[supahonkey]

I started working on a tube bender as my senior project. I am a senior at Indiana University Purdue University – Indianapolis. I am studying MET. My goal was, and still is, to build a tube bender and to make it CNC controlled via a Windows PC. My goal with this post is not to high light the making of the tube bender. It isn’t difficult to buy a JD2 bender and add hydraulic power; we’ve all seen it done hundreds of times. Instead, I will focus on the design and implement of CNC control.

I know a CNC tube bender isn’t practical for most, hell, it’s not even practical for me. This is more of an exercise to go beyond the norm, exposing myself to areas I normally wouldn’t have ventured.

I can't figure out why the forum is programmed to not allow the original creator of the post to not edit his post after a certain amount of time.
Here is how my system works: I have a Grainger pump, relief valve, and Dayton motor. This connects to a Vickers directional control valve. The valve arrangement is a spring center, 3 way tandem form. What this means is when I release the foot pedal, incoming fluid from the pressure line is automatically dumped back to the reservoir. Ports “A” and “B” are locked. The fluid then flows through hydraulic hose into a Parker 2.5”x24” cylinder. I am proud to say that all mechanical components of my bender are made in the USA.

I am currently in the beginning of the CNC implementation. I purchased a breakout board and a relay board. The breakout board serves two purposes: it allows me to interface the valve and rotary encoder to a parallel port cable. It also serves to optoisolate the I/O devices (valve, limit switches, stepper motors, etc) with the computer. If a surge occurs, $1 chips on the breakout board will get destroyed, not my motherboard.

I have some pictures to start off. As work progresses, I will post more. The project is due on December 15, leaving little time.

My bender as it sits now. Finish work is still required. I have to weld in a shelf for the middle. This will store tube templates and dies. I also need to plate with 20 gauge steel. I am going to powder coat the whole thing once it is complete.









I will have many more pictures to follow in the next few days.

[supahonkey]

Here is my breakout board:



Here is the relay board:



Offsetting the boards from the case with motherboard risers:



The box with the boards mounted:




I haven't decided whether I will make a small 5V DC power supply, or if I will just tap off the PC. I like the idea of having a separate power supply for the boards. Both boards require 5VDC to run.

[jamiedaugherty]

Where did you get that breakout board? I'm going to need one of those for use with my Microkinetics Optistep Plus card and haven't found any decent ones yet. The one you are using looks pretty nice.

[DSL PWR]

It's printed right on the board: http://www.cnc4pc.com

[skippy]

Now here's a project that got my attention. I'll be watching closely and I wish you all the luck with it.

[Al_The_Man]

Do you anticipate getting the positional accuracy with a directional valve rather than a proportional as is traditionally used for this application? how are you going to get the arc position feedback?
Al.

[powerglider]

great project! I'm very interested in the results and comminicating after you are done.

I would love to try and implement the same cnc response on my Diacro #3 tubing bender in the future.

How are you ending the stroke? Is the stroke end time based? Or limit switch based with the limit switch position manually adjusted in a mechanical way? Or are you measuring stroke or on the central bending die as an arc?

[supahonkey]

Die position will be determined with a rotary encoder. It will attach to the center pin, which rotates about the same axis as the die. End and start stroke will be determined by one of two ways. I am not familiar with Mach3 and its abilities to handle rotary encoders. Method one will either use small normally closed infra-red limit switch at the begininning and end stroke. Method two will use the rotary encoder. Simple if then statements will constantly analyze die position. If it is near the end of the cylinder stroke (100ish degrees) I will have Mach3 turn off one of the relays, closing the valve. If Mach 3 can handle simple scripting then it will be easy.

[Al_The_Man]

Mach does not see the encoder, there is no feedback to the control.
Normally this is the responsibility of the servo drive in this system.
If it was a proportional valve, a step/dir to analogue amplifier could maybe utilized.
Al.

[supahonkey]

I don't know much about Mach3, but I did see a selection for the Encoder and a selection for the pins for input. I also saw a pulses per rev option. I can't imagine Mach3 can't receive input from a rotary encoder! Please explain how a proportional valve would work better? Don't they just limit flow? They don't offer feedback do they?

[Al_The_Man]

Most commercial benders use a proportional valve, also I used one in a retro-fit conversion of a semi manual machine, it is actually a servo valve, the spool is shifted a proportional distance in either direction in order to control the hydraulic flow to move the cylinder proportionately, rather than just on off.
The valve itself does not offer feedback, but an encoder or other means is fed back to the control for PID control.
The control for the valve is usually a servo drive amplifier.
I made mine from a National Semi LM759, but I was using a Motion Card so I had analogue output to the LM759.
I have not used Mach either, but the servo encoders are fed back to the drives, The manual Pulse Hand wheel may go back, however.
But even this may prove a bit difficult to get any kind of precision with on/off control and not servo.
Al.

