bigger than the smallest amount that the machine can move. For example, I made a wing model recently, and I noticed the CAD model had some minor imperfections. Well the real model came out with the same minor imperfections, so i think that make when I run the CP command, the software simply isn't producing a high resolution circle. Maybe I can test this by using other software to produce the .sbp.

I think that typically "chatter marks" is a misnomer in itself. chatter usually describes soem randon oscillation of the bit that causes surface roughness. when I cut a big circle in 3 steps, I get the same exact set of "chatter marks" on each step.

I suppose the test w/ the CAD file would prove that either the software is causing the problem or some kind of post-processing/motor drivers are causing the problem. I don't think it's the drivers, because they work with way more than enuf resolution from command to command. so, maybe the usb connection can't keep up? maybe I need to ditch Win98 (even tho it's supposedly running usb 2.0). maybe I can get away with a change in ramping values or that look ahead setting.

ramping values, that what i wanted to talk about. i got side tracked. so about the random changes in ramping values (see above post)...(way above) it's just that. Random. It doesn't seem like I could change a setting and make it better. One thing that did catch my interest is the comment about speed values. I often re-spec the speeds after compiling a file while running it, but I always use the stop command and insert, so it should re-calculate the ramping. and it does, and works great most of the time until it decides to get adventurous and throw that ramping crizap(ha, filter that.) out the window.

The only thing that catches my eye so far is the look ahead setting. maybe that will help.

any suggestions?

Tony.

[Mayo]

I have noticed "chatter" in some material but not in others. This leads me to believe it has a lot to do with choice of bit, cutting speed, and hold down. When I slow down my cutting speed I get less chatter. I think this allows the bit to actually cut rather than take chunks out, and it produces less horizontal force against the material as it is cutting. It's a combination of several factors at the same time and I think all the ideas mentioned above have some part in the equation.

One way to disprove my theory about it stemming from choice of bit and cutting speed and hold down would be to run a test of a file that produces chatter on a bot equipped with plasma or laser.

[patricktoomey]

Mayo, I've reproduced it using a dragknife vinyl cutter on my PRT. I think you're right though that there are many factors here and some of it is due to horizonal loading, gantry deflection, cutter deflection, etc. The vast majority in my case though is the low resolution stepping of the drivers/controller. I cut corian using a high bit speed and low feed rate and have the same exact "chatter" patterns no matter how slow I go. As someone mentioned earlier, this isn't what is usually referred to as chatter which would usually mean a bit issue. It's a resolution issue that most people are concerned about and that can be eliminated with a different control box.

[kent]

I've been lurking and learning for quit awhile. This thread has raised issues that at the moment has me about to back out on purchasing a shopbot. How bad is the chatter that yall have been discussing? Light sanding or sanding out "saw tooths".

[scottcox]

Kent,

I've had the Alpha with a spindle since April and have only encountered chatter ONCE. And that was in the middle of an unsupported workpiece. I braced my work better and the chatter was gone.

I cut some 3/8" acrylic today and was expecting to have to clean up the edges. It wasn't neccessary. I got a very smooth cut first time.


7704
No chatter.

[gcoding]

Scott
Looks nice, just wondering what speed was you running. It seems alot of the chatter is happening at higher speeds

[scottcox]

Thanks Jeremy.

This was done in three steps, all in Parts Wizard 2. Material is 3/8" Acrylite GP. Viewport is 6.25 in diameter with eight 1/4 inch holes.

1) Peeled top paper and V-carved the Arial text at 18Krpm/2ips,
2) Straight plunge into 1/4 inch holes with a 1/4 inch upcut spiral bit at 10Krpm/0.1ips,
3) Cutout using 1/4 inch upcut spiral at 16Krpm/1ips with a stepdown of .1 per pass.

No burning/melting, good chipload, good edge. My customer is happy so I'm happy.


