Belt reduction suggestions

[Jtucker7000]

I have an old plasmacam table that I've modified with new servo motors,gecko drives, and mach4. I've struggled with the mix match parts and lack of knowledge trying to improve cut quality. I intend to build/buy a larger more conventional table in the future but would like to make the best of what I have until then. I used the existing Belt reduction it came with and it's still the original rack and pinion. On the large axis it has a 10 tooth XL pulley on the motor and a 48 tooth XL pulley I assume this means 4.8:1 reduction. The small axis has the same pulleys but are 10 and 24 teeth 2.4:1. I looked on the website at parts and noticed the replacement parts are a 40dp style acetate pullies that are 16 tooth and 120 tooth on the large (7.5:1) and 16 tooth and 60 on the small axis (3.75:1). I'm new to all of this but that seems like a significant change or maybe I'm not understanding something. The diameter of the gears seem similar but the 40 dp pulleys have much finer teeth. So as it is I'm getting cut quality well enough to sell a few things but the cut edges are wavey. Motion seems smooth, I don't know any specs for the original motors but the ones I have are nema 23 350 oz/in peak torque and 50 oz/in constant torque, 4600 no load RPM. Currently my best quality cuts on 14 guage is around 80 ipm. Does the current belt reduction seem appropriate or would I benefit from the
Replacements? Or maybe something completely different. I've contemplated moving to steppers to eliminate servo tunning as a possible cause and direct drive. The variables seem endless I'm just trying to narrow down one at a time.

[acourtjester]

I don't know about Plasmacam tables, have you ever had smooth cut edges. If you have then you have a motion problem, loose setscrew, poor belt tension, it is hard to look at the movement to see a problem. A simple thing to check things is to mount a ink pen on the torch and mark on paper in place of cutting. NO wasting of metal and the pen will give you a thin line to check movements.
Now the main difference between steppers and servo is the servo has a feed back to be sure it move the amount it was told to, stepper don't.
You may need to redo the axis calibrations to be sure things are moving correctly. If you change pulley sizes you will need to do this also.
From what has been said Mach 4 is not for plasma.

[jimcolt]

Properly sized servos most often need about a 10:1 reduction (there are other factors that would change this, but most cnc tables are in this range) and servos, again properly sized for the mass....typically are 3 or 3.5:1 ratio. This allows you to take the most advantage of the speed torque curve of the servo (wide and higher rpm) and the speed/torque curve of a stepper (narrow torque curve, best torque at lower rpm).

The Plasmacam that I have had for 14 or so years cuts incredibly smooth with its original electronics and motors. Jim Colt Hypertherm

[tcaudle]

I don't think wavy cuts are from the wrong belt ratio.

Belt ratios are only part of the equation. YOu have to take into account the drive gear DIAMETER (DP) . It steps UP speed at a ratio of PI times the diameter. Rotary torqe decreases with the same ratio in reverse and linear force is rotary torque divided by the radius of the drive gear. Resolution decreases by the smae ratio as the speed increases. Then you factor in the belt reduction ratio it it does the opposite (increases torque, increases resolution and decreases speed. I have no idea on the motor specs on the PCAM but they are high voltage DC brushed servos.

Something is wrong because 14 gauge at 45 amps should cut at about 225 IPM. Servo tuning CAN cause some motion problems. If the gain is too low and the dampening to high the servo gets "spongy" and will overshoot and even oscillate. The Geckos can be tuned off the table by clamping one to the table , putting a 4 to 6 inch long lever arm on the shaft and turning the three pots all the way down. Power up the motor so it locks and increase the gain (P) pot a little and push down on the lever. The motor should resist your trying to move the shaft. move the gain up until and feel the resistance increasing. At some point it will start to oscillate. At that point de power the motor , turn the gain down a little and re-power . Increase the dampening (D) pot) and take the gain up past where it was before . when you have it pretty stiff reach and thump the lever . It should bounce a few times then stabilize. if it bounces more than two times turn up the gain but leave the dampening alone. Keep doing this alternating between gain and dampening and thumping the lever to cause a 'disturbance and getting it as stiff (as few bounces as possible ) without it breaking into oscillation (watch your fingers! Oscillation can be violent) On those motors turn the torque limiting down to about 1/4 full turn . as a final setting very carefully advance the I (integral) pot a tiny amount while hitting the shaft lever to see if you can get it even fewer bounces. We use little or no I on the smaller servos. The fewer the bounces the less over shoot from inertia direction changes.. Once you have the servo loop stable to the load changes introduced by hitting the lever then you can move them to the table.

What encoders are you using? IS this a parallel port setup with MACH3 (which will limit your upper motor RPM due to the max kernel speeds) Pulse train out of a parallel port can be pretty jerky. I would say setup some tewst cuts with the torch off where the gantry is making lots of back and forth and sise to side moves and put hand lightly on the torch top and "feel" how smooth the moves are.

A loose Z or torch mount can be a source of wavy cuts if the rest of the gantry is tight.

