Accuracy of PRT vs. Alpha?

[sol]

I'm not concerned about speed at all, so I ordered the regular PRT for $6200.00. Since there is quite a long lead time, I figure I could change the order pretty easily. My question is, does the Alpha model offer higher precision? Thanks.

[Brady Watson]

Eric,
Yes it does...but do you need it? Depending on what you are cutting, you may not need the additional accuracy & speed that the Alpha affords you.

-Brady

[sol]

Hi Brady,
I am cutting patterns and inlays for high end furniture. The precision required is mostly 2D. I don't plan on tackling 3D at the moment. What are the tolerances for the PRT and the Alpha? Thanks for your help.

Eric

[Brady Watson]

Advertised resolution for my PRT with old-style gantry:
Step Resolution 0.002”
Repeatability 0.005"
Positioning accuracy +/- 0.005”

Advertised resolution for *new* PRT with Alpha-style gantry:
Step Resolution 0.0013"
Positioning Accuracy +/- .005
Cutting Accuracy +/- .015

Advertised resolution for Alpha*:
Step Resolution 0.0015”
Positional Accuracy +/- 0.005in
Overall Cutting Accuracy +/- 0.015in

Keep in mind that these are advertised specs...and depending on how well you set up your new machine and maintain it, your mileage will vary. I was very careful when I set up my machine and as a result, I have excellent repeatability and dimensional accuracy. One advantage the Alpha gives you over the standard PRT is closed-loop feedback and correction...meaning that it prevents lost steps. This shouldn't be a problem though if you operate the PRT within the designed limits. You shouldn't have any problems doing inlays with the PRT. I have done many inlays with my PRT and they are dead-on the money. The money you save with the PRT could be spent on more robust software like Insignia or ArtCAM Pro.

-Brady

* I am not sure these values are correct for the Alpha...I could have sworn that the advertised overall accuracy was much better. You might want to call SB to check.

[sol]

Thanks for the info brady.

I guess I'm sticking with the regular PRT. Wood can easily move way more than 5-ten-thousandths of an inch in the same day, so I won't sweat it. Besides, shellac and sawdust make an excellent wood filler!

The painful part is the wait. They said 4-6 week lead time. Anyone have any comments on how accurate that is? Thanks.

[richards]

Precision, accuracy, repeatability are words that some use interchangebly but actually have different meanings. Without getting into a 'word' war, let me simply say that I would buy my Alpha again without hesitation, BUT, at least now I know that a hand-held router following a well-made template will give a SMOOTHER finish than I've been able to get with the Alpha - at all speeds and with all kinds of cutters. The Alpha has given me accuracy within my ability to measure. It's also totally repeatable - assuming that I have the ability to write a file that keeps all moves at reasonable and sane speeds, using the right cutter, etc., etc., etc. Precision, however, still eludes me. No matter how I try, I still get 'chatter' in portions of each cut. Sometimes a curve has 'chatter' marks. Sometimes an X-Y move has 'chatter' marks. Sometimes the ramp-up or ramp-down portions of a cut have 'chatter' marks. Whatever the cause or the case, I've found that a little sand paper and a little elbow grease needs to be part of the production cycle. Since I believe that I have a fairly good understanding of stepper motors some small understanding of the physics involved in cutting parts with the Shopbot, I don't see any easy way to improve accuracy, repeatability or precision. The tool is very good. It is better than advertised. And, with the exception of requiring a little sanding on templates (so that the 'chatter' marks don't telegraph onto the piece being cut via the template), it's almost a perfect machine.

[gerald_d]

The Alpha and the "special"PRT (http://www.shopbottools.com/prtspecial.htm) have identical frames, cars, rails, v-rollers, racks, gears, hold-downs, spring-tensioners. Even the gearbox ratios may be identical (3.6 : 1) and the steps per rotation of the stepper motors may be identical (200). Therefore, you are probably getting identical precision, accuracy, repeatability unless you are working in the speed/feed range where the "special"PRT is at risk of losing steps while the Alpha would compensate for losses.

[stevem]

The Alpha does not have gear boxes on the steppers, therefore should be somewhat more accurate due to reduced backlash.

