How would you use Mach to control a bender?

[panozeng]

I posted the following in one of the generic CNC forums on here, and have since read a few posts where people suggest using Mach to control a bender. Here is my post:

Ok, I have an older mandrel bender that has been updated with an Automation Direct / Koyo PLC but it is still a manual machine. The bend angle is not electronically measured or controlled; instead it uses a mechanically adjustable stop and limit switch to stop the machine at the desired bend angle. It is VERY obnoxious to get this dialed in - I want digital control. The machine still uses a control panel from the 60's with industrial type switches as well - most of which have very dirty contacts and can be intermittant.

I would like to upgrade the machine in 2 phases:

1. Add a rotary encoder and some form of digital angle display that could toggle between the current bend angle of the machine and an adjustable set point. (Or have 2 displays, one for current angle and oe for setpoint) I am thinking about trying to do this part with the PLC that is already on there - but the high speed counter alone is around $300 to read the encoder, which is another $90 - not to mention a display unit at $135 and some rather expensive software to program it (and I would have to learn how!)

2. I plan on adding a tube positioner with stepper or servo control. This would be 2 axis - one for tube rotation and one for linear position. I would really like to have a desktop computer controlling things at this point so it would be easy to program. When bending you position the tube in the 2 axes I just mentioned then bend to a certain angle and do it all again.

I can handle all the mechanical parts no problem, but am struggling with what way to go as far as computer controls and software. Everything I've seen so far on this website is for XYZ control and doesn't really translate well to benders. Also, the machine has a BUNCH of solenoids that need to be operated by the controls - I believe there are about 16 limit switch inputs and 12 solenoid outputs on the current PLC being used.

I would like to hear ideas on how to control this beast. By the way, it is capable of bending up to 6" OD steel tubing - so it is a beast!

So, how would you implement this in Mach? Is there a way to reconfigure the axes to work well for bending? One axis would be a linear feed. One a rotation angle. And the last variable is bend angle which involves switching a hydraulic solenoid on until the rotary encoder reached the desired bend angle. I guess this could be done as an axis?

Thank you!
Jason

[CNCgr]

You think you can control Bender? Ha!



On the serious side:

Originally Posted by panozeng One axis would be a linear feed. One a rotation angle. And the last variable is bend angle which involves switching a hydraulic solenoid on until the rotary encoder reached the desired bend angle. I guess this could be done as an axis?

Is there a control for reverse? Maybe you could output the step and dir pins for the "bending" axis to a PLC, and if the dir pin was high and the step was pulsing it would go forward. If the dir pin was low and step pulsing it would reverse.

Nikolas

[NC Cams]

You could easily set up a 555 chip running as multivibrator to provide a clock to output the necesary pulses (either direct or via a pulse count divider chip.

A thumbwheel programmable up/down counter (CD4029 or equivalent TTL chip) would provide the step count to bend whatever angle you prescribe. A Red Lion counter (they might have one that up/downs) would be your non contacting digital gage to count the step pulses. I like the CMOS stuff for better noise immunity, trouble is, its not looked at in high regard by the computer boys due to its power and speed issues, but for this, you don't need megahertz speeds now do you???

Whether you go with a servo or stepper is your choice (gecko's would do). Controller could be built for peanuts (under $50) and fit in a small kids metal lunch bucket if they still make them. Power for control could come from at 12v battery or small wall plug deal. You'll have more issues coming up with power for the servo/stepper than the control.

For what you want to do, you don't need a PC or a program - a simple presettable pulser would easily do the trick.

[Al_The_Man]

Originally Posted by NC Cams . You'll have more issues coming up with power for the servo/stepper than the control. .

Traditionally Pipe bending machines use Hydraulics & servo proportional valves with encoder feedback due to the extreme pressures required, for example bending 6" OD tube as the poster requires.
Al.

[NC Cams]

The step/direction part of the system could still be handled by a simple pulse generator instead of Mach/PC or sophisticated PLC.

Whether you'r bending 6" tub or 0.060" wire, the only difference is the scale factor of the force involved and system tuning and proportion establishment.

I got the 4029 counter idea from some previous counting based work we did with a pulsed batter charger. Simple deal of sending out so many preprogrammed evenly spaced charge pulses then a single discharge pules, reset, repeat until charge sense said your were done.

A 3 to 6 stage U/D counter would provied the number of steps (degress in this case) needed for a bend. Simply figure out the step versus movement relationship. THen set the thumbwheels to the desired degree of bend (with windage factor for individual pipe size memory).

The counter would send out pulses and count down as the high force side moved as prescribed. Once you go to 0, the count would stop, hold for prescribe time then return the same count to home by changing the U/D pin on the counter to go the other preset direction.

The feedback to the stepper or servo system that does the bending (hyd or electro mechancal) would work the same asside from mounting. I do not disagree that this side of the system would need to be robust.

The point being that you wouldn't need a PLC or a PC/Mach to do the "thinking". It is essentially a simple, single axis motion of going so far out and back out. Simple step and direction. Send out the steps, set direction via 1 or 0.