[Jtucker7000]

I've seen newer plasmacam tables that look great mine is the first model and I got it with bad motors and a burnt up control box. They've all but stopped any support for it and parts are only what is still available in stock. I assumed it would be best to go this route. The encoders are cui amt102 with adjustable resolutions mine is set for 500 cpr and as I understand really means 2000 because of quaduature? The original optical encoders it came with are listed at 96cpr. It's a mach4 setup with a parralell cable using a c11 BOB I think from cnc4pc. My pulse rate in Mach is set at 35000hz (I've adjusted this to both extremes with no major difference except too low and it stops movement) In motor calibration setup both x and y are at 1918.2500 count per unit, 325 velocity, and 35 acceleration.The spur gear measured across is .875. I measured out 1' marks on the table and and adjusted the Counts per unit until it traveled the proper distance, I didn't expect both axis to have the same number but it seems accurate. I intend to move to vital systems setup that's Ethernet and adds the thc functions to mach4. I haven't seen many people with this hopefully it's worth the cost. I'll try the suggestions on loop tuning, i tried tuning both on and off the table last time using geckos method which is similar. One thing I did notice is on the motor I attached the drain wire to the motor from the shielded encoder cable and a hum I thought was oscillation went away but performance had no change. I've tried to methodically adjust all the above variables but I end up with the same result. If I change the encoder resolution all the other settings seem to move proportionally ending in the same result. As far as the "T" trim pot on the gecko as I was told fully ccw is max and cw is min opposite of all the rest, and I thought that was an amperage setting(0-20) mine is less than 1/4 turn from the fully cw setting. The T setting never seemed to do anything I have the dip switch for max torque for 1 second off. My x axis has a plastic material on the slides no actual bearings that I see, it has a small amount of play side to side and the home made torch mount isn't great, but I had the same result with the factory holder. I'm planning to move to the velox z unless there's a better suggestion. I appreciate the responses I spend about 2 hrs a night reading on this forum and others, I keep thinking I'll find something I'm missing. I'm sure it's a combination of problems and I'd like to spend my small budget in the right ways until I can afford better. The pictures I attached show it at its worst that was 120ipm on the 14 gauge the 1/4" piece was cut at 70 and I think looks ok. At slower speeds the wave and divets aren't as pronounced. Thicker parts seem to cut cleaner but they arent very intricate.

[tcaudle]

At slower speeds the wave and divets aren't as pronounced. Thicker parts seem to cut cleaner but they arent very intricate.

Yep, Wobbly mechanics. Possibly oscillating Z from THC. The faster you cut the more inertia you have and that can cause vibration on any thing that is even slightly loose.

You can't get a Prom Queen using Dollar Store lipstick. The torque control is basically just current limiting so until the motor is stalled and starts drawing max current it won't do anything. Those small servos won't take 10A of continuous current...they quickly overheat and let the smoke out.

The Velox is okay but its pretty deep and wide so you are going to lose some cut area. Then there is the added weight.....

How did you get the Promo to work with MACH4 running a parallel port? Are you sure you don't have MACH3....that works with the Promo. MACH4 has no THC hooks in it to my knowledge.

In my humble opinion you will get more for your money if you focus on the mechanics and get all of the slop out of it. Without that you won't get better cuts with a different control system. Its like having a car with loose steering, bad shocks and worn bushings and dropping in a new fuel injected motor and expect the driving and handling to get better.

[Jtucker7000]

It's Mach 4 for sure I'm using the proma sd which is a stand alone option with no control or feedback to Mach. I re-zero after each cut. I can't turn it off and no anti dive functions which would help. I do notice the z is constantly moving maybe some adjustment on the proma would help. Vital systems is the only motion controller that add thc controls for Mach 4 that I know of that's why I've been looking to move that direction and ditching the parallel port, but it's around $1300. Did all the numbers I'm using for Mach and encoder resolution sound correct? Is there a slimmer z-axis you recommend? The one a I have is new but probably not the best. I'm hoping to move to a machine torch to help with mounting, I haven't seen an aftermarket hand torch holder that was reasonably priced. Do you recommend any specific mechanical components I could upgrade utilizing my existing gantry or is the recommendation to replace the entire gantry with something like precision plasma offers? I see a lot of people using pbc ivtaaq linear v rails I thought about bolting one to the existing 3x3 tube but I'd still need to fabricate the carriage, and that would only improve one axis. I have about $1000 I could put to upgrades. Sorry for the 1000 question there are so many options out there its hard to know what works. I have already replaced all the cam followers, springs, and cleaned/replaced pulleys and gear racks. I also put loctite on all the set screws. I've seen carriages on eBay for nema 23 motors that ride on 2" or 3" tubing, they would be similar to what I have just new... I welcome any suggestions

[tcaudle]

The SD is as close to not having a THC as you can get . YOu have NO control over it before or during a cut. My option, but gathered from other posts on other forums from SD users. One look around the finished table vendors and the builders here, you won't find very many trying to use MACH4 or the SD Promo to do plasma cutting. The authors of MACH4 openly stated they did not consider Plasma as a useful market segment over two years ago when they stopped support or development/fixes to MACH3. The biggest impediment you have is your thin budget. Its gotten you to where you are now and its probably going to leave you there. Honestly to make that machine into a good cutting machine is going to cost you in the 3500 to 4000 range. You have already started down a road that may be painful to deviate from. You have the bones of a decent controller but the mechanics need help and the front end controls need help. Perhaps rethink the whole approach (including using MACH4 ) might be in order. Do some research on other options like the MP3600 - DTHCIV front end control that will interface to the G320X drivers you have

[RAD]

I would recommend looking into UCCNC + UC100 (superior to mach) and going with the minithc