[gerald_d]

Without gearboxes there may be less backlash, but the steps become "coarser" and the torque at the pinion is much less. It is a trade-off. SB has at various times built Bots with and without gearboxes and there isn't a huge favourite either way for getting better specs. (The money_no_object guys tend to put in top quality gearboxes to get finer steps and more torque out of the drive system)

[dingwall]

Mike, the chatter you mention is possibly scalloping due to alternating steps. When cutting in either X or Y direction the cut will be smooth. Things get jagged when the steps start to alternate between the X and Y motors. I.E. X moves 1, Y moves 1. On my machine with DOS SW it's most noticeable at 45 degrees. I haven't checked it out with the Win SW.

I was frustrated with that too. Once I measured the depth of them (.002") I felt a little better.

[richards]

Sheldon,
You're absolutly right. The steps are shallow and they almost always occur when multiple axes are moving simultaneously. In addition, the smaller the cutter, the more pronounced the chatter, i.e. 1/4-inch cutters show more chatter than 3/8-inch cutters (cutter flex?). A quick pass with some 150 grit sandpaper usually rubs the chatter marks right off; however, when the goal is to cut a template (for all the neighbors that have a seemingly endless need for router templates), the chatter telegraphs through onto the piece being routed.

On my next set of templates, I'm going to try cutting them 1/16-inch oversize with the 3/8-inch cutter and then running the final pass with a tool path that fits a 1/2-inch cutter. That would mimic the way I did things pre-shopbot - rough cut 1/16 - 1/8-inch oversize on the bandsaw and then cut to size on the router table via a template and a 1/2-inch cutter. My thinking is that a 1/2-inch cutter would produce even less chatter than the 3/8-inch cutter. I'm picking nits here because EVERY tool has limitations, otherwise, all we would need would be that one perfect tool that did everything perfectly. The Shopbot is a great machine that performs better than advertised. Whether it produces a cut as good as, better than or worse than the 'big iron' machines doesn't matter to me because I don't have access to a 'big iron' machine. (And whether the 'big iron' machines actually make single pass cuts with a single cutter that are chatter free is another matter.)

[gerald_d]

Am I assuming correctly that the change from PRT to Alpha caused little or no change to the chatters at 45 degrees? Steve, you are running a stiffer gantry and a different control system, did you see a difference in the "45 chatters"? I am trying to figure the primary cause of these marks - motor steps, resonance, flex, whatever. Sheldon, you changed gear ratios - did you see a difference?

Do the big iron machines do this? Apparently they do too if you see this on the Komo (http://www.komo.com/CNC%20Routers/Machines/solution_series.htm) website: "Know the best way to choose a CNC router? Run your finger along the edge of a finished part, glide it around a corner. If it feels like it needs sanding, it wasn’t machined on a Komo. The Solution reliably produces the finest, chip-free finishes you've ever seen."....and they certainly aren't comparing themselves to ShopBot, they are talking about other machines in the big iron class.

[richards]

Gerald,
I've read that also, and wondered whether they were touting the ability of their machine to cut a 'smooth' part with one tool, or whether they assumed that the operator would use their automated tool changer and use as many tools as necessary to produce that 'smooth' part.

Now, back to the original question posted by Eric Sun, I have no idea whether the new Alpha design is more accurate than the PRT design, since I don't have a PRT, but I do know that by tweaking the ramp speeds, feed speeds, and especially by adding a spindle to the Alpha, that I'm getting amazing results. I've also learned that the machine has certain limitations, at least as I have it configured, that would limit what I can do without post-processing the cut parts; however, knowing those limitations means that I should cut parts that the Shopbot can cut efficiently. If the parts require a lot of post-processing, then either that cost has to be added to the bid, or the customer has to do the post-processing.

[stevem]

Gerald, there is a measurable improvement in the quality of the finish with the stiffer gantry, although I’ve never had problems with chatter since my machine was converted to Gecko drives immediately upon assembly.

The picture below shows an MDF prototype mold cut in a single pass on my home built “big iron” style router. On a recent job cutting Acrylic into circles on the Bot, the edge finish was very smooth and required no sanding. I will be cutting wood as I am just finishing the installation of a new cyclone dust collector. Will report on the results.