7705

[mikejohn]

Has anyone tried this.
Cut out a piece that is showing obvious chatter.
Register exactly the position and orientation of the piece, creating a registration mark.
Cut out a second piece in exactly the same place and orientation, with registration mark.
Now, line up your registration marks on the two pieces, and see if the chatter occurs in identical places on both parts.
As I have said before, I don't get chatter (don't know why) so I can't do the experiment.

..............Mike

[cnc_works]

Mike, I can already tell you where it happened for me. Always worst at each 45deg to the axis, where x & y were handing off.

Donn

[richards]

It seems that the increased torque from the 3:1 gearbox is more important to 'chatter' free cuts than the 3X resolution increase. Let me throw some numbers out. The factory defaults for my Alpha is a one-inch pitch diameter, 20-tooth, 20-pitch, 20-dregree pinion gear and 1,000 steps per stepper shaft revolution. That means that each step pulse would move the axis 0.00314 inches. After installing the 3:1 gearbox, each step pulse should move the axis 0.001047 inches. Changing the steps per revolution to 500 should move the axis 0.00209 inches. In short, the factory direct drive 1:1 setup should have a resolution of 1/300th of an inch - plenty accurate out here in the West. However, both the 3:1 cuts at 1,000 steps per revolution and at 500 steps per revolution are extremely good and nearly identical. It would seem that the 500 steps per revolution setting would produce 'chatter' right in the middle of the factory setting and the 3:1/1,000 setting, but the 3:1/500 setup is as good as the 3:1/1,000 setup. That makes me think that the 3X torque increase is the major factor in reducing 'chatter'. It's all a little confusing, and probably not worth the time it takes to think about it because if the 'chatter' problem is solved, I don't really care whether it was increased resolution or increased torque, but inquiring minds want to know. What do you think?

[jsfrost]

Kent,

It appears from inputs of others, that some machines chatter more than others, and that chatter is due to a variety of factors, including cut speed. In my own work, the quality of cut on Shopbot produced work is superior to the quality of cut I can get from my tablesaw, scrollsaw, bandsaw, or manual router table. And the machine can do things I can not achieve by hand. However I never push for maximum speed, it is not required for the parts I produce.

But don't take my word for it, or, on the other hand, reject SB because others have issues. I fully believe that cut quality is a problem in some applications, and insignificant in others. I would recommend you find a Shopbot operator in your area, and arrange a demo doing cuts that show how the tool can work for you. Or not.

[richards]

Here's a photo of the 3:1 belt-driven gearbox.


7706.

As you can see, this model is cut from 1/2-inch thick polycarbonate. The spacers are 1-inch aluminum. (The extra-long shaft and rusty gear are left-overs from other projects.)

Starting at the shaft, there is a 1/2-inch clamp collar, #9 rubber O-ring, 1/2-inch shielded ball bearing, motor plate, 60-tooth XL gear, main plate, 1/2-inch shielded ball bearing, #9 rubber O-ring, 1/2-inch clamp collar, 20-tooth, 20-pitch, 20-degree spur gear. The aluminum spacers are 1-1/2-inch high, threaded for 1/4-20 machine screws. There is a 20-tooth XL gear bored out to 14mm to fit the Alpha stepper motor. The belt is a 150XL037. The original 3/8-inch mounting bolt was replaced with a longer 3/8-inch bolt, due to the extra thickness of the polycarbonate. The motor mounting bolts are 1/4-20 X 1-inch socket head bolts (so that the belt can easily be adjusted from the rear of the motor).

This version took about 1.5 hours to cut (Y-axis, 2 X-axis, no Z-axis) using a 1/4-inch O-flute cutter at 1.25-ips. Machining the spacers took forever. (Whatever made me think that using a metal mini-lathe would be fun? Can you say one full day machining and threading spacers?)

[Brady Watson]

Hey Mike, nice job!

Have you given any thought (during your mental engineering sessions) to using off the shelf BWC gears as a reduction setup? It could be done with 2 or 3 gears depending on whether the direction of the motors can be reversed. No slipping or stretching of belts...although they would be a little noisier as all gear drives are.

