[mungaro]

Eric:

What are you going to do with the old controller? I have the book and manuals for the machine.

Mark

[erikjgreen]

The old controller (dynapath delta 10) is sitting on a shelf. It and the old keyboard/monitor arm are going to be sold for whatever I can get, I guess.

If you're willing to scan in the wiring diagram for the machine, plus maybe the manual that shows how to do service on the ways/lube system and the head, and the coolant system, I'd be willing to trade

The dynapath controller worked ok, it's just that it was slow and only held a limited number of program lines. So I decided I'd rather run the machine with EMC.

I'm also upgrading to a VFD for the spindle which is almost a spindle servo controller. With a few limitations, this should give me more or less equal functionality to the Lagunmatic 310, which had a spindle servo and drive for tapping etc.

Erik

[erikjgreen]

Also, over the weekend I tried one config for the limit switches, it didn't work as well as I wanted (translation: I made a major mistake in wiring and got nowhere).

On the bright side, all the limit and home switch wires are identified and labeled, and trimmed to the right size for routing in the cabinet.

Tried to get my 24v power supply working, found out a fuse is blown on it. Planned out SSRs and contactors for controlling power to the VFD, coolant pump, air line (for chip clearance) work light, etc.

I'm running out of cabinet room, so I'm being careful placing items.

I think I'm going to upgrade to EMC2 beta 2.5, since it has some features I want, including the touchscreen interface.

Erik

[mungaro]

Eric:

I will have the books completly copied and bound for a complete trade of what you have that you are not going ot use on the Mill. I am not very savy on change outs but I would shure like to change mine out also. I did a change on a CNC lathe to Mach3 Turn and it was a pain in the rear and dread doing it again. Thst is the reason for trying to keep my machine in spare parts.

Mark

[erikjgreen]

That would be great!

The stuff I have that I'm not using is basically the Dynapath control chassis, the monitor and the electronics in its box. I think I posted pics earlier in the thread with condition info... it's all functional, the keypad and monitor work but show some age, and most of the cables are there... I re-used some of the ones that connected to the drives.

If that works for you, PM me your address and I'll start packing it up.

Toward the end of this thread I'll post pics of my electronics conversion, including the pinouts for the encoders and how I re-wired everything, to serve as a guide for anyone else replacing the electronics on this mill.

Erik

[mungaro]

Eric:

Send me your email address because I have trouble sending messages through CNC zone.

I will take all that you are not using.. I don't need any of the metal etc. just the electric stuff.

My address is: [email protected]

[mungaro]

Eric:
I sure would like change my unit out but really do not the in's and out's of doing so.
I am just a remove and replace type of guy unless it is absolutly necessary. I like my equipment to just work properly.
Mark

[erikjgreen]

Email sent.

Not much work on the mill last night, I was doing a car exhaust instead.

I'm expecting my VFD delivery today, so I'd better get moving on the whole thing.

Erik

[erikjgreen]

Well, I had a pretty good weekend. Got a lot done (but I always want to get more done than I do. Maybe I need to set more realistic goals?)

Friday I went to pick up an e-bay purchase from a seller near me, and he gave me the item I bought plus a great deal on a brand new 1 horse 3 phase motor with encoder and a salvaged control board from some large mill or factory machine.

The 1 horse motor I'll use for something, what I really wanted was the encoder... it's set up with a nice sealed enclosure, military style screw connectors, and an 8 foot cable (it's differential/quadrature 1024 counts/rev, with an index channel too). So I'll get a coupler that fits it and my mill spindle motor, make an adapter to mount it, and I'll have my spindle encoder with absolute positioning too.

The control panel thing I mostly got because I wanted some solenoid valves... the numatics kind that can handle 140 PSI. I'll use two for upgrading/replacing the one controlling my drawbar's impact wrench, so I can push a button instead of hauling on a lever (and eventually maybe get the mill to pick up its own tools) and at least one other for switching on/off compressed air chip clearance. He gave me the whole panel for that price of those three valves, so I have to figure out what to do with the other 16 CNC brass band, maybe?