8547

[dvanr]

My understanding of X,Y movement in CNC machines was that the step direction commands are loaded serially ie X gets a command then Y gets a command ( and then Z if doing 3d). The commands are then executed on one move signal, so the steppers should move in concert on a 45 degree angle, not in a 1,2 fashion which would give a staircase pattern.

One other thought,

I haven't seen any complaints of chatter in the Z and either X or Y axis ie plunge or climb cuts.

XY chatters, does XZ or YZ chatter as much?

[Brady Watson]

Dick,
You are correct..the steppers DO receive step and direction signals individually from the controller. One thing to note on both Steve's Gecko-drive bot and the newer Alphas is that they use microstepping drivers, opposed to 1/4 stepping drivers in a PRT. Microstepping drivers divide each step into 10 microsteps, and 1/4 stepping does this in 4 steps. Microstepping is obviously smoother in operation and sound than a comparable 1/4 stepping machine. Now...let's not completely dog Shopbot's logic in choosing the Allegro 1/4 stepping drivers, because on a bipolar configured stepper, there is only 'detent' torque at each 1/4 step...so even with the silky smooth microstepping driver, you can still lose microsteps in between detents, or in this case 1/4 step increments. A microstepping driver will transform a 200 step/rev stepper into a 2000 step/rev stepper when microstepping, opposed the the Allegro's at 800 steps/rev.

Microstepping drivers do make a big difference with the saw-toothed pattern that you are referring to. It's not the "You move...then I'll move" action of the X/Y steppers creating the sawtooth pattern, but rather the 1/4 stepping action of the drivers doing this. The Alphas and other machines running microstep drivers show much less of the sawtoothing than a bot running the Allegros.

-Brady

[dingwall]

Mike, a larger cutter should improve on the depth of the scallops.

Gerald, when I upped the ratio, the scallop size shrank to acceptable levels.

Brady, are you saying that Microstepping drivers altough smoother are more prone to losing steps?

[Brady Watson]

No.

Running a microstepping driver on an open-loop stepper has no advantage over a 1/4 stepping driver when it comes to lost steps. A 1/4 stepping driver will always sit on a detent that will hold at rated torque (in theory)...a microstepping driver will also do this for you in 4 detent positions...the other 6 positions are in-between areas where the electical flux is not strong enough to hold the stepper at the same torque as the detent positions. A bipolar stepper is in fact, bipolar. Think of it as having 2 North and 2 South poles in a X pattern. Those are your detent postions regardless of driver. (or at least this is how I understand it)

-Brady

[gerald_d]

Very interesting discussion. I just want to add some numbers so that we know what a "step" means in real life....

A pinion gear diameter is around 1" (25mm) and one rotation of the pinion gear moves a carriage about 3" (75mm). A stepper motor needs 200 full-steps for one revolution.

Therefore, for a directly driven stepper:pinion, one full step equals about 0.015" (0.4mm) of linear travel.

When there is a 3.6 : 1 gearbox between the stepper motor and the pinion, one full step of the motor equals about 0.004" (0.1mm) of linear travel.

For a 45 degree move, the resulting diagonal distance is 41% greater than each of the x or y moves.

(The above figures are rounded off - they are not meant to be a precise baseline)

[sol]

Thanks for all the replies.

However, now I'm a bit confused. A few threads down, there is a thread, "Upgrading PRT Frame to Alpha Frame" where users stated that the Alpha Frame is quite a bit beefier. I have a milling machine and a mig welder... can I beef up the frame myself with bolt on attachments?

[gerald_d]

Eric, the good news is that the machine you ordered will come with the "Alpha" frame. The older (up to middle last year) ShopBots had a different frame.

[gerald_d]

On the "45 chatters issue", it has struck me how similar (in pitch and depth) these chatters are from one job to the next. (And looking at pics here at the Forum, even how similar they are from one type of ShopBot to the next.) For example, changing from a single-flute to a 4-flute cutter does not make the chatter pitch 4 times finer. Halving the move speed does not halve the chatter pitch. The ShopBots with and without gearboxes give rather similar chatters. Etc.

