Wells Index Analog Amplifiers, DC Permanent Magnet servo W/Tach, Resolvers...

I recently picked up a Wells Index 750 CNC Knee Mill. This machine has analog amplifiers, Permanent Magnet Servo Motors with a tachometer on the tail of the motor, and resolvers on either the ball screw or the nose of the motor shaft for the Z. It appears the amplifiers are getting about 85-90VDC and when I force power to the machine the z axis starts slowly jogging leading me to think the amp is bad. I cant find a spec sheet on the motors after days of searching. I am not familiar with servos at all and trying to decide the best route to go. Initially I was thinking of buying one servo driver and testing each axis with an existing Linux and BoB until I discovered they were resolvers and not encoders. Then I thought I would get the Mesa cards I would need to run the existing analog amps, and resolvers, but if any components are bad it would probably be better to just upgrade the servos and drivers.

This is my second Wells Index CNC, My first machine is a Model 700 that has stepper motors and it was easy for me to get movement out of the machine with existing stepper motors and some cheap drivers and BoB I had from a router project.

Amps:GA4559-1
Motors: ElectroCraft Model# E7230110 Part#072312020
Resolver: Singer 09301

With the experience here I am curious to hear some input of what route to go

@riversvic

Tom is correct about the servo drift as my Hurco KMB1 would do the same thing when I was having problems closing the loops in LCNC.

Since the control is not working, all the amps know to do is output however they were tuned.

When you get the resolvers feeding back and tied into LinuxCNC, then the control will close the loop and stop the drift. When you did the previous machine with Step Direction, you did not need feedback to close the loop. With servos you need feedback to get the control to understand what to output to the amps so that it will offset the drift and keep it sitting still. I have not used resolvers before as all of my machines have encoders. But simply put the Step and Direction puts out pulses and it is counted in LCNC, your analog amps only see -10 to 0 to +10 volts to get working and therefore cannot "count" the amount of voltage that has been sent to the amps. This is why you need the feedback to so that LCNC can see how much to add or subtract in voltage to the amp. Hope this makes it a little clearer.

I am super busy and trying to help someone else, but saw your youtube video with all the Kwik Switch 200 tooling. That is a lot of money alone, so you are way ahead of the game for being tooled. I skimmed through the video and did not have the audio on, so I will have to go back and check it out.

Not trying to jump in over Tom as he is way more advanced than me and very knowledgeable plus will help you big time. Time is his only problem, just like me. Just wanted to try and help. I have Electro-Craft servos and Servo-mate amps in my machine, which were very popular for Wells Index back in the 80's also and that is why I wanted to chime in.



Robert

Also meant to add that if it were me, I would add encoders and use a Mesa board as there is a lot of documentation for help when you have problems.

Using a BOB is not too good as you are going to run out of input and outputs really quick. Plus I have not played with a BOB in a long time but I only remember the cheap ones having 0-10volt analog to the spindle only. This will not work for a standard analog amp for any of the axis as you will not only need 0 - 10 volt positive, but will need 10volt negative to 0 for moving the opposite direction. No BOB that I am aware of has -10V to 0 to +10V for analog output.

To understand this you could take the 2 wires that are the analog input to one of the amps, say X axis and use a little AAA battery to apply a little DC voltage (I think its about 1.6V) to the input and it will make the table move. Only quickly touch the wire to the battery so that you can get it to move, if you hold it its going to keep moving. The table will move one direction and if you flip the battery around it will move the opposite direction, hence the -10 and +10. Be very careful as you do not have the limit switches (probably proximity switches) wired up and you can crash your table messing up the machine. I would wire up the limit switches to LCNC so that it shunts the amps and therefore not crash if you did try this. OR at least have a way to kill power really quick to the amp if it did take off. If you are super brave with the machine power off you can manual turn the ball screw and put the table in the middle so it is not close to the end, then power up and try what I am saying. DO NOT USE A 9V battery or anything larger than AAA as 9V is nearly the full 10volt input and will send the table wide open one way. This is just a quick and dirty way to check if your amps and servos are working. IF there is a problem and cannot get it to move this way, then there is something else to look at. Just a cheap way to check this out. I actually built a little box with a joystick wired to the amp inputs so I could move the table around and do some drilling before I got mine fully working. Of course I wired limit switches to the amps to shunt them and I never got close to the end of the table travel.

My disclaimer for anyone else that reads this and attempts anything.... I am not a electrical engineer or mechanical engineer.... so this is all at your risk.

I would never advise anyone to try this, but from the youtube video, you remind me of myself and my dad. I have picked up all kinds of old American iron from up north and brought back to Georgia to get working, and with a lot of elbow grease we have gotten this far.