[Kevin Taylor]

Nice project If you have used a JD type bender you must know that a slight amount of over bend is required Ex 93dg to get 90 because the memory of the material I havent looked at the comerical CNC bender's enough to know how they do it but you might be able to treat it like backlash comp on a mill ot lathe A second note to do a large bend you have to move the pin in the die even if you have a long cylinder some of the high end bender's move both halves of the bender somthing to look at . Keep up the great work Kevin

[powerglider]

in manual bending we need a mechanical stop and as consistent a movement as possible right up to the final contact into the mechanical stop to acheive any consistency.

Wouldn't you need the same in your setup? Or is the stop of hydraulic fluid so instant, precise and repeatable that you don't need a mechanical stop? I've never observed a comercial cnc bender close enough to see if they use a mechanical stop.

[Al_The_Man]

Wouldn't you need the same in your setup? Or is the stop of hydraulic fluid so instant, precise and repeatable that you don't need a mechanical stop? I've never observed a comercial cnc bender close enough to see if they use a mechanical stop.

Commercial benders use the proportional valve in conjunction with either an analogue pot or encoder on the bend Former.
The position is then programmed just like a servo motor, to precise angular position, an overbend factor can be programmed in for spring-back.
This is the method used by Pines Benders and others .
Al.

[supahonkey]

I have no idea whether my fluid will be so instantly perfect. I will first test accuracy by using the rotary encoder. It's easy to loose focus on the purpose of the machine. Bending tube is not like machining; holding tight tolerances is not only unnecessary, it's not possible. The JD2 bender is sloppy, it only costs $299! The JD2 bender flexes and twists. The pin is steel against steel; there are no bushings in the system. I don't require anything more accurate than +/- 1 degree.

My purpose of a limit switch is to home my cylinder and to control end of travel. If I don’t have limit switches all that happens is the fluid opens a relief valve and dumps to tank (upon end of travel).

Using limit switches would prevent me from having a CNC machine. I would have to move mechanical switches to a certain position. That system wouldn’t even be a NC machine. All I can do is experiment with each system and see what works best.

Al:
How would the position of a servo within a proportional valve describe die position? I understand that the servo controls flow by partially opening a valve, but how would that translate into die position? I looked at proportional valves, but a Vickers was around $750! My cylinder actuates slowly and the valve reacts quickly. It will be interesting to see my accuracy using a non-proportional valve.

[supahonkey]

I got a lot of work done. Unfortunately it was nothing worth positing pictures. I built the PC, installed XP and Mach3. I did manage to blow the 5V+ voltage regulator (7805T) on my relay board. I used the RJ-45 cable to connect it to my relay board. The wiring diagram states that Pin 1 is 12V+ and pin 8 is ground. I have been working with computer networking equipment for many years. Apparently what I consider pin 1 is what CNC4PC considers pin 8. I wish there was an actual part on the board or print to say what pin 1 is. The front face of the regulator exploded and hit me in the face. It actually drew blood! The part number was blown off, therefore I coudln't tell what it was. I realized a few hours later, (after going to the electronics store) that I had pictures of it from when I bought it.

[Al_The_Man]

I think you at least want encoder feedback, this is normally done off the pivot pin or moving arm of the bender, I recently retrofitted a rolling machine that used this principle of tracking the encoder on each axis rotation point and when the position was in coincidence, the hyd. valve turns off.
It is crude, but worked for this application.
I used a servo feedback card so I could closely track the encoder.
I know the proportional valves can run in the $k's .
Al.

[VWSatOz]

what about a cnc controlled stop screw? move the stop on a heavy screw to get stroke limit you want then bump into it using on/of valve. I saw this on Curvit mc years ago

[NEATman]

Along the lines of the stop screw, how about a stepper motor mounted below, with it's axis aligned with the pivot - but not connected. The stepper shaft could have an arm with a limit switch mounted to it. The stepper would rotate to the angle needed, then the bender would rotate until a pin or arm on the bender hits the "swinging" limit switch mounted to the stepper, which in turn dumps the pressure from the hydraulics, and ends the bend. You could then drive the stepper in either direction, and even add a second "home" switch to home the axis. I would also suggest a hardstop to prevent the stepper from ever rotating more than 1 rev, as you will be rotating wires along with it...

NEATman
Keith

[skippy]

Hi everyone, sorry to hijack the thread but I'd like to set myself up with a cheap mandrel bender and also watch this project of yours very closely. Buying one from the USA is out of the question due to the super high dollar value at present and also the surcharge for metric dies. I found this chinese bender the other day and it seems like it might be a mandrel bender but I'm not sure as the picture is not close enough for me to decide and you know what it's like when you ask "is it a mandrel bender?" (no answer). I know what the cheap non mandrel benders look like and I also know what mandrel benders look like but this one.... I'm not sure of. Could one of you please take a look and tell me?
thanks

http://www.alibaba.com/product-gs/20...showimage.html