Throughout all these variables, there are two factors which stay fairly constant and I would like to test people's thoughts on these:

1. Gantry stiffness in terms of natural vibration frequency. I wonder if we get more chatter when the y-car is in the middle of gantry (most 'flexible' position) or when the y-car is near the end of the gantry ('stiffest' position). I think the older utility strut gantry and the newer alu extrusion gantries are very similar in stiffness in the x and z directions, while the newer gantry is certainly stiffer in the y-direction. (Steve's gantry is much stiffer in all directions and this is why I am particularly interested in what "45 chatters" he may be getting). However, if the chatters are caused by gantry resonance, then halving the move speed should make the chatter pitch half as fine, but this doesn't seem to be the relationship.

2. The one thing which is absolutely consistent through all setups is the pitch of the gear racks. Here I have a theory that the spring-loaded pinion gears are transmitting their "rough ride" through to the finished cutting face. Sometimes the roughness causes resonance in the gantry, making the chatter mark deeper, but the pitch stays fairly consistent. A ShopBotter once mentioned that reducing the load on the tensioning springs gave an improved cut, and this made sense. Tightly pushing gears against each other is a rather crude technique that will seldom be found in the engineering world. In the precision gearing world there are such things as anti-backlash gears (http://www.google.com/search?sourceid=navclient&ie=UTF-8&rls=GGLD%2CGGLD:2005-07%2CGGLD:en&q=anti%2Dbacklash%2Bgears) but they have a more complex spring tensioning system. We might get better cut qualities if we removed our turnbuckles and springs, fixed the motors rigidly to the carriages, and then shimmed the racks to get a good mesh to the pinions.

Any thoughts?

[fleinbach]

Gerald,

I don't think it would be a good idea to fix the motor rigidly to the carriage. As you say the pitch of the gear racks is absolutely consistent as well as the pitch of the pinion gear but the tolerance between the rack and the top rail is certainly anything but consistent. I have not done it but I would be willing to bet you will find a varying thickness of several thousands between the top surface of the rails and pinion gear. Of course this means once you adjusted everything in one spot on the rail and move it to another you would find it to be either to lose or too tight.

A secondary problem due to the soft metal of the rails would be constant maintenance in readjusting to keep a tight tolerance.

I too have been working on this problem since last August. I have several ideas I would like to implement but just have not had the time. If or when I ever get around to it I will post them here.

[daveiannone]

Gerald,
I believe you could , with enough fiddling around and patience get the racks set accurately enough if you were to lock the motors in place....I am not sure if Frank's second point about soft rail is true or not, but that could be worked around also......My fear about locking the motors would be that the spring loaded system has SAVED me a couple times that I made a mistake , and I question what would happen if you hit a clamp for example and had no springs....It would be an interesting setup to try though.

Dave I

[gerald_d]

Frank & David, what I had in mind was first to mount a dial gauge on top of a spring-loaded motor and to see how much variation there really is down the length of a rack. If the readings were fairly regular then shims could be placed between the racks and the rails to get a constant reading - before mounting the motors rigidly to the carriages. All of this in theory only, but it is something that I will seriously consider if the chatters are so bad that I think that I need to do something about them. For our work at the moment they are quite tolerable and I havn't even gotten to the very simple first step of putting a dial gauge above a motor.

If someone looked at those linked "anti-backlash gears" and wondered how the concept could be implemented for our tables; We could mount a second rack against the side of the first one, and use a wider pinion gear that will span both racks. The second rack must be mounted such that it can slide a small distance (0.1") axially along the rail, and once the pinion gear is engaged, the second rack is then springloaded, with the effect of widening the gear tooth that the pinion engages with.

[Brady Watson]

Interesting discussion, Gerald.

I totally get what you are saying about resonance with the gantry...However, I don't think that this is what is contributing to the 'steps'. I have a theory that, while the Alphas are much smoother and show less signs of these steps, they still are processing their code through the same CPU as the PRTs. The BlueEarth board that sticks up on the Alpha board is the same as the ones the PRTs used. Since everything from the PC to control box data stream goes thru this little processor, I would have a look at that.

A friend of mine has a Bot that is running aftermarket steppers and Geckos. He can cut at 2 or 3X what he was cutting on the PRT...and his acrylic circles are PERFECT! He is running a stock PRT gantry...but will be adding gussets to it in the near future. This leads me to believe that the control box is a real bottleneck and the BE board might just be what is causing the resonance because it can't calculate moves fast enough...In theory, it might just be 'pulsing' it's output and causing the steps...not sure.