Robert

@riversvic

Ok I went back and listened to the video and watched the whole thing pretty much and realized that you were only getting drift on the Z axis. I know you thought something was wrong with Z only as the others were not doing it, but that is normal.

One of my Hurco machines did exactly what yours did the first time I fired it up without the control working. The table sat solid, but the Z was slowly drifting, and when I say slow it was super slow. The amps will have a few pots for adjusting/ tuning the amp/axis. I have 6 different pots on my Servo-Dynamics amps and it is a pain to do. Do not adjust them as they are fine for getting working with LCNC. Tuning the amps should be the last resort as you can tune your PID loop in LCNC and get them tuned that way.

Ok also in the video it sounded like you needed several Mesa boards, I think you really only need one setup such a plug and go kit and maybe one more board for the resolvers. I can't help you like Tom (Tommylight) can with Mesa as I am more verse with other hardware that I have used in the past, plus now at this point I am only going to use EtherCAT and Beckhoff hardware.

Also one other thing to keep in mind. I know in the last couple of post we were discussing the BOB's. Another major problem with a BOB is that you are going to be using the parallel port (db25) and that is one of the worst ways to do a servo system. It's much better to use the Mesa plug and go so that it plugs into the PCI slot and not the slow DB25 port. The PCI slot being on the bus will keep the speed up. It's all about the speed when it comes to your encoders, feedback speed will rule the system. If you have high count encoders such as on my machine (1000 ppr), then you will loose counts and the machine will keep moving in whichever axis it lost count. I can speak from experience as it is frustrating to keep all of your code with slow speeds to avoid loosing count. If it happens you will break a end mill, or worse crash the spindle.

Once you go to another type of hardware and get away from the DB25 port, you will never go back (My Opinion LOL). Now with that said, there are some Omron amps and I am sure Yaskawa has some also that allow step direction input like a stepper system, plus there are cheap amps from China that have step/dir input. If you were going to change the amps and wanted to go that route, then you could still get away with a BOB, but just won't have closed loop feedback and like stated before you will run out of inputs/outputs really quick on a cheap BOB. CNC4PC use to have some better BOB's to get more I/O, but I have not looked at those in a few years, so I can't say who is the main player now. Plus, I have not tried this way of doing it, so I cannot speak about the speed or reliability compared to traditional servo systems, but I have seen people on the web that have went this route.

Just some things to think about.


Robert

Thank You Robert for the very in depth post. Currently I have decided to hold off on the 750 until I get the 700 going and I am learning a lot about the 750 in the process. For the 700 I went with the 6i25 pcie and 7i76 and the I/O amount is ridiculous. For the 750 I am leaning towards removing the analog amps and going to a step and direction system and swaping the resolvers to encoders. I have a friend at Rockwell looking into the Electrocrafts for some specs on them.

@riversvic,

Not a problem, was hoping to help in any way possible as I know how it is to waste time trying to find info.

I actually went back and re-read about resolvers as it had peaked my curiosity to refresh on them. They are super reliable, but not as good of feedback as encoders for the time period back then. It does look like some newer resolvers actually outperform encoders in todays world, but this is military space grade parts... not affordable to me LOL.

I did not check out the model number to the Electro-Craft motors you have, but I do have some info for the one's that I have if you run into a dead end.

Hope the retrofit is easy and goes well, as there is no better feeling than finishing and running the machine.


Robert

So initially when I was forcing power to the machine I was doing it in an unsafe manner. I now applied ac in a safer way and discovered that all 3 axis are drifting slowly when I turn the machine on. I did not notice This before because the position I was in when forcing power.

Tom is correct, since the controller is not fired up and giving commands to the servo amps then they are simply operating with no input to offset the output from the amps. In other words, since you do not have the controller operating totally the whole system from computer/mesa board sending analog output to the amps sending power to the servos moving the resolvers to send feedback to the controller loop is missing. In other words the hardware loop is open since you are missing the computer to tell the amps what to do, is not tied together and therefore is why you are just drifting when fired up. You may be able to understand it this way, the output from LCNC is to offset what is happening downstream of the amps and since there is no LCNC to give the offset then the amps are giving a very small output and the servos are drifting. That is normal and ok, once you install a computer with LCNC and hook everything together on the mesa board, that will close the loop in your hardware. COMPUTER/LCNC to MESA to AMPS to SERVO moves RESOLVER feedback goes back to LCNC through mesa board... completes the loop

So you don't get confused, there are 2 loops. The hardware itself and then in the software when using LCNC, so you are technically open in both loops. You can put all of the hardware together and have the hardware loop closed, but since the software is not there to tell the amps what to do, your loop is open on the software also.


Hope this helps clarify.

Robert