As for work on the mill this weekend, I got the wiring done to get power to the computer from the main power cord (I'd been testing with a power strip plugged into the wall) via a GFCI outlet. The original wiring had GFCI outlets on the 110 volt AC and the 83 volt AC for the servo drives, so I decided to put that back in. I had to use new GFCIs though, because the old ones didn't have screw terminals or much good wire left.

So, I got the GFCIs mounted, some prep done for the VFD (Hitachi WJ200, arrived Friday), re-connected the tachometer wires via one of the old terminal blocks to all three axes, wired the X axis command wires to the first servo drive, and carefully powered everything up. Nothing new worked, although the computer booted up ok.

I have to look and see if I'm supplying power to the individual servo boards, and also check to see if 110 is used for something besides the fans in that chassis. I did read 94 DC volts across the capacitor, so some power got through, but I don't know where that number came from... it should be higher.

I also don't know if the EMC2 config I'm using has the servo drive wires set up to work out of the box on the mesa 7i33 card (the encoders did work, but...) or if the SD1525 servo drives need to have an enable pin set before they'll do anything with the command voltage. For that matter, I don't know if the wires in the mill labeled "x servo in" and "x servo out" match "x drv" and "gnd" on the mesa card properly.

I did spend time checking the config and setting things like the SCALE parameters for the axes... 40000 for each, I think is right (counts/inch). All three axes have indexes on the encoders, set that parameter too.

I didn't get an "ok" LED on the boards when I powered them up, but I don't remember seeing that before I redid the wiring either, so I dunno.

Basically, I skipped a whole bunch of careful step by step in the hopes that I'd get a quick result, but it was not to be.

Next time I get to work on this (tomorrow?) I'll start with checking out my power wiring.

Anyone know if a standard GFCI will work with less than 125 volts? I wanted to keep the safety feature because the mill had it before, but I confess I don't know if it does anything except give me the ability to switch the drive power on and off at the moment.

Erik

[erikjgreen]

I also found a good step by step for getting the servos working, I think I'll follow this:

LinuxCNC Documentation Wiki: Tuning EMC2/HAL PID Loops

Erik

[erikjgreen]

Okay... did some more testing last night, then spent the rest of the night and today (during work breaks) doing more research.

I've decided to get the machine hooked up to my home network (I've been holding off due to laziness in getting the cat5 run) so I can update to Linuxcnc 2.5beta. Other than the new features I want, this distribution has the latest copy of pncconf, which seems to be the way to go for configuration.

I was assuming I could get things working just editing the HAL files, but given that I'm also trying to get a dissimilar pile of electronics working together, that's one more variable I don't need.

I did verify the polarity and scale on the mill encoders last night, so that's good. They appear to be on or very close to on as far as accuracy goes.

Today I got a nice reply from a request I sent to Servo Dynamics... they sent me a PDF manual for my servo amp chassis. So, mysteries solved on that.

I found out that the 110v supply I suspected for the fans is in fact for the fans, but is also required for a +-15 "bias" supply in the chassis. Each card needs the bias voltage in addition to the DC bus voltage to work. That's why I couldn't easily identify all the assemblies on the amp chassis... there's an extra power supply in there.

I also found out the exact voltage specs for the drive. Peak voltage is +- 100 VDC, peak current 25 amps, continuous voltage 92 VDC (so that's why I measured 93 volts yesterday!), continuous current 15 amps, 1.85hp. The amp will fault out and stop if DC voltage goes over 130VDC. Per the manual, this should only happen if regenerative braking current from the servo blows fuses.

The LED indicators are all "fault on" only, so no surprise I didn't see any when I powered the drives up. That's good, it means no internal problems.

The amps have a differential or single ended input plus an auxiliary single ended only input... not sure why, maybe for testing?

Switching frequency for these amps is 5 khz and bandwidth 1.5 khz. Still interested in finding out whether this gives me an audible whine.

There's an output current monitor pin on each amp capable of driving a meter.. so I could attach one, or else find a way to capture the values in real time with an ADC input each so I could have per-axis current (load) monitoring. I have that with the VFD I bought, too.