-Brady

[richards]

Brady,
I've thought about using gears, but I'm leary of backlash. Without designing in a lot of adjustments, I would think that I'd have significant backlash. The belt-drive, if tensioned properly, is backlash-free.

[pfulghum]

Great job Mike...
Now How much for the rest of us?


I'm wondering if the belt is also taking mechanical vibration out of the system to clean the cut as well.

-- pat

[richards]

Pat,

This has been a labor of love. Costs were not so important as results; however, realistically, it would cost about $150 - $175 per motor mount to duplicate what I've done. That's figuring actual costs plus a reasonable per-hour charge for labor. A do-it-yourselfer, who has more talent than I, could probably save a little. However, so far, I'm into the project well over $2,500. Is it worth it? In my opinion, getting rid of the 'chatter', or at least reducing it to a minor level, is well worth the cost.

By the way, I've received several emails asking whether polycarbonate is strong enough to hold up in the long run. Since I have no personal experience with polycarbonate, I can't answer that question; however, I've literally pounded the polycarbonate with a 3-lb sledge (to remove a bearing) and could not inflict any visible damage to the polycarbonate (it's the same stuff used in 'bullet-proof windows). The 3/8-inch hinge bolt is the weak link in the whole assembly. As far as I can see, all the twisting and torquing that happens to the motor mount, first happens to the 3/8-inch bolt.

[gerald_d]

Mike, the polycarbonate might want to break around the 3/8" bolt, but the bolt itself won't be the first thing to go (if it is a half-decent quality bolt). A big steel washer under the bolt head will spread the forces to the plastic more gently.

[richards]

Gerald, The bolt is grade-8, so it won't fracture before the plastic. The radius of the mounting washers are greater than the radius of the plastic, so I don't think that the bolt/washers will be the weak link. The plastic is 1/2-inch thick, so I think that it will withstand all forces - but time will tell.

If I had my way, and if I had a good CNC vertical mill that could cut aluminum, I would have built the bracket out of 1/2-inch aluminum (and saved some money on materials). But, at a projected cut-time of more than 3-hours for aluminum, vs 1.5 hours for polycarbonate, I decided on polycarbonate. If I'd had any experience in cutting aluminum on the Shopbot, I might have decided on aluminum, but searching the forum didn't give me enough confidence to try aluminum; however, I did buy a piece of sheet stock large enough for one bracket. That piece of aluminum will be put in my 'next project' stock pile until I get access to a metal cutting CNC-mill.

This is version 4 of the belt-drive gear box. Two of the models were cut from Delrin, which is twice as expensive as polycarbonate. Delrin seems to be even better than polycarbonate, but spending $200+ every day to test-cut motor mounts, got old in a hurry. Also, Delrin seems to be more flexible than polycarbonate. I don't think that flexibility is a real issue in this design, but I didn't want to take a chance.

[mikejohn]

Haven't done a lot of thread drift lately, but I will try one here
.
Apart from the slower spindle/router (or whatever it's called for metal work) is there any difference between a metal CNC mill and a Shopbot?

............Mike

[gerald_d]

A huge difference. The metal cutting machines go for rigidity and hence high mass. The wood routing guys are speed maniacs they want light weight. Truck tractors vs racing cars. Somewhere in the silly world in between, there are those racing trucks, but they are neither good trucks, nor are they good racers.

[pfulghum]

Mike,

I would like to try this.
Can you recommend a place online the gears/belts/axle/bearings?

Can you take a photo from the spur gear side?

-- pat

[richards]

Pat,
MSC is one of the places that you could use via the Internet. The shaft/axle is just 1/2-inch drill rod that you can get at any machinist supply (instead of paying air-freight to have it shipped with the pulleys, bearings, belts, etc.) A local machine shop could mill the 'flats' onto the shaft, or you could grind them using a Dremel type tool.

