To mach3 or not mach3, opinions requested

Hey all,

I'm Ken Cameron from Austin Texas. I am a former CNC machinist (mostly early 90's Haas mills), currently a metal fabricator and business owner. Most of what we make is used in our business, as such we are our own customer. This gives us a lot of flexibility when making machine tool purchases. I recently purchased a ~1980 KMB1-M from Brian Greul in Houston Texas. Thank you Brian for all your help getting the machine loaded on my trailer.

I am a complete newb with the the the BX controls that this machine seems to have.

The machine was in pretty rough shape when we got it home. The monitor went out. I was able to hook up an old Sony Triniton TV up to the video out signal to get a functioning screen with out having to buy and OEM replacement for something like $790:p. We have already replaced the spindle control snubber board for 279$ and fixed a bunch of damaged wires what were burned up when one of the snubbers got extra hot and crispy. Currently, the y and z axis jog just fine, and the spindle turns on and off, and I can move around through the control system. At this point, I seem to be having trouble with the X-axis, getting a motion error when I try to jog the machine in that axis. I'll be working with Shane at Martin CNC to diagnose that one as soon as my new multimeter arrives.


Long story short, I am wondering if it might be a better route for me to try to upgrade the control in this machine rather than struggle through learning it's antique control system, and trying to maintain 35 year old hardware. I am entirely intimidated by the prospect of an upgrade, but I think that I could do it with enough free or cheap advice and support. Learning the control might be a worth while prospect, but I am concerned that I won't be able to make any useful parts on this machine for my business.

When Brian owned the machine he used BOBCAD and some sort of internal windows software to DNC the g-code to the machine. I'm pretty sure that I can do the same once the X-axis issue is resolved.

So in making this decision:
1) I need to learn more about if my x axis is dead in the water, has a loose connection or what.
2) I'd like to better understand what a my machine is capable of doing in its current configuration. (Can I profile out the 2d parts that I have attached? Can it do 3 axis simultaneous interpolation?)
3) If I were to do an upgrade, What would be the best route to go? (replace servos with steppers?, keep servos and use Granite devices drives to control them.
4) Is mach 3 the logical choice in control systems? IS flashcut a good option, if some what more costly.

Sometimes a picture is worth a thousand words, so I've attached a few photos of the machine to give the circle of more knowledgeable folks that I am calling out to some more back ground.

thanks in advance,

-Ken Cameron

While it's pretty easy to be intimidated by the prospect of an upgrade, IMHO it will be the best option for the long haul. If the older controller can operate, it wouldn't have any problem doing 2d profiling, pocketing, etc, but I doubt that it would perform 3d operations very well. It may be capable of it, but usually controllers of this vintage have limited capacity for memory, and they have to pull the code into memory, run the code, the get more code over and over. This is usually a slow process compared to the programs it can fit completely into memory as most 2d type operations likely would (at least that's been my experience).

The servo drives look very recent and are probably good. Can you take some better photos of them so that all the text on them can be read?

My personal preference for Mach 3 CNC controllers is CS Labs in either the IP-S or IP-A versions. The IP-S is a step/direction controller and the IP-A is an analog controller. I've installed several now and they are both highly capable and very reliable. They are ethernet based controllers, but with a more industrial 24volt system for inputs and outputs which most Mach 3 controllers lack. Judging by the type of machine you have, either of those controllers probably has enough inputs and outputs built in to handle all the functions you'd need, and I'd venture to guess you'd have a few that go unused.

The main thing that saves time, in my experience, when replacing a controller is to just take everything out right from the get go. The first time I retrofitted a new controller into a machine, which was a bit more of a complex machine than this, I tried to document what the existing system was wired up for and how it was wired. As it turned out, almost nothing stayed the same and all the time I spent doing that was wasted. I think it would have cut my installation time in half to just tear it all out and then install the new controller.

You should expect to spend about $2000-2500 to replace the controller with a CS Labs system. In that estimate I'm including the controller itself, the wire needed, wire ducting to keep it tidy, the computer needed to run the machine (something like a basic touch screen all-in-one perhaps), and Mach 3. You should also figure perhaps 100 hours or more to install it, given that you are learning. I know that if I did a project like that today, I could probably install it in about 3 days to fully operational, so 25 hours is probably the time an experienced installer could do the job and verify that all the operations are working correctly. One thing that will guarantee that the time to install takes longer is using old components. If there are bad relays, or bad motors, etc., you'll be chasing those issues around for a lot of extra hours. Use as many new components as you can afford as it will save enough time that it's probably worth the extra expense.

>>>IS flashcut a good option, if some what more costly.

Flashcut is an excellent option. Frankly, many who have gone that route will point out that by the time you purchase all the extra aftermarket components to make Mach3 actually work reliably, you've spent nearly enough to buy Flashcut.

If I were you, I would give them a call, tell them what you have and let them advise you as to what options they would have for you.... and I am sure there will be a few. You may be able to use the drives and motors you already have, and that means you just need the software and signal generator. With that, you could easily explore what may be going on with your existing drives by swapping them around to find out which may be less than stellar.