I also have remote shut down (estop/fault indicator) and remote reset pins... probably I'll end up wiring these to reset the amps when EMC starts up, and to display a message on EMC when the amps fault out due to overcurrent or temperature, or ground fault short. I could also use the limit switch inputs for estop too, since my actual limit and home switches are connected in software rather than on the amps. I may have to jumper the limit switch pins to get the amps to work before I do that, though.

The amps are shipped and come tuned for velocity mode control, but also include a special note with the manual (a later addendum) that tells how to convert them amps into a "Power stage which has the appropriate linear and proportional gain" to work as a "Tension (torque) control system". I probably won't mess with it for now.

Should be an interesting evening....

Erik

[samco]

It sounds like you are having a great time learning! We where in your shoes about a year ago. Keep plugging away and it will all fall together. You will have a machine that will serve you very well and be very solid.

Large Kearney & Trecker HMC conversion. - CNCzone.com-The Largest Machinist Community on the net!

I cannot say enough how impressed I am with mesa and linuxcnc. Solid.

As far as analog signals back to emc - you might want to look at the arduino.. One of the linuxcnc developers made a nice driver for it.

Improved Analog & Digital Interface with Arduino

I am using it to monitor spindle temp.

sam

[erikjgreen]

Cool! I'll have a look.

I made a lot of interesting progress tonight, but I found a mystery (well, a couple, but only one I care about at the moment).

I started with the manual for the servo drive and the pncconf program. Pncconf is definitely the way to go for setting up Mesa cards, I don't know why I was trying to get a hal file written by hand before.

I connected the 110v power to the drives, as I now knew it needed it. I also connected the command wires from the mesa 7i33 board to the SD1525 servo amps.

Then I powered it all up.. wow. The machine immediately started oscillating back and forth between the stops on the Y and Z axes... not X. I hit power off as quickly as I could... whew.

I tried disconnecting the command wires from the Mesa, thinking I had mismatched the signals somehow... same result (except I hit off quicker this time).

Then I went back to the manual, read, and thought things through. The only thing connected to the servo amps at this point (other than power) was the tach feedback. What if that had reversed polarity? I checked the wires, and I'd reconnected all three axes the same... red to red and black to black. I decided to try reversing Y and Z anyway, to see what happened. Then I figured "well, the mill manufacturer wouldn't have one oddball axis color code, maybe the X axis amp or tach isn't working"... so I switched X, too.

Yep, you guessed it. This time the Y and Z axes stayed put and the X went wild.

Sooo... after powering off, reversing the X tach, reconnecting the command wires from the mesa card, and powering back on, I was ready to try pncconf. At this point I noticed two things. First, most of the noise in the electronics enclosure is from the fans on the servo amps. Those things blow about twice as hard as any box fan I've ever used. The computer power supply fan is dwarfed by comparison.

The second thing I noticed was an insane high pitched whining from the servos. I'd heard the same thing before, when I'd tried to use the Dynapath controller with them. I had hoped I wouldn't get it again.

So, I figured I'd try to set things up and see if the whine stayed. I started up pncconf sitting by the mill, and noticed that the X amp had faulted out with an RMS error... checking the manual, it says that happens when among other things mechanical binding (stuck axis) happens. Turns out the X axis was slowly creeping in the + direction whenever it was turned on, and it was hard against the stops.

So, I put aside pncconf for a bit, and ran through some basic tuning as described in the SD1525 manual. Adjusting the balance for each amp I got them to the point where the axes (yep, all three) wouldn't keep creeping when no signal was applied. So when I reset the amps via the button, they stayed put and didn't fault out. So far, so good.

Then I started pncconf. I initially set up the X and Y pins swapped, and Z was swapped with an unused encoder. That corrected, I started tuning the X axis with the open loop control panel. I noticed that my encoder line count was off from what I thought (10,000 pulses per inch, not 20k). I also noticed that the drives were creeping again... they were showing both velocity and position change.