-Brady

[gerald_d]

Pretty soon I will also have a 80V/Gecko/Mach2/PMDX-120 driving a pre-Alpha PRT and I am looking forward to seeing the results. This (http://www.scapenotes.com/notes/messages/2020/4721.jpg) (and earlier detail (http://www.scapenotes.com/notes/messages/2020/4709.jpg)) is what is lying next to me on my desk at the moment. I am not putting it together for better cut quality, I am doing it because this will be a simpler system for me to maintain out here.

[Brady Watson]

Nice! Are those 202s? (I don't think you need the capacitors with the 202s anymore...FYI)

Let us know when you get it up and running!!!

-B

[gerald_d]

Those are 201's, which are stocked over here. The 202's are wider by the width of the capacitor which is internal. Have a look at the G101 (G-Rex) being launched next month - there BlueEarth will be replaced by Rabbit. G101 will take USB and allows for encoders on the steppers - it will match Alpha performance. The dope is at the Yahoo list (http://finance.groups.yahoo.com/group/geckodrive/messages)

[Brady Watson]

I know...It's pretty slick! All you need now are some double shaft steppers and some encoders (500 cpr would do nicely)...BUT you will also have to factor in your gear reduction when you choose your encoder, and make sure that your PC can handle the increased step-pulse rate. I am not factoring in the G101 and the fact that it has an INSANE pulse rate AND the ability to run stand-alone after you have fed it your G-code. It has it's own processor in it...life is getting good!

My friend ran a non-gearbox stepper on his bot with no reduction and it really didn't have the resolution he was looking for. A 3:1 or even 4:1 reduction works nicely...and I have seen that machine rapid move 30 IPS (because Mach2 limits that as the max speed....I don't see any reason why you would want to move faster than that...but 'she's got more in her'...) Not to take anything away from the Alpha...they can rapid over 30IPS too without the software governor.




-B

[bleeth]

Brady and Gerald: I'm glad to see you continuing this discussion and getting down to specifics. The gist in practical terms to me seems to be that if one modifies their PRT similar to how Steve did above and as shown in the other thread one ends up with a unit that runs at Alpha speeds with the same stepper motors without the chatter and sacrifices only that it is no longer a "pure" shopbot and now runs G-code rather than SB code. I assume that this also means that one would no longer be using SB control software. With the research you have been doing on this and the resulting conclusions you are reaching,what "package" of currently available components do you like best, what would be the approximate cost, and how might I have a look at the controller interface?

Dave

[paco]

Hey Dave!

You might to lurk some at the CNC zone (http://www.cnczone.com)... and the forum (http://www.cnczone.com/forums/)...

and this link (http://www.artofcnc.ca) for Mach controls...

[bleeth]

Yeah Paco-I've been doing some of that. I see that Mach 2 is $145 but must run on 2000 or XP. My current operation computer in the shop is an old 98 runner and the office computer on which I have everything in my world loaded on is ME. The gecko site is down right now (seems to happen a lot) and the consideration is the total package in dollars and time down and redoing it all. It seems that it would still come out more favorable than the Alpha changeover but need a lot more "construction" than I've messed around with as far as building a new board etc. I can't afford to tear it all apart and then find myself in over my head on putting it back together again!! On the other hand I just lost a client due to "chatter" and that really hurts.

Dave

[Brady Watson]

Dave,
It's a real toss-up depending on your situation. Naturally if I had the $$$ I would upgrade to an Alpha in a heartbeat. The advantage is that I don't have to fabricate anything, I still retain SB support and I don't have to change my design or operating software. I would give myself plenty of time to learn all the old tricks all over again using Mach2 & G-Code if I went that route. My main interest in doing a Mach2 conversion is for my Bridgeport BOSS CNC project...which STILL isn't done due to time constraints. If you are not into doing R&D and can't afford to have down-time, then you just have to bite the bullet and go with an Alpha.