However, due to the force factor, you'd probably be better off with a servo than a stepper. This would require a bit more sopnistication as you'd have to feedback to the control if you moved the amount you wanted. Coming up with a closed loop feedback system that would automatically compensate for the hysterisis is not something that I think a simple step/dirn system would/could achieve.

In the concept:

Red Lion counters, cheap (they show how to drive reset in their catalog).

CMOS chips, cheap

PCB, cheap (via expess PCB)

F/B encoders, market price

Hyd servos, market price

Building on a breadboard what does the work of a computer and fancy programming, priceless.

BTW, this is exactly the sort of control that we bought from a guy for our cam grinder - only it had a long obsolete G&L computer that isn't supported and huge array of discrete and CMOS computer chips. Cost nearly $100K when new, we got it for near scrap.

What is the size of a decent console TV can now be fit into a couple of IC'S (L297/298), a handfull of fets (vs clunky darlingtons) and a similar handfull of CMOS chips.

Hope the ideas are found to be of a stimulating nature.

[panozeng]

Well, I am not totally following the 4029 idea, but the Red Lion counter was a good tip. I don't see why the controller needs to send out a pulse to control the hydraulic solenoid. Unless - I wonder if you could run a conventional hydraulic solenoid in a PWM manner to ramp down the flow prior to stopping the flow? Probably way too violent with with pressures involved...

But anyway, all of the current functions on the bender are hydraulically driven and controlled by solenoids. You ought to see inside the hydraulics cavity - a dual outlet hydraulic pump driven by a 20hp motor, a bunch of large Parker solenoids, and a massive amount of bent stainless steel lines connecting it all. It currently has a hydraulic ram for clamping on the tube, a ram for moving the pressure die in, a ram for the pressure die assist, a ram for the mandrel advance/retract, and a big mama of a cylinder (with 4 ports on it - 2 in the middle) to move the bend head. Right now the Koyo PLC uses inputs from a pushbutton switch box to control all the solenoids that control the cylinders. It does have a slightly automated sequence program as well - but it is VERY dependent on having a lot of limit switches dialed in exactly to not mangle the tube and/or tooling, so I don't use it as often.

For my immediate desire of having a digital display with a setpoint Red Lion would be easy to set up for about $260. They have an up/down counter that works with a quadrature encoder and gives you setpoints that trip relays. I could use that relay with all the current controls to replace the current adjustable microswitch stop. The display has a correction factor to multiply the number of counts by to get the disired display, so I could use a fairly high pulse per rev counter and get nice angle resolution. This would make the machine MUCH easier to set up and maintain accurate bends while I contemplate ideas on adding NC to a tube positioner (feed and rotation).

Thank you,
Jason

[NC Cams]

The 4029 is an up/down counter chip - it or a variation of it is in the Red Lion.

You program in a binary coded count with thumb wheel switches and a direction (up or down, 1 or 0) and feed a pulse train to it. IT then counts up or down until it reaches your set number of count pulses and stops. Reset it and then give it a new GO signal and it does it all over again.

Depending on how you setup the carrover or carry in logic, you can get it to count down, dwell, return, whatever. By stacking them in sequential order or getting them to trigger another U/D only with a different count pulse rate, you can get different input vs output speeds, dwell variations, whatever.

Essentially, Red Lion is doing this for you only in a pre-packaged industrial suited package. All things considered, probably the better way to go for you.

Don't forget to check out the Red Lion digital panel meters (CUB and SUBCUB's). Nice simple self powered devices that run pretty quick too. Their LED ones are nice too. Nice stuff to integrate as they have a lot of appications info in their catalog. Wouldn't hurt to have one.

[Al_The_Man]

I discovered an interesting item some time ago when looking for Hydraulic servo valves, they traditionally are very fussy as to oil filtration due to being easily made inoperable due to oil contamination.
There was one company that overcame this and not only produced a cheaper valve (some parker valves are $7,000.00) but made it possble to use an ordinary servo amp.
The construction of the valve used a stepper motor to move the spool instead of the traditional 2 coil differential pilot-valve method.
If any one is interested I could look the name up.
Al.

[jderou]

Originally Posted by CNCgr You think you can control Bender? Ha!

HAHA! LMAO. Gotta give credit where it is due! ok now back to the post.

[panozeng]

Al - I would definitely be interested in the name of that valve you found. With the bender's current on/off solenoid she stops quite violently at higher bend speeds. In fact, I am going to have to anchor this machine to the floor - it walks the opposite end from the bend arm about 1/2" every time you hit the stop at middle to high bend rates! It has no current provision to be mounted, so I'll have to add some tabs. Actually might make a frame for it to sit in that has levelling feet as well.

[HPbyGD]

Don't they put a air chamber similar to what you put in a house to keep the pipes form hammering just a lot stronger to take the pressure to take the hammering out of hydraulic equipment?
Gary

[Al_The_Man]

Originally Posted by panozeng Al - I would definitely be interested in the name of that valve you found. With the bender's current on/off solenoid she stops quite violently at higher bend speeds. l.

I will try and find it, it has been a while, the other thing that some do with on/off type valves is when the end approaches, there is a paralell flow control solenoid kicks in for the last few degrees for slow approach, the catch is that you need the feedback or a settable limit switch to detect the correct point to slow down.
Al.