So, back to tuning. I tweaked the balance pot ever so slightly on the X amp until the numbers stopped moving, then did slow moves + and - to see how long it took to settle. I got it finally to the point where almost instantly after I stopped the jog, the encoder would show 0 motion. Hooray!

Then on to do the same to the Y and Z axes. I hadn't measured the total travel distance nor the limit switch locations (which are adjustable) yet, so I did some quick measurements with a tape and put the values into the program.

Finally, after a last check of the pncconf values, I saved the config and started emc2.

The axes started creeping again! Argh. I did an even smaller series of tweaks with the screwdriver until all three were still. I still had the whining noise, but I thought I'd try an actual move command in emc.

No luck... the limit and home switch config is wrong. I immediately got limit switch errors and emc stopped. So, I have that to work on.

Then I noticed the axes were creeping AGAIN. Dunno what happened. So, more tweaks, and all were still... until, while working on the last one, the pitch of the continuous whining changed, and at the same time all three axes started moving again!

So, now I'm wondering two things: 1) What's the whining from? 2) How is it related to the axes' motion issues?

My only guess at this point is that it's a power supply issue? Like maybe the 15 volt bias supply or the rectifier/cap for the servos is failing? If I understand how servo amps work, a voltage fluctuation in whatever value the amp uses for comparison with the input voltage could cause creep?

Obviously I need the servos to not move when not commanded, and to stay tuned where I put 'em. I'd also dearly love to be rid of that godawful whine... it sounds like it's at about 4000-5000 hz based on my (rusty) musician's ear.

Other than the whining and amp issues, the only other sounds I noticed was that the Z axis sounds a bit rough, like the servo motor or pulley isn't balanced.. I hear some kind of sound from that. I also hear some minor grinding from the X axis servo, like a bearing is going. Can't really tell though, since there was a lot of other noise and it wasn't a priority tonight.

Next steps... try to make the servo amps work right, and figure out the limit switches. Also need the enable/inhibit lines for the amps to work, or when emc2 shuts down the axes also move a bit.

Erik

PS: If it would help anyone diagnose this, I could make a video of the whine and servo movement.

[samco]

I would guess this is why....

Switching frequency for these amps is 5 khz and bandwidth 1.5 khz. Still interested in finding out whether this gives me an audible whine.

as far as drifting...
You don't have to get it perfect. Once hooked to emc (closed loop) - it will compinsate for drift.

Keep plugging away!

sam

[erikjgreen]

Okay, so if I get the loop closed it shouldn't drift... is it ok that it drifts in emc some?

I've been reading up trying to figure out how to cure the high pitched whining noise, it sounds like it might be the tach gain.

I can cure the creeping outside of emc by wiring up the amp enable circuit... basically unless emc is running (actually the HAL, I think) then the amps will be shut down (but with power applied).

I'm debating putting the VFD outside the electronics cabinet... I wanted to use a contactor to control power to it, and that plus the VFD itself will be really cramped in the current cabinet space. I should really attach a second cabinet somewhere, because I still need to put in controls for compressed air, flood coolant, and a couple other things.

More tuning this weekend, I guess.

Erik

[samco]

This is quite normal setup. My servo drives only get enabled when emc is up and on.

sam

I can cure the creeping outside of emc by wiring up the amp enable circuit... basically unless emc is running (actually the HAL, I think) then the amps will be shut down (but with power applied).

Erik

[Al_The_Man]

Off the bat you should not adjust the balance when the drive is under control, if this is what you were doing?
The command input should be disconnected and the drive command input shorted.
Did you replace the old drives? I can't recall if you have tach feedback now, but if you do you cannot normally run a drive in torque mode and use tach feedback as well.
If using a Tach the drive has to be tuned in the velocity mode and tuned with a variable DC command voltage input battery box before connecting to the control loop, balance, gain etc.
Al.

[erikjgreen]

Mostly I was adjusting balance with emc2 not running, but once or twice I did it while watching the emc2 DRO screen. The manual for the drives said it was ok to do under power, so I assumed it was ok while running emc if emc's "power" was off and emc wasn't holding the drives in place.

I have left the amps in velocity mode, since that seems to be their native setup.