Right now the machine is in the middle of a long job, so I can't remove a motor for photos; however, let me describe the process. Starting at the end of the shaft/axle, install the new spur gear. [Buy one that fits a 1/2-inch shaft either from MSC or Shopbot. Be certain that the pitch and degree angle match your machine. You can vary the number of teeth to suit your needs. Ted suggested 30 teeth - which I'll try on my next version. The spur gears from MSC need to be drilled and tapped for set screws.] Install the clamping-collar onto the shaft. On the X-axis leave about 1/8-inch between the spur gear and the collar. On the Y-axis, butt the spur gear against the collar (and add two fender washers to the 3/8-inch mounting bolt to move the entire motor mount about 3/32-inch away from the machine so that the spur gear rides the rail properly). Install the #9 rubber O-ring. Slide the axle/shaft into the unit. With the shaft, barely protruding from the bearing, position the large XL gear onto the shaft (be sure that the XL belt is draped around the gear). Slide the shaft through the 2nd bearing. Attach the O-ring to the backside of the shaft and then attach the collar onto the shaft. Mount everything onto the shopbot and adjust collars, O-rings, gears, until everything lines up properly. The whole process takes about five minutes or less.

The O-rings keep the collar from rubbing against the bearing. Since the collars and the bearings that I use are almost exactly the same outside diameter, I had to find somekind of spacer to put between the collar and bearing. The O-ring was exactly what I needed. The bearings that I use have an outside diameter of 1.125-inch and are 0.3125-inch thick. Make sure that the bearing pockets are sized properly for the bearings. Press the bearings into position with a vice or arbor press - don't use a hammer (don't ask me how I know). You might want to cut the plates from MDF for testing purposes and then switch to polycarbonate or Delrin when you're satisfied with your design. You could also try aluminum, but when I looked at the suggested feed rates and depth of cut on the forum (0.05 depth per pass and 0.5-inch per second), I realized that I didn't want to spend 10 - 15 hours cutting aluminum. During August, with afternoon thunderstorms, I usually lose power at least once a day.

By the way, use the basic dimensions for the factory motor mounts. My version 3 motor mounts were as compact as I could make them and interchangeable for any motor (the standard mount needs to be 'mirrored' for the farside of the X-axis). However, the stop blocks on the rails had to be moved - and the Y-axis stop blocks can't be moved. Anybody want to buy $200 worth of Delrin that is slightly used and shaped somewhat like motor mounts?

If I can find a local shop that is willing to charge a reasonable price to mill everything from aluminum, I'll offer a commercial version so that you won't have to waste time and materials experimenting. So far, the shops that can do the job quickly want a guarantee of several hundred units - and I'm only willing to guarantee a dozen or so units, meaning that only small slow shops that charge by the hour are willing to talk to me.

[supertigre]

Mike;

Don't you already own a mill that can handle aluminum (think carefully - big blue thing in your toy room)?

Ooops just reread your post, it was considered.

Guy

[pfulghum]

Mike,

Per your specs, here is what I've found so far...

Bearings:
http://www1.mscdirect.com/CGI/NNSRIT?PARTPG=IMLMKD&PMPXNO=1872952

Timing Belt: Item #35393958
http://www1.mscdirect.com/CGI/NNSRIT?PARTPG=IMLMKD&PMPXNO=1869927

The XL series pulleys from msc are on
http://www1.mscdirect.com/CGI/NNPDFF?PMPAGE=3773&PARTPG=N2DRVSH&PMT4NO=0&PMITEM= 35375823&PMCTLG=00&PMT4TP=*LTIP

#35375997 - 60 tooth (but its 3/8" bore)
#35375864 - 20 tooth (but its 1/4" bore)

I can't seem to find the right pulleys.

Where did you get your pulleys from?
Can you post your frame design that I can use as a starting point.

-- pat

[richards]

Pat,

I would substitute a shielded bearing #35433259, istead of a sealed bearning. (They turn easier and the large chip size of routing wood keeps them free from contamination.)

The timing belt and XL pulleys look correct. You'll have to bore out the pulleys to fit the shafts, 14mm for the 20-tooth and 0.500 inch for the 60-tooth.