They also will have plug and play systems...... Obviously one pays a higher premium for that, but you know, a member in this forum bought a complete plug and play control for his Grizzly mill and was moving the machine around in just a few hours. That is impressive given that there are other discussions of the same mill that go on and on and on, taking months to get in motion. It will come down to how much money you have to plunk down to get it done.

I've bought and used Mach..... Sorry, not a fan. It is really an awesome "control software"...... for $150, but that's about all I have to say about it. Of course someday, they will have a new version... so we are told. Could be the best thing since sliced bread for all we know at this point.

I have a Novakon that came with Mach, and after a few weeks of diddling with strange unexplained phenomena (unwanted motion, unexpected motion, Erratic spindle operation or no spindle operation at all and subsequent and broken tooling. Add to that, kiddie crayon screens unless you fiddle some more to make your own (not always successful) or spend even more money on what was once a $150 "control.

I've used Flashcut (and a few others) since 1997, and I have done quite a few retros to date using Flashcut, some for myself, but also some for others, and on a variety of machines. I like the fact that it is super easy to teach others to use, completely reliable, works with practically ANY windows PC including Laptops.

The best part is, you don't have to dig thru thousands of messages in a forum to find out which version of mach did this or that better with this or that aftermarket hardware. Come to think of it, MANY people bought into 3rd party hardware that really did not work at all (The "brain", usb "smooth thing" and a few "mpg's" come to mind). When I have a issue (and that is not often these days), I just call them and they tend to solve any problem a guy may be experiencing.

By the way, I saw their newest version 5 down at IMTS and got to play with it for a while. I was surprised to see that they have integrated Cad and Cam into the control and by the look and feel of it, it is going to be an excellent program to use. The drawing aspect is very much like Moi3d, and the Cam seamless, but still allows full control over the code created or direct access to the code for editing. Looking forward to my next Retrofit !

Gentlemen,

Thank you so much for the time that you took to answer my many questions. I really appreciate your input and knowledge. I have a 10 day long event that will be taking up my time for the next several days, so I cant spend full time on this at the moment. 100 hours is a lot of time to spend on a project for me, and given my current level of knowledge, that might be optimistic. A more or less turn key solution with stellar support sounds like a really good way to go for me. I'll give the flashcut folks a call and see what they may be able to do for me.

Mach motion is another thorough at your machine and get running in hours package but by the looks of the wiring and rust in there it probly take a bit to get running unless you know what all the wires are

1

Well I tried to get through the Shane at martin CNC this afternoon, but he was out on a service call. Gotta try to get him in the morning.

I took a few more photos of the machine. I pulled the x axis servo cover and found out that it is a Electrocraft Corp servo. Unfortunately there is no model number on the data tag. I took a few photos. As requested I got some more photos of the servo amps and also of the servo it self. There is some tape on the end of the servo with some very old dried up adhesive. I am wondering if it might be holding the encoder in place or some thing.

The servo amplifiers look an awful lot like the ones found at this eBay auction:

Hurco Servo Amp Retrofit | eBay

From the looks of it my machine at least had it's servo amplifiers upgraded some time ago. They say "Advanced Motion Control Brush type PWC servo amplifier" on them.

if you can work out what each wire does what on the machine it wont be to hard to upgrade its just how much money you wont to spend and if you wont a 5vdc system or 24vdc

Well I am not sure if I made it better or worse at this point. I unwrapped the tape around the servo encoder and started gently poking at it. I removed a couple of allen bolts that held the servo encode casting in place and gently tugged to see if it would remove easily. It felt as if there were a wire holding it in place, not being able to see what is going on I stopped, and put the allen bolts back in place. I then took off a plastic housing that was covered in the tape that I removed earlier, it looked like there were a few tuning knobs on the encoder under the housing. I did not see any loose connections, so I left it alone and put the cover back in place.

Now when the machine is fired up and I start to jog the x axis, it moves freely , but there is no change of position on the X axis read out on my screen. Also the x axis will keep on moving until it either hits a limit switch, or the emergency stop in activated. The other axis behave as expected, showing a change in position on the screen and stopping their motion as soon as I stop telling them to move. So I am thinking that at this point that the servo encoder is on the fritz.

-Ken

- - - Updated - - -

Daniel,

What would the advantages / disadvantages of a 5v or 24v system be?

- - - Updated - - -

confirmed, I spoke to Dave at Midwest CNC. Looks like I have a bad x axis encoder. Gonna order one tomorrow during business hours. I did not really want to spend the afternoon trouble shooting this, but it hard to put a new toy down when I start playing with it.

I ordered a new servo encoder from Midwest CNC. They said that they had a new style that would not use the aluminum casting pictured in the post above. I spoke to the Flashcut CNC folks. They are waiting on some pictures from me of the servo controllers. It looks like if the servo encoders are optical, and the servo amplifiers are digital, then they can be used with the flashcut CNC controller and pulse generator. I also reached out to the folks at advanced motion control (ADVANCED Motion Controls Servo Drives Brush Brushless Motor) who made the servo controllers that this machine currently has in place. So far no response back from them.