Let's get clear on what we are talking about...Steve, so far, has upgraded his PRT, retaining the shopbot 44/48v PRT power supply (took out of case), 4 Gecko drives, a 'breakout board' and Mach2 software. This combo gives you the advantage of using the Gecko's silky-smooth microstepping drivers. Cutting speeds are NOT increased, although due to Mach2 ramping tuning, you can increase your rapid moves a bit. You can very easily double your cutting speeds and rapid speeds by simply doubling the power supply voltage. This would put you right on the ragged edge of what the Geckos can handle @ 80v max, using the PRT Vextas.

Another option is to use another type of stepper that can move faster & with more torque than the PRT motors. The issue here is that you still need gear reduction and this is the major part that needs to be fabricated. A belt-drive reduction makes the most sense, but plan on spending around $150 each for these to have them custom made. With good aftermarket steppers in the 600 Oz range, AND 3:1 or 4:1 reduction, you can have Alpha-like performance...but no feedback...if you want that you'll have to invest in a G101 or similar to take advantage of feedback/closed loop performance...

Yes..it is all possible. Is it the right thing for everyone? ~ Probably not depending on your skills and needs, plus your workload. For those of you that want to take this on, it might not be a bad idea to make it so that you can go back to SB controls if things don't work out to your liking.

-Brady

[stevem]

Brady, I used none of the ShopBot electrical components in my conversion, other than the motors. The new transformer was purchased online for $29.00. It’s 45V 5A.
Cutting and rapid speeds are double that of the original control box. The original box was sold to offset the cost of the conversion. Higher voltage would give better performance.


I am also waiting for the G101 to become more “ignorant user” friendly, which would make it suitable for my use.

The real benefit to having a welded gantry is never having to adjust “square”.

[gerald_d]

On squareness, the Mach software has a standard routine for squaring a gantry. If you have home switches on both x-rails, Mach will drive, and zero, the two x-motors individually.

Anyway, changing the electronics and software for smoother control is a given, I wonder if we can explore some of the mechanical aspects a little further? Is the rough ride of the pinion-hard-against-rack an issue? Do people with ball-screw experience have similar stories about chatter?

[Brady Watson]

Gerald,
I don't think that there is a huge disparity between a ballscrew and rack in terms of cut quality, strictly in regards to chatter. Many CNC routers use R&P. A non-rigid gantry that flexes under load is not a good thing. You would think that a solid/rigid gantry would be prone to resonance from the rack, but I have found that a loose gantry is much worse because it allows the resonance from the rack to jiggle the Y-car in particular, if it is not really secure. I would think that the harmonics of the router would be far worse than that of the rack...and the gyroscopic action of that router when running will continually try to vibrate and move around in any slop you have in the mechanicals. I think that the ideal setup would run a ballscrew on the Z only, and R&P on the XY...and I still think that the majority of the chatter is a result of the electronics.

-B

[richards]

Brady,
From my cutting, I would basically agree with your statement, "...and I still think that the majority of the chatter is a result of the electronics.". However, I've noticed that there is chatter on single axis moves during the ramping part of the cut. Even though I've played with the ramp variables a little to find the smoothest cut, I still get chatter on those single axis moves. The chatter is not as pronounced as chatter from a multiaxis move, but it can be heard when the move is taking place and easily felt by running a finger tip along the cut afterwards. Single axis chatter has to be caused by machine vibration.

Multiaxis chatter can be caused by both machine vibration and electronic non-syncronization between the motors. You can see the chatter caused by the electronics fairly easily when cutting a 3-5 inch diameter circle. By drawing a circle on graph paper and then determining how many steps each axis has to make at each graph point, it shows that there are points where one axis is almost idle while the other axis is quite active. It becomes apparent that unless the steps are timed exactly, that you're going to have chatter. The chatter on the circle is almost non-existent at the 45-degree point of each of the four quadrants of the circle, the very spot when both motors are basically making equal steps. The chatter is most pronounced near the 0-degree and 90-degree portions of each quadrant when the ratio of steps between the motors is the highest.

It seems to me that to entirely eliminate the chatter, the mechanical side of the machine would have to be worked on (perhaps causing even worse grief when the router/spindle 'crashes'), and the electronics would have to be enhanced to produce pulses exactly when needed from multiple axes.