I bought the spur gears from McMaster-Carr, part no. 5122T12 (Catalog page 985). On that page you'll also find other 20-pitch, 20-degree angle gears. I picked 20-tooth to duplicate what I started with. (Remember that the Alpha stepper has a shaft diameter of 14mm, so the original spur gears are bored too large to fit a 1/2-inch shaft. Messing around with shims, etc. probably was the cause of out of round holes in version 2, which cost about $200 in wasted Delrin - which was entirely my fault.)

(When I looked at the shipping box, I noticed that all gears, pulleys, bearings, collars, etc. were purchased from McMaster-Carr instead of MSC, due to the fact that McMaster had the 20-degree spur gears.)

The frame design is fairly basic. Take the right-side x-axis motor mount off of your machine. Lay it onto a piece of paper and draw the outline, 3/8-inch mounting hole, spring hole, and large center hole. The center of the large hole will be the center of your bearing mounting hole. Three and 1/4-inches above the center of the bearing mounting hole will be the center of the Alpha stepper motor's shaft (so that a 150XL037 belt will fit properly). You'll find all dimensions for the Alpha stepper on Oriental Motors website under AS911AA.

[gerald_d]

Sorry Mike, a sealed bearing in this application is way better than a shielded bearing:
http://www.talkshopbot.com/forum/messages/26/7470.html
Shielded bearings are better at the 20 000 rpm's (router) because of heat and friction, but down at the slow steppers, the extra friction is a very minor issue.

[pfulghum]

Brady / Mike

A standard alpha has...
1) 1:1 gearing
2) 20 tooth Spur on a 20 Pitch Rack
3) 1000 step/rev
4) Unit Values = 1273.xx

On a 20 Pitch rack...20 Teeth = 3.14159".

On a stock alpha, one revolution of the motor (1000 steps) will be 3.14 linear inches. The unit values for Alphas are 4000/PI or 1273.23.

There is this 4X "multiplier" in the "VU" command that comes into play in the above calculation calculation.

Gleaming from this thread, Brady is running...

1) 7.2:1 geared "special" alpha
2) 30 tooth Spur
3) 500 step/rev
4) UV = 1527 (I'm questioning this).

The effect on the unit value for each of your "Modifications" is...

1) Multiple UV by 7.2 (more steps per rev)
2) Multiple UV by 20/30 (spur gear increase - reduces torque - increases speed)
3) Multiple UV by 500/1000 (driver step rate decrease)

That would be..
1273 * 7.2 * 20 / 30 * 500 / 1000 = 3055.

Brady you say your unit values are 1527

Did you change the multiplier?
Where is my math error?



I've talked to Gary back at the factory about chatter, and I bought some 25T spur gears to help. It seems like they help (barely) but that is contray to I don't know why. Going to a bigger spur gear decreases both torque and resolution.

Here is my dilema...

- Mike has great success with his 3:1 gearing giving hime 3x the torque. (maintaining a 20T spur).
- You seem to really like your setup 4.8x normal torque with 30T spur. 7.2 * 20/30 = 4.8x
- Factory is pushing me to bigger spur (decreasing torque and resolution)

Should I ...

a) Build a 3:1 belt system with 20T spur like Mike

b) Build a 4:1 belt system with 25T spur

Pros/Cons ???


-- pat

[Brady Watson]

Pat,
Using the 1000 setting my UV is 3819.xxx, step multiplier is 5. Using the 500 setting, UV is 1527.xxx and multiplier is 2. Torque is approximately 2 to 3X a standard Alpha, not 4.8 times.

One thing to bear in mind that is not readily apparent is that a larger spur/pinion gear has more teeth meshing with the rack and offers smoother operation. Take it to the extreme by imagining a 12 tooth pinion cogging down the rack compared to a 48 tooth one...

I'm not going to *should* on anybody...you'll have to make up your own mind. I'm sure that Mike will be the 1st to point out that making your own gearboxes is doable...but requires time and money to do it right.