The servo encoder is setting me back $325. Ouch. One way or another I'm going to get this machine running again.

-Ken

noise can be less of a problem with 24vdc system than a 5vdc system.
I have a 5vdc system on my big machine and its a pain in the a__ cant find what's causing an occasional spike if it was 24vdc it`s probable it would not effect the machine

Quote:

---

Originally Posted by KenCameron

Daniel,

What would the advantages / disadvantages of a 5v or 24v system be?

- - - Updated - - -

---

A 24volt system will just be less likely to have problems picking up electronic noise. Basically, the signal is stronger is how I see it. If you're working around a lot of other equipment in a shop environment, noise can be a factor that can make things go haywire at times. You rarely see an industrial machine with 5 volt signals to things like relays, at least that's been my experience. When I retrofitted my controller, the outputs matched the rest of the system, which was already 24 volt. My machine was originally a $150k CNC router, so I guess there is some assumption that the engineers who made it know what they are doing and would have used 5v if that was better.

The servos/drives look pretty nice actually. If I understand correctly, the Ref + and Ref - are analog inputs from the controller. Also, you don't usually see a "Tach" input unless the drive works with analog controllers either. In some cases, drives can be either digital or analog. If that's the case, I think I'd strongly recommend against going with anything digital even if the servo drives can take a digital input. Reason being is that with an analog controller, you'll have feedback directly to the controller from the encoders, or what's called closed loop. This allows the controller to make real-time adjustments to the trajectory based on the actual position being achieved by the motion, and providing corrections to the trajectory to keep the motion within tolerances set by the controller parameters. IMHO, you'd have a hard time finding a high end CNC machine with digital drives, and there is a reason for that which is that they do not have fully closed loop feedback. Some digital signal setups provide feedback from the encoders to the drives, but not to the controller. This allows the drive to go into a fault mode when the position does not reflect the intended motion beyond a tolerance setting, but it does not provide trajectory correction and it does not provide position retention.

Position retention, as I'm calling it because I just don't know if there is a true industry term for it other than "closed loop", is essentially that the controller is aware of the position of each axis regardless of whether the system is in a state of reset or actively running. If the system is in reset, and you manually move the table or the head to a new position, a digital system will not account for the change in position because the controller is not aware that any change occurred. If the motors are running a the the system goes into reset (perhaps by e-stop or by other means), the position of each axis is lost because it's unlikely that the stoppage happened exactly as it shows on the controller DRO (the encoder info is not used, only the steps counted). With a true closed loop analog system, I can move the table around, the head around, hand crank an axis to a new position, etc. and the controller will read the change in position constantly so long as there is power to the controller. You could hit the e-stop button, move an axis to a new position by hand, reset the system and then resume a program with no loss of position.

For my CS Labs analog controller (CSMIO IP-A), the system is true closed loop, meaning that the encoder feedback goes to the controller, which in turn provides trajectory correction where needed. It also maintains awareness of the position after a reset or e-stop, again even if you were to manually try to trick it by moving the table around. It also provides what is called "homing on index" using the closed loop feedback system. In this system, when you "reference" (home) the axis, each axis moves toward the home switch until the switch is triggered, then the axis moves away from the home switch until the index mark on the encoder is found. It stops the axis on the index mark precisely, and sets the axis position to zero. Because the index mark is mechanically linked to the axis ball screws, the home position is always exactly the same every time you home the machine within the machine's mechanical tolerances. For example, I have an encoder with relatively low resolution of 2000ppr, and one rotation of the ball screw is 20mm of motion, so one encoder pulse is approximately .01mm (.00039"). Since the homing process uses an exact pulse to stop on, it always homes within a theoretical .01mm as it never misses the index pulse. Mechanically, the resolution of the ballscrew is only .002" (I have a router, not a mill), so the homing process is more precise than the hardware is. In the event of a power failure, I can re-home the machine and it will align with the work just as if it had never happened. That is nearly impossible without the "home on index" feature. You can get that feature with a digital system as well, but it requires more specific drives that output a pulse when the index pulse is found by the drive, since the feedback goes to the drive alone and not the controller. With an analog system like the CS Labs, this information is already there for the controller and it's built in with no special drives.

You may spend a little more time installing a CSMIO IP-A (there are 2 other analog systems for Mach 3, but I haven't used them), but the advantages of using an analog controller over a digital controller are more than enough to make up for it in my opinion. Also, I am being pretty conservative in the hours I gave you. It is possible that you would have the machine moving around in 20 hours or less (I could probably do that in 3 or 4 hours), but that's not quite the same thing as working exactly as you intend it to. Even if you get a "kit", there is a big difference between the machine moving around and being 100% complete. When I say 100 hours (or more), I mean to the point where there is nothing left to do and it works exactly as you want it to. I'd give that some serious consideration if the drives are capable of analog operation.