I have some more info on the problems I'm having. Today, I did a fair bit of work on my setup, trying to get the drives' remote stop enabled using an NC relay, so emc2 would enable them when it ran. The relay I had turned out to have a 110VAC coil, so it wouldn't work with the output from the Mesa 7i33, which is 48VDC. I'll have to look at how other people have done that.. 48 volts seems higher than would be useful for the Mesa to output.

I corrected the issues I'd been having with the limit and home switches, verifying pin numbers and gpio port numbers until they were all set FALSE when not being pressed by the stops on the machine, and in the process double checking my mapping of wiring.

That done, I was able to move the Y axis back and forth. X and Z gave following errors right away. Then I checked the FERROR and MIN_FERROR settings, and found X and Z were fairly small (.005 in) but Y had gotten set to .125 in. I decided to set them all big for now and tune them when the amps were working better. I set them to .25 inches, which won't work for cutting anything, but will keep emc from erroring out while I test.

I couldn't determine a reasonable following error setting, but I figured that 1/8" was too large to be useful, and if the servos weren't following more closely, the servo amps probably needed to be tuned. I hadn't done that since I started the rebuild.

I powered everything off, got the manual for the amps printed, and followed the directions in it to the letter. Starting with doing initial settings on the amp pots, as given in the manual by number of turns CW and CCW.

I had problems with setting the pots at first since I had expected them to have stops (so when I put one entirely CCW, it would stop turning). They just kept going, and I was being gentle so I know I didn't break something trying. Per the manual, all but the RMS pot are 20 turns, RMS is a 1 turn.

So, I decided to connect an ohmmeter to each pot via the test points... I came up with a connector that would stay on the TP pins and watched the ohm values change as I adjusted. When I turned each one CCW to zero ohms, I took that as a starting point and counted CW turns per the manual.

At the point in the process where the manual said to adjust CMP (compensation), I got stuck. The pot on the Y and Z axes for CMP wouldn't change value. Even triple checking I was looking at the right pin and pot, they wouldn't move. X worked sort of.. if I bumped the pot with the screwdriver too hard it would jump values all over the place. I decided to leave the cmp settings at the original ones, about 1.9kohms, since I couldn't alter the Y and Z settings anyway. I matched X to that with some work.

I followed the rest of the process, until I got to the point where it says to adjust CMP clockwise until the servos enter a high pitched oscillation... I'm wondering if that's the whine I'm hearing all the time? Or would it be more of a hum?

Since I couldn't adjust CMP, and the rest of the pots were more or less adjusted per procedure, I decided to try moving under power again. I booted up, and tried, but got following errors again right away.

With that done, I found that EMC would keep the axes still once booted. I could move Y and Z axes okay, with the display in EMC keeping up, except Z motion was inverted. Easy to fix later. X just sort of jumped when I tried to move it, then faulted out again.

I decided to make a video of the whine and current status and ask for help.

At this point I'm not sure what to do with these amps. I can't follow the tuning process described in the manual, since I can't change the values on the CMP pots. I'm hopeful that if I could, I'd be able to tune them so the whining would go away and following error would be something usable (and of course so the X axis would work).

But now it looks like the adjustment pots might be fried... maybe these drives are just too worn out to work and need servicing?

I'd like opinions on the whining sound and also directions to go in... check out this movie I posted to youtube, complete with bad camera work:

My lagunmatic servos

Also, so everyone knows the process I'm trying to follow, here are some images of pages from the manual. I don't want to post the whole thing here since I'm not sure Servo Dynamics would be happy with that, but I wanted the experts to see what I was working with.

Here they are:
http://i280.photobucket.com/albums/k...ervoamp/p1.png

http://i280.photobucket.com/albums/k...ervoamp/p2.png

http://i280.photobucket.com/albums/k...ervoamp/p3.png

http://i280.photobucket.com/albums/k...ervoamp/p4.png


I'd be happy to avoid buying new drives, if anyone can think of anything. I might be able to manage replacing those pots in these boards if I'm right that they're fried. I also have to honestly say I'm a bit down on these old amps... they seem like a lot of trouble to get working, in my limited experience.