-B

[richards]

Pat,
The maximum ratio with a single-stage belt-driven gearbox is 3.6:1 (20-tooth and 72 tooth). Gears smaller than 20-tooth are too small to bore to 14mm (if manufacturer recommendations are followed). The largest standard gear is 72-tooth - and it is large enough to look awkward on the Shopbot.

Going to a multi-stage gearbox would allow you to have almost any ratio, but at the cost of greater complexity and greater bulk. The main problem is that belts need to be adjusted/tensioned properly. The only two ways that I've tried is moving/sliding one gear while holding the other stationary or using a tension roller. A tension roller is more elegant, but with only 1/2-inch between the gears, there's not much room for a tension roller. Separating the gears far enought to add a tensioner means a bulky gearbox. In a multi-stage gearbox, moving/sliding shafts around really adds to the parts count and adding tension rollers adds inches to the gearbox.

The reason that I stayed with a 20-tooth gear was to keep the ratio as high as possible with my single-stage unit. So far, I haven't tried any other spur gears - but I will as time goes on and I have time to experiment a little. The simple fact is that the 3:1 ratio solves most of my 'chatter' problem, and that was the entire reason that I started messing around with gears.

Let me caution everyone about the torque. It is significant! Be careful that you don't slam into the stops! (Don't ask me how I know. Nothing broke, but I really felt stupid.)

Now, just a word about the cost. Anytime I design and build prototypes, I always count on the cost being at least 4X the cost of parts - and that doesn't count the downtime/labor that I contribute. Version 1 sometimes works. Version 2 fixes most of the bugs. Version 3 usually is pretty close. Version 4 is where I normally stop until I discover a major problem. Right now, I'm on Version 4 of the gearbox. Just as expected, it cost a whole lot more that it should have and it took a whole lot more time than it should have. Had the 'chatter' problem not been severe, I would have just been wasting time and money. However, because a problem was solved, it was worth it - to me. Your mileage may vary.

[gerald_d]

On 20 vs 25 tooth pinions....

Forcing teeth into contact by spring-loading the pinion against the rack is very unusual in the gear world, and a 25 tooth will ride smoother because of its bigger diameter.

Also, a 20 tooth pinion on a 1.8 degree stepper motor (200 steps) will always have the gear teeth in resonance with the steps. Moving away from 20 teeth (either more or less teeth) gets away from the resonance situation.

[gerald_d]

The major selling point for the Alpha when it was launched, was the the higher cutting speeds. Putting in gearboxes and belts surely reduces those speeds, but not by as much as the ratios would calculate. Can those with Alpha's and gears/belts comment on speed losses?

[Brady Watson]

Zero loss in cutting speed. My geared Alpha has full torque @ 12 IPS (due to the electronic governor via SB3) and jogs a full 30 IPS...although I keep it down around 15-20 IPS on the jogs to reduce unnessesary wear & tear on the rails and geartrain.

-B

[mikejohn]

Brady
You do a lot of 3D stuff I believe.
How much time saving do you get cutting 3d files on the Shopbot compared to what you got on your original system?

.........Mike

[richards]

Gerald,
No speed reduction; however, I'm running at 500 steps per revolution instead of the original 1,000 steps per revolution. (No loss of quality at 500 spr.) Like Brady, I've slowed my jogs down even though the machine can still jog at 30-ips. My unit values are 1909.8592 instead of 1273.2395.

Before changing from 1,000 spr to 500 spr, I was still able to run at fun speed and jog at full speed.

That the Alpha stepper/drivers can still deliver full speed is no real surprise. At 30-ips, with a 3:1 ratio, the Alpha stepper is only running at 1718 RPM. Factory specs say the limit is 5,000 RPM - so there is still plenty of headroom available. Finding a way to send pulses that fast would probably be the issue.

[richards]

Maybe I should be a little more exact when making claims for Oriental Motors/Alpha stepper/drivers. The charts actually show speed/torque data for speeds up to 4,000 RPM. The main data sheet also shows a maximum pulse rate of 250,000 pulses-per-second (50% duty rate) and the Connection/Operation data sheet lists 5,000 RPM as the speed at which the overspeed alarm is activated.

All of those figures are far beyond speeds required to match the unaltered Alpha's speeds and far beyond the speeds required to maintain factory speeds with any reasonable belt-driven gearbox.

In fact, I don't think that it would be possible to build a reasonably priced CNC router that would ever be limited by the Oriental Alpha steppers. When I've gone through the thought process to try to determine what parts would have to be replaced or modified on my Alpha if I were to run steppers at 4,000 RPM (jog speed) and maybe 3,000 RPM cut speed, the cost, just for parts was a multiple of the cost of the original machine - and that didn't include dust covers to protect parts that would be much more intolerant of contamination, nor did it include the cost of a modified controller that could send out 250,000 pulses-per-second or fully syncronize all axes during fault conditions.

In short, the Alpha, as shipped is a pretty good machine. Adding a gearbox (for those of us who need less chatter) is a fairly low-cost modification. Probably one of the greatest features of the Shopbot is that it can be easily modified, at reasonable expense, to suit our exact needs.

[Brady Watson]

It depends on the size of the parts being cut and the intricacy of the part. Smaller parts have a smaller increase is speed, much like trying to cut a 1" circle at 10 IPS...not gonna happen. I've cut 3D at a max of about 8,3.5 IPS on larger parts. The big leap ahead for me is being able to run over 3 IPS without losing steps...I've ruined countless 3D parts on the PRT from running on the ragged edge and lost steps. 3D is faster if you learn to tweak your ramping values properly. What runs in 2D doesn't always work for 3D.

-B

[frank_hav]

Brady,
I also do alot of 3D work. Most of mine fits in a 12" cube. I would like to adjust my ramp speeds so they closer match my move speeds. As you pointed out, alot of the time the cut never gets to full move speeds. So I increase the speeds so it cuts faster 3D, but this makes the occasional straight cut to fast. Any thoughts on where to start.

Thanks
Frank

[Brady Watson]

Frank,
Read pages 20-22 in the Command Reference PDF, under Help in SB3 to get a better understanding of how ramps are affected in 3D. Check out the VR, and VU commands and read about Small Circle Definition under that command.

My tool is a bit different but on a standard Alpha (or PRT) try adjusting your VR-> Slow Corner Speed (SCS) down to 35 and your Minimum Distance to Check (MDC) down to 0.08 for 3D. If you are getting 'banging' when the X reaches it's end and the Z plunges or raises, then reduce the SCS - If it goes too slowly or crawls over your 3D relief, raise the SCS. 3D reliefs vary greatly and a smaller intricate relief will have very different ramping values than a large relief with long smooth sweeping moves.

Ramping can be pretty subtle, so I would recommend running a 3D file in styrofoam and stopping the tool movement, <I>nsert command VR and make an adjustment. Resume the cut and notice the change. Etc etc. Make sure that your move speeds are fast enough to get out of ramp speed (if doing a 12" sq. then set them at 4,2 to start, slower if it has a lot of peaks and valleys). I haven't found the 3D ramping threshold setting to be particularly useful for me, but after you get things close, you might want to change values to see if it helps your tool. I only change SCS and Min Dis Check. They seem to exert the most influence. When switching gears and cutting 2D, if you find that it is too slow in the corners, then leave the MDC and raise the SCS to 50-65.

Messing with ramps is like tuning a carburator...give yourself some experimentation time to find how the tool responds to your adjustments. If you do this a few times, you will be able to adjust ramps and fine tune your tool's behavior like a pro. I've found that these little adjustments can make a big difference. If you get into trouble with ramping, just reset the defaults and start over! (Note: This post regards SB3 & is not applicable to v2.39 and earlier DOS versions of the control software)

-B

[frank_hav]

Brady,

Thanks, I was playing with the VR today. I was setting the ramps to 2,1 both move and jog. And then i set the normal move and jog speeds to 2,1. This produced a really smooth motion but I'm not sure what the side effects are. Basicaly I was setting all speed motions to 2,1 and eliminated speed changes. GOOD OR BAD ???