[Brady Watson]

Mike,
I agree. Also note that when you focus on chatter at this level, the runout of the router or spindle becomes a factor that may contribute to the chatter you are highlighting. For example, a factory-fresh PC router has .008" of runout ~ which will be directly transferred to your cut even with a perfectly aligned/square/aerospace linear motion setup.

-Brady

[gerald_d]

Brady, I wouldn't agree with the blanket statement that all PC routers have 0.008" of runout when new. Some of them probably have that much runout, while most of them have a lot less?

[erik_f]

I'm not as advanced as alot of people on this board...but in my case I have found most of my problems with "smoothness" not to be the limits of the machine...but either settings in my software...or, and this is usually the case, something in my machines set up on the mechanical side of things.

[Brady Watson]

Gerald,
It was not my intention to describe that as an absolute condition ~ only to point out another possibility on our quest for higher cut quality.

-B

[richards]

I'm going to need a little help understanding how much of a factor runout is in producing chatter. In my pre-spindle days, when I ran the PC7518 router at 3.5-inches per second and 16,000 rpm, each rotation of the cutter would have advanced the cut 0.0131 inches. Now taking 0.008 total runout as an example, I would expect to see 0.004 inch chatter marks every 0.0131 inches (assuming that the point of maximum runout was to the side of the cut and not along the axis of the cut).

What I normally see is a repeating 'chatter mark' pattern that is spaced from 0.1 to 0.25 inches apart. In carefully looking at some doors that I just cut, the marks are most obvious in the ramping portions of the cut and in the areas of the arch where one axis is moving much more than the second axis.

One other factor that hasn't been mentioned is the machining marks on the shopbot rails themselves. The rails have a repeating pattern that more closely matches the 'chatter' pattern that I'm getting than would a chatter pattern produced by the cutter advancing through the material (mathmatically speaking).

It seems that we might have several factors involved: 1. Flexing of the rails during the 'high torque' portion of the ramp. 2. Electronic induced chatter caused by the signals being sent to the stepper drivers. 3. Runout of the cutter. 4. Pattern following caused by machining marks in the rails.

[dingwall]

I was dissappointed with the lack of smooth ramping when I started using the Windows SW. I haven't had time to look into it though.

Doesn't the new Windows SW have S-shaped ramps?

If so, shouldn't it be easy to tune the stutters out of the ramps?

[richards]

I'm not sure if this is the correct place to post this, but, after spending the better part of the day experimenting with some old Oriental PK296-03 (12vdc, 1-amp) stepper motors that I never could get to run properly and some G212 Gecko stepper drives, I'm truely amazed. The steppers act like some of the expensive PK motors that I've used in other designs.

It seems to me that if you have an older machine (pre-alpha) with older drives, you might be able to significantly increase its speed for about $500 - $1,000 by adding a higher voltage power supply and four Gecko G212 drives.

If you're like me, when you first looked at the Gecko drives, you noticed that the G201 (the most popular model) required 2000 pulses to rotate the motor's shaft one revolution and you also noted that the drivers only had provisions for a four wire stepper motor. Both of those 'limitations' kept me from looking closer. The G210/G212 have built in pulse generators that can 'manufacture' pulses to enable you to use the more standard 200 pulses per revolution. A little reading and an email to Oriental motor cleared up the mystery about using a six-wire motor with a four-wire driver. Reading posts on the forum from other users of the Gecko drivers cleared up the rest of the mystery.

So, the bottom line seems to be that you can get 1.5 times to 3 times more speed from your existing stepper motors for a reasonable price. You will have to hack your machine a little but it might be worth the effort.

(If you're wondering, I'm NOT going to hack my Alpha. It works great the way it is - although I may add a second Y-motor and/or additional stiffening at some future point. I do have a need for a second custom machine to replace my horizontal boring machine, hence the reason that I started reading the posts by Gecko users which lead to a little experimentation and finally to a full-fledged mockup using four G212 driver cards, four old (ten years on the shelf) PH296-03 steppers, a PMDX-122 break-out board, a PMDX-135 power supply, a toroid 25x2 500 watt power supply and Mach2/3 software. The breakout board and Mach2/3 software should not be necessary for Shopbot users.)

If you're interested, drop me an email and I'll try to help without loading up the forum with non-shopbot posts.