Thanks for reading,

Erik

[erikjgreen]

Okay, I've been researching and trying to figure out "what now?"

I decided to order some replacement trimpots for these amplifiers. I couldn't find exact parts, so I matched the form factor and value.. 15k pots, 15 turns, with the same layout as the ones on the boards (which I think are 15 or 20k, 20 turns).

I'll try the X axis board first once the parts arrive, and see if I can successfully replace the part. If that works, I'll try tuning that axis and see if it functions, what the following error is, etc.

If that works ok I will try the other two axes. If not, I'll probably sell the SD chassis and boards on ebay or somewhere and replace them with either Viper or Dugong digital drives.

If I can make the SD drives all work, I'll use them for some length of time, depending on how loud the whine is and how much it makes me want to poke my eardrums out

Pics when the parts arrive and I start soldering.

Erik

[erikjgreen]

Okay, I finally got some more time to spend on this thing tonight.

Having received spare parts, I got my soldering iron out and proceeded to replace the broken trim pots. Not too many problems there... having a good soldering iron (Weller in my case) makes it easy. The new ones fit side by side with the other trimpots in the place allotted, but were a bit taller (no problem).

I checked the trimpots with my multimeter using the board test points, they adjusted like the manual said they should. Then I installed the boards and powered 'em up, and started the tuning procedure. This time, it went smoothly, and I even heard the "high frequency oscillation" sound at the right point in the tuning process.

Having gotten through that, I started up emc2 to see how they worked. All three axes moved and didn't trigger the following error at low speed (following error was set to max 0.05 inches). Then I tried homing... oops, it's going the wrong way.

Turns out the direction emc thinks the axes go is inverted from where they actually go. When emc2 seeks for home in the X- direction, the quill is moving toward the + part of the coordinate system. Verified this with pncconf... a press of the + button in the open loop test moves the axis in the - direction. But, I found there are "invert motor" and "invert encoder" buttons in pncconf!

Tried those... emc2 faulted out as soon as I tried to move an axis. I think the problem is that you can't invert the tach, and the drives do look at which way that moves.

So, I got out my manual and tried the configuration for tach-free torque mode. Jumper across pins, tach pot to 0, compensation to max, signal pot proportional to max current... power up. The axes move in this mode provided I set the signal command to more than 1.1v, but once moving they simply accelerate to max speed and keep going, regardless of the command voltage. I made sure the tach wires were disconnected, no help there. No information in the manual on tuning in this mode.

So, back to velocity mode. Reconnect tach wires, power on, fix polarity, re-power on. Re-do tuning, goes smooth again. Same problem with following error.

Turn off encoder invert and motor invert.... works ok again, at least until the axes start wandering too much. I probably need to add signal gain or something so the card responds better to emc's attempts to hold the axes in place. I try tweaks of balance to minimize wandering and increase and decrease signal gain to improve response. I also adjust the tach pot to minimize overshoot... each axis settles quickly after command goes to 0, cool.

But, the axes still move in the wrong directions.

So, positive outcomes from all this... I think the servo cards are probably functional at this point, and I was able to do slow moves with very little following error allowed. On the down side, the axes are upside down and I can't change 'em in software.. I think I might be able to reverse the motor wires and encoder wires, but I'm not sure that would work either... tach in the way again.

One more item... the whining. At this point I'm pretty sure it's a normal sound for these amps... 5 khz pwm. If these servo amps were working perfectly I might ignore it or try to tune it out or something... but as things are, it's driving me nuts.

To be fair, it might be my lack of experience with analog electronics helping, but these servo amps seem really complicated to tune, and I'm not sure if it's my ineptitude, aging electronics, or some incompatibility between Mesa and SD hardware. Or a combination.

So unless someone has a brilliant idea for fixing the axes and tuning up the amps, sometime tomorrow I'm going to order 3 Dugong drives, the big ones, to upgrade amps. Those can supposedly be tuned live using a laptop connected to the drive via USB.

Anyone want to buy some SD1525 cards and a chassis, cheap?

Onward and upward,
Erik