Frank
PS It is a 2 month old Alpha 48 x 48 x 12

[Brady Watson]

Frank,
Think of ramp speeds as a percentage of move speeds. E.G. - You set your MS to 4 IPS and ramps for X & Y are set to .4, then it will make a 90° turn @ 4 X 0.4 = 1.6 IPS. There are other factors that affect this speed, so this is just for a mental illustration.

I never touch the X,Y,Z ramp values. I leave them at the default value of 0.4 ~ I DO however change the Slow Corner Speed and Minimum Distance to Check values, and recommend that you start out by doing the same. Restore the default settings and try just messing with those 2 values. What works for me and the type of work that I do may not be the best setting for your machine. I will say that when I set up other people's CNCs for them, I only change these 2 settings to smooth things out if necessary.

Are you sure that you set the VR to 2,1 or did you mean your MS was set @ 2,1 ?

-B

[frank_hav]

Brady,

I set both to those settings. I was assuming this would basically eliminate any ramp change. I will go back to defaults and try changing the values you talk about. I'm interested on the differences it produces. I use this machine for R&D projects so for me this is more of a learning session then any problem I'm trying to cure. I cut most of my long projects at night while I sleep, so I just slow them down anyway. A slow cut that finishes 5 minute before I get there is better than arriving to a broken bit and an unfinished piece.

Thanks Brady

Frank

[frank_hav]

Brady,

I just noticed you are on the other side of the river. We'll have to visit each other sometime.

Frank

[flyboy]

It's been a while since I've been on here but I've been tinking about the upgrades too. It seems everyone wants more resolution and more speed. Of course we do.
My question is at what point would I just want to go with servos? My freind built his own cnc router and used servos and that thing hums alons really fast and is very very accurate. he used ball screw linear slides though at considerable cost. I just want my shopbot but want to upgrade it. Would a servo system be better or not? Seems like i can do it for about 4K myself with a little scrounging.

[richards]

Tom,
I'm in the middle of that very project. Adding servos is not at all difficult, except that you have to have significant gearing (especially with a rack and pinion drive). The servos that I have on my test bench have usable speeds in the 2,200 to 6,000 RPM range. A 30-tooth spur gear only needs to turn 382 RPM to give you 30-ips jogs. So, depending on the servo motor selected, you would need gearing between 6:1 and 16:1.

Also, a servo motor has much less torque than a stepper - given the same basic size, so gearing is necessary to multiply the torque.

The ID3002 motor from MCG runs at about 3,100 RPM when using a 70VDC power supply and Gecko G320 driver. It produces 120 oz*in torque. With a 12:1 gearbox (which would limit jogs to about 20-ips), I would have 1,400 oz*in of torque. That motor costs about $160. An encoder would add another $75 and a G320 would add $115. So, the basic costs are very reasonable. Gearing, per motor, would cost between $250 and $1,000, depending on whether you built your own gear box or bought one from a place like Bayside (a part of the Parker company).

If you're not the kind of guy that likes to play with electronics, you can buy AC servos and drives from Automation Direct for $600 to $800 per motor/drive. Of course you would still have to add gearboxes.

One of the big factors that you'll have to be aware of is that the Shopbot Alpha controller realistically puts out about 30,000 pulses per second - which is very respectable. A 100 cpr (400 quad) encoder attached to a 3,100 RPM motor would require about 21,000 pulses per second. However, with a 12:1 gearbox, you would end up with 0.0002 resolution per step. That's right two ten thousanths of an inch! I think that native resolution would no longer be a concern.

Basically, you would need the motors, encoders, Gecko drivers, power supply and a simple signal inverter to catch and invert the pulses produced by the Shopbot controller - and some gear boxes. Don't forget that a servo can require a lot of power. The ID3002 motor needs about 5 amps continuous current and can draw as much as 26 amps before frying, so don't ignore the need to use a